In light of “Liberation Day,” we’re taking this show out from the archives — about 1930’s trade policy and the dangerous search for national self-sufficiency with Nick Mulder, where we discussed his book, The Economic Weapon. (You can listen to part one here.)

In this episode, we’re going to get into the juicy stuff around the late 1930s, the leadup to World War II, and interesting parallels that you might see today with what the US is doing with respect to China, trade and technology.

We address:

• Why countries yearn for autarchy aka rohstoff freiheit, “raw materials freedom”

• Why states start wars because of “temporal claustrophobia,” and what it has to do with Japan ultimately siding with the Axis;

• Parallels between the “ABCD circle” (America, Britain, China, Dutch East Indies) and the semiconductor export controls today;

• Why having an empire was a liability for Britain;

• What sanctions had to do with the Czechoslovaks — even with a larger army — falling to the Nazis.

Have a listen on Spotify or iTunes.

Sanctionomics

Jordan Schneider: So, Mussolini invades Ethiopia, the the League of Nations lowers the economic boom on him, but too slowly to make him lose the war.

Watching the world really economically squeeze a middling power focued minds in Nazi Germany and Imperial Japan.

What was the reaction that you saw in the sources of the leading Axis powers as they saw Italy struggling under the weight of global economic sanctions?

Nick Mulder: Italy is put under sanctions about a month after it invades Ethiopia — so November 1935. And what you can see from that point onward in discussions — in Nazi Germany particularly, but also in Japan in the months and years following that — is an increasing focus on what sanctions would mean for them.

In German, in fact, this already precedes the sanctions against Italy. Of course, they had been exposed to a particularly nasty blockade in World War I — they have that memory; the Nazi ideology is very concerned with food security. So they have plenty of reasons to be focused on that.

In Germany — which at that time also was operating really on a shoestring amount of foreign exchange reserves — this was very worrisome. Germany, of course, was engaged in one of the largest armament efforts ever seen in a capitalist economy in peacetime, as Adam Tooze shows in Wages of Destruction. And their external dependence was massive: n order to run all those steel industries, all the energy, the coal, the oil — you need to import it. And interestingly, because of the Great Depression, imports of these commodities had actually become cheaper because there had been a huge commodity downturn.

So you think that the Great Depression causes trade to collapse and the sanctions will no longer work. Actually, the opposite is true for a number of key commodities. The commodity downturn is so severe that it becomes extremely cheap to source oil, coal, iron ore, textiles, raw inputs for a variety of industries, scrap metal from abroad.

So that is the weak Achilles’ heel that Nazi Germany and Japan (with a very similar industrial structure) both have — and that is what they choose to then start protecting.

And in Germany, there’s a really direct effect of the League’s sanctions against Italy on [Germany’s] thinking. The main body that’s in charge of national defense planning — it’s called the Reich Defense Council (Reichsverteidungsrat) — gets together in early December 1935. And it has all the key people there: Hjalmar Schacht (the Reichsbank chairman and Finance Minister), the heads of the General Staff and planners, [Alfred] Jodl, [Wilhelm] Keitel; and soon, in the spring of 1936, Hitler joins them, too. And at each of those meetings they emphasize, “We need to look at what’s happening to Italy, we need blockade resilience, and we need to figure out how we move not just from a kind of trade-commercial protectionism, but to an economic model that is immune to sanctions.” And what they mean by that: it’s immune to having raw material imports severed; they call that rohstoff freiheit, or “raw materials freedom” — they have total autonomy because they have all the raw materials they need for war.

That becomes the aim of their planning going forward, and the main thing that it manifests in is the famous Four Year Plan that is announced in the spring of 1936 — while the League’s sanctions against Italy are still in effect. And it’s then given a particularly powerful head: Hermann Göring becomes the head of the organization running that, and they have a goal: “within eighteen months we want to be independent in terms of fuel from the rest of the world economy, and within four years we want to be totally ready for an aggressive war of conquest.”

Jordan Schneider: What’s the difference between autarchy and autarky?

Nick Mulder: Yeah — so we spell it different ways. Some people write it with a k — that’s the most common. But you also see it sometimes with ch. And there’s an interesting etymological difference between them.

The older version is autarchy, which comes from autos — so it means to rule oneself, to be independent, to be in command of oneself in one’s own position. And that basically just means that you have autonomy, political or otherwise.

Autarky actually comes from the verb arkhein, which is to suffice or to subsist. And that means that you could actually survive off of your own resources. So that’s a narrower definition.

And the interesting thing at the time (one of the famous Italian economists, Luigi Einaudi, notes this): some states that try and become autarkic — actually have access to all the resources that they need within their own territory — lose the ability to have full independence, because they need to engage in policies that are so radical that they effectively close off lots of options for themselves politically. And that’s actually what I think ends up happening in the 1930s: the road to full self-sufficiency is a road that goes through conquest. And that ends up accelerating this war — that had already been in the air was already very possible — but it ends up bringing on a kind of war that is particularly virulent and aggressive and even genocidal, I would argue, because of some of these dynamics.

Jordan Schneider: Let’s talk a little bit about import substitution and the role of German industry in not only becoming self-sufficient by conquering other people’s lands that had coal and cotton, but also changing the way that they consumed those raw materials in order to make themselves more self-sufficient.

Nick Mulder: Yes — there’s a number of schemes that they launched in order to become self-sufficient, and some of them had already been pioneered in World War I and in the 1920s.

So World War I has this big scientific breakthrough that today still powers a lot of global agriculture and sustains a huge part of the world population: the synthetic fixation of nitrogen, which allows you to make fertilizer using simply oxygen. And it means that you no longer need to use saltpeter and some nitrates that you get out of the ground. You can actually use atmospheric components. That’s one thing.

The other thing is fuel hydrogenation. And IG Farben, which becomes infamous for creating Zyklon B — the gas used in the Holocaust — also is one of the main huge German chemical corporations that pioneers a technique for turning coal into oil. [Coal and oil are] both forms of carbon energy, but coal is a lot harder — it’s kind of as if you imagine (even for people who don’t do chemistry) that you add a ton of water to coal, and you put it under an enormous amount of pressure, and you heat it and actually you get something approximating some sort of oily substance; you don’t need to refine it. It’s extremely energy intensive, very wasteful, and very inefficient — so you need an enormous amount of feedstock and fuel in order to get this reaction going. But it is possible for countries that have only coal to turn it into gasoline and a variety of other things, particularly aviation fuels, through fuel hydrogenation.

So that’s the technology that the Nazis hope is going to make them ultimately independent of imports of oil. And as the League of Nations considers this sanctions measure — extending the sanctions against Italy with an embargo on oil imports — [that technology] becomes very important.

The Japanese also take [hydrogenation technology] over. IG Farben — the chemical corporation in Nazi Germany that is accelerating this with huge subsidies from the Nazi government — also sends people to Japan. And in Manchuria and North Korea, the Japanese have access to huge coal reservoirs, so they build a number of plants — both the Imperial Japanese navy and the Imperial Japanese army have their own competing fuel-hydrogenation projects.

And apparently, the North Korean regime today still has some of these plants in the same places that the Japanese Navy built them in the 1930s; and North Korea has massive coal reserves. So there’s speculation that Kim Jong-un might be able to make it through a full fuel embargo by using, basically, Nazi-era technology.

Jordan Schneider: Wild.

Nick Mulder: The other really wild thing, by the way, is that the main postwar user of fuel hydrogenation is apartheid South Africa. They, too, have the same thing: massive coal reserves. They get put under an oil embargo, and they use [hydrogenation] in order to circumvent that [embargo], partially. And today, actually, the largest fuel hydrogenation plot is in South Africa, owned by the South African state-owned company Sasol. So this is a really interesting afterlife with technology.

Economic Asphyxiation and Pearl Harbor

Jordan Schneider: You started taking us to East Asia, so let’s stay there. How does Imperial Japan’s thinking change post-Ethiopia?

Nick Mulder: Japan is — even more than Italy, I would say — a country that is really on the fence for a long time about what its posture toward the West should be. And this is the general thing that I try to emphasize in the book about the interwar period: the war with Nazi Germany was, to some degree, probably inevitable at some point (it was just in the nature of the Nazi regime that they were going to try and use violence); but the fact that Mussolini ended up fighting on the side of the fascists was already less necessary; the fact that Japan sided with the Axis is even more remarkable.

And there was a much bigger split within Japan about who the opponent should be. It’s equally imaginable that they would have gone to war against the Soviet Union and  remained on the side of the Western allies, Britain and the United States, who they rightly saw as much bigger adversaries.

But one of the things that ends up derailing the Japanese liberals’ (so to speak) or the more pro-Western camp’s plans is the war in China. They, of course, are partially themselves to blame for this, because Japan has already invaded Manchuria with a false-flag operation — in 1931, the Manchukuo Imperial Army becomes a sort of state within a state that ends up undermining the central government and essentially running its own foreign policy.

But by the time it’s 1937 or so — we’re now in the immediate aftermath of the Ethiopia sanctions — the Japanese state is in a situation where it still could go either way. And what ultimately ends up happening is that one camp of its officers in China ends up in a fight with Chiang Kai-shek and the Nationalists. And actually, it seems now (according to most historians of China) that the Nationalists, too, were actually kind of pining for a confrontation at that point: in 1931 China didn’t want war — but in 1937 the calculation seems to have been, on the part of some people in the KMT ruling elite, that Japan was going to get stronger every year; if they were going to fight Japan, better do it now than later.

So interestingly, you can see a whole number of countries and groups in the spirit have this sense of “temporal claustrophobia” (a term from Chris Clark in his book Sleepwalkers). And it’s not just the Japanese and the Germans — it’s also the Chinese, actually, that want to have a confrontation with Japan sooner rather than later. So that ends up triggering a war in 1937 that is arguably the start of World War II, because it directly carries on into the Second World War.

So that complicates the picture dramatically, and it ends up triggering a slow drift, essentially, of Japan into an anti-Western alliance — because the West ends up siding with the Chinese resistance, because of course, [the West wants] to make sure that Japan doesn’t take over China entirely.

Jordan Schneider: “Temporal claustrophobia” — what’s your take? The causes of World War II are multivariate, and your book is front and center of my mind. So I’m curious: thinking back, in the final moments when Japan is thinking about starting Pearl Harbor, when Hitler is thinking about invading Poland and then invading the Soviet Union — this very human fear that even if our odds are bad now, they’re going to keep getting worse, seems very clear. And instead of reevaluating whether or not you want to play the game that gave you these bad odds in the first place, you decide to take the plunge and roll the iron dice, as the case may be.

Let’s talk about what the US ended up doing after 1937, as we get into 1939, 1940, and 1941 when it comes to economic sanctions on Japan.

Nick Mulder: The US has already been considering economic sanctions on Japan since 1931, since the original invasion of Manchuria and the creation of Manchukuo. But at that time, Herbert Hoover is the president, and he holds back on it. It takes quite a while into the [Franklin Delano] Roosevelt administration, really Roosevelt’s second term, before he decides to start getting tougher on Japan — and he has his famous Quarantine Speech in the fall of 1937.

And after that there are a number of incidents. By the summer of 1938, he for the first time begins to call on American companies to institute what he calls “moral embargoes” — voluntary restrictions by American firms. One of the reasons that he’s doing this is because there are Neutrality Acts in effect, which make it impossible for the US president to discriminate by cutting off trade with one country that’s party to a conflict and not with the other; the Neutrality Acts actually obliged the US government to break off arms trade with both parties to a conflict. So this is very tricky for Roosevelt — he has to negotiate these neutrality acts. And if he declares there is a war going on in East Asia, then China also loses access to American arms. So this is why he needs to first go through the private sector and try and have them do it voluntarily.

Now, at some point, they find ways around it. And by 1939, world war has broken out in Europe, too, with the invasion of Poland, and that makes it a lot easier. And that summer, Japan keeps pushing further and further, not just in northern China against British diplomatic presence, but also into Indochina.

And the other thing is that Japan at that point is even more dependent on US trade and on US exports of these commodities than it was in 1935 and 1936 — because the British empire is totally focused on producing for its own war effort, because it’s fighting against the Nazis and it has prioritized its own colonies. So the British empire goes into essentially full economic lockdown mode; Japan can’t really trade that much with them anymore. So Burma, India — those markets become a lot more difficult to access. So [Japan] becomes more and more dependent on trade with the US.

And then Roosevelt steadily ratchets up the pressure in 1940: he lets his commercial treaty expire, so trade becomes more onerous between the US and Japan. And by the summer of 1940 — after the Nazis have taken over all of Europe, and Japan also pushes into Indochina and is now trying to take over French colonial possessions there, because Paris has fallen to the Nazis — he decides it’s really time to start putting a limit on this. He begins to openly restrict a lot of iron-ore and scrap-metal shipments to Japan.

So these are actually the first full discriminatory economic sanctions. He’s targeting Japan openly. He’s trying to throttle this key raw material, making sure they just cannot produce enough to sustain their war in East Asia. And in the summer of 1941, the situation had escalated further still because Lend-Lease has gone into effect — so the US is now also bankrolling the war effort of the British Empire, of Chiang Kai-shek of the Nationalists, and of a number of other countries. And [Lend-Lease] actually needs to prioritize raw materials for [the US].

So part of the story of economic sanctions against Japan that end up triggering the Japanese attack is that the US cannot simultaneously mobilize its own war industry and keep exporting at the same rate with Japan — there’s just a limited amount of North American raw materials. And this, then, means that even if there hadn’t been really severe restrictions, Japan would have seen some dip in what it would have been able to obtain from the US.

But what Roosevelt ends up doing: he puts restrictions in place in July 1941, and then he leaves on a trip to meet Churchill on this big cruiser where they draft the Atlantic Charter together in August 1941. And while he’s away, Dean Acheson and [Henry] Morgenthau (at the Treasury) actually end up — on their own initiative — wrapping up and increasing the sanctions, making them very hard to take off. They freeze all Japanese foreign assets. The US has not declared war on Japan at all — so these are actions where they openly target Japan’s foreign financial reserves. And they cut off oil supplies — and that’s really the thing that sort of sets the final stage of the temporal claustrophobia in motion.

Jordan Schneider: So let’s play the counterfactual game. There are two fantastic books on this: Eri Hotta’s Japan 1941 and Michael Barnhart’s Japan Prepares for Total War, both of whom hint at the idea that perhaps America could have nudged Japan to go invade the Soviet Union instead by not putting on these sanctions.

I’m curious if you think there was a way in which these sanctions could have been rolled out more deftly, giving Japan more of an exit ramp than they felt they had.

Nick Mulder: I think it’s a very interesting suggestion that they could have pushed them toward invading the Soviets — but ultimately I don’t think it would have made a difference, and it wouldn’t have been a feasible solution for the Japanese leadership. And here’s why.

The core commodity that they are extremely anxious about is oil. They have none of it on the Japanese aisles. They have some technology to turn Korean and Manchurian coal into oil, but it’s still not the full amount they need. And what they really can do: they can import from the US, they can import from Mexico, and then there are a number of other places like Iran, Venezuela — those are the main producers in the world. And finally, there’s only one that’s within a reasonable distance of their own territory: the Dutch East Indies.

And the key factor, I would argue (and not just because I’m Dutch), is the fact that this oil embargo is a three-country embargo — it’s a British-American-Dutch embargo. So that is important because the Japanese have simultaneously been negotiating with the Dutch East Indies over preferential access to oil production from Indonesia — and that would have given them maybe as much as 60% of the entire Dutch East Indies’ oil production, which would have taken care of their basic needs.

But the two important things that happen: one is that the Dutch East Indies government is by that time isolated, because the Netherlands has already fallen to the Nazis; the actual Dutch government is in exile in London. So that means, essentially, the Dutch don’t have a lot of independence anymore because they’re now hosted by Churchill — effectively, the Anglo-American leaders can determine what the Dutch do. The second thing is that the Japanese are so desperate for commercial expansion that they end up over-egging the demands they make to the Dutch East Indies government — and the trade treaty goes nowhere; they don’t get that access. And that ultimately is what makes them realize, “Look, this is an encirclement.” It’s an ABCD encirclement, as the Japanese nationalists call it: America, Britain, China, and the Dutch East Indies — that’s the box that they think that they’re in.

And it bears a really interesting parallel with the current [chip export controls]: I’m not saying we’re in the same situation yet, but the current restrictions on chips (including ASML and the Japanese) are an American-Dutch-Japanese-English embargo, now against China. So Japan and China have just switched roles here — the other three countries are actually the same.

Jordan Schneider: What was the Dutch political economy? How are they thinking about managing their negotiations with Japan in 1940, 1941?

Nick Mulder: They are traditionally neutral, and they have been trying to play that role for a long time. They didn’t participate in World War I. They had no desire to enter World War II. They weren’t really planning to enter on the side of the Franco-British Expeditionary Force. They would have opened their territory if there was a need to, but they were trying to do what Switzerland, and Denmark, and the Scandinavian countries were doing — but those also got  invaded by Hitler, so Switzerland is really the only one to get away safely.

And so this traditional Dutch idea of neutrality was already under threat — and this is what ends up pushing them into joining the Anglo-American oil embargo.

And if you also think about it: Shell — which is one of the main oil producers that is not American in this period, and controls most of the Venezuelan and also a lot of the Indonesian oil production — it’s an Anglo-Dutch firm. It was called Royal Dutch Shell for a reason. It’s like Unilever, one of these Anglo-Dutch capitalist enterprises.

So they’re increasingly drifting into the Anglo-American camp and losing this traditional middle position that they had between Britain and Germany.

When Sanctions Actually Work

Jordan Schneider: Can we say FDR had temporal claustrophobia, too, in starting Lend-Lease? Is this a unified theory of everything?

Nick Mulder: It’s an interesting question, and if you read the accounts of people who’ve recently written about this shift in thinking — like Stephen Wertheim in his book Tomorrow, the World — I think that there is a kind of sense that the whole world had changed for FDR after the fall of Paris.

The summer of 1940 is this moment where, for the first time, one of the original three victors of World War I, a beacon of liberalism in the eighteenth, nineteenth, and twentieth centuries, is under the rule of a new kind of totalitarianism — and that’s when even fairly neutralist Americans, for the first time, become amenable to this idea that “Nazism is really a threat to Western civilization,” and they need to do something. And it’s from the summer of 1940 onward that you, I think, start to see in the American elite an increasing preparedness to use these measures.

And the first place where they do it is Franco in Spain. The experience of using coordinated oil sanctions between Britain and the US also starts in that summer. And one of the reasons, I think, that the US goes into the oil sanctions against Japan in the summer of 1941 so blithely is that the oil sanctions in the summer of 1940 against Franco work really well.

They were extremely small. They imposed them for only a few weeks, and then they lifted them again. And they did it just to prove the point that Spain is entirely dependent on American oil: they only have to stop two tanker ships in the Port of Houston (Spain’s entire oil supply can be provided by six vessels a month) — that was enough of a demonstration to show that Franco better not join the Axis.

And it’s the confidence bestowed by that sanctions’ success in the summer of 1940 — a kind of “almost” deterrence, very light usage that issues a clear threat — that, I think, makes [the US] think that they can do the same with Japan.

And of course, racial attitudes play a role here: they just think that the Japanese, ultimately, are easier to manage than the hot-headed Spaniards, and that ultimately they won’t do [something like Pearl Harbor].

But Japan is much further away, and it actually does have a major oil producer right next to it (the Dutch East Indies) that it hopes it can secure. So the main objective of the Japanese campaign in the winter of 1941, 1942 is the Dutch East Indies’ oil fields. There’s a lot of other useful stuff for them, but the general staff is extremely clear that that needs to be the priority. And in order to get there, you need to conquer the Philippines, you need to boot the US naval bases out of that part of Asia — so a lot of these other things become necessary as a way of getting to Sumatra.

Jordan Schneider: The story of Franco being scared off joining the Axis is illustrative of my biggest takeaway from your book: sanctions are great when you go all in. The half measures — say, “Oh, we’ll do this cute thing and have it be financial sanctions,” or, “We’ll just have this nice little escalatory ladder to slowly try to make our adversary realize that we mean business” — don’t work nearly as well as the times where the countries just say, “No, you’re not allowed to import any stuff, and the stuff you’re not going to be allowed to import is going to be the most important thing for your economy, and we won’t let you get it again until you do what we want you to do.”

And there are a number of moments — particularly with Nazi Germany in the mid-1930s — where that really could have stopped rearmament. And we talked last time about Bulgaria, we talked about Paraguay, potentially Japan in 1931 as well: if the pain spigot was turned all the way on early enough, then maybe you end up not having these horrific, world-shattering eventualities of what World War II brought us.

First, am I wrong? And second, what was it about the 1930s that stopped more aggressive economic actions from being taken earlier on as the tides of revanchism ended up ebbing?

Nick Mulder: So to your first question: you are not wrong — but you are right that maximum sanctions are the best only under certain highly specific conditions.

The two particular factors that are key to expanding the success of oil sanctions against Spain in the summer of 1940: [firstly] Franco has just come out of this really grueling civil war — so reconstruction is paramount, and he’s ruling over a devastated society, and he needs all the resources he can get for reconstruction. So [the sanctions] get him and threaten him at a moment of weakness.

Secondly, he has an alternative — the Axis — and he goes to Hitler and asks Hitler what resources Hitler has for him. And of course, Hitler himself is extremely anxious about his access to these things and has nothing to spare, and says, “Sorry, but you’re going to have to fend for yourself and conquer Morocco or something.” So that’s not exactly an attractive proposition. And ultimately that’s one of the reasons why the Axis is a weak alliance — because they cannot meaningfully compensate for each other’s weaknesses. So this is one of the things that makes the situation for Spain and Japan different.

And Japan still has hopes that they can win in China. It’s the classic story of: you are committed to a war that’s not going anywhere; it’s devolved into a guerrilla war; it’s gobbling up ever more resources (like Afghanistan in 2010 or something) — and the Japanese military keeps telling the leadership, “No, but we need one more surge, and then we can win in China.”

And surely they want to go back to peace in East Asia. They don’t want to have to fight the British Empire, the Royal Navy in Singapore, and the US Navy across the Pacific. But they end up being in this war against an opponent that is now receiving steadily more aid from the West — and the spring of 1941 [brought] Lend-Lease, and then it became clear to the Japanese, “Look, the Chinese are going to be in this confrontation for as long as the Allies want them to be. So we need to get to a deal with the Allies. But they now also are not only funding our opponents, but also turning the screws on us. How is this not already a war against us, essentially?” And that accounts for one of these crazy things — that they declare a war, that they actually know that they stand a very small chance of winning and they’re almost certainly going to lose in the long-run. So that’s one aspect.

The other thing you asked about — what are the factors that are holding back tougher sanctions? Part of it has to do with the states in question needing to come up with these sanctions plans on the fly. They have some studies — I used a bunch of them as source material for my book, and they were really interesting to read because they give you great analysis of different import vulnerabilities, and they’re very useful as inside intelligence accounts of the economic history of the 1930s. But they do not always have a good understanding of the world economy. That’s one thing.

They also have large amounts of interest involved in these international, intercontinental sanctions campaigns. And the main trading partners of Japan are the colonies and the dominions of the British Empire — and they actually are not in favor of sanctions on Japan. Australia, Canada, India, New Zealand all have an enormous amount of trade at stake with Japan, because Japan is the only rich, industrialized country in Asia that can buy their exports. So they either trade with Europeans who are now all at war — and then the only other place for them is the Japanese Empire. So for Britain, having an empire is actually a liability. It prevents them from being able to put tougher pressure on Japan early.

Jordan Schneider: Let’s come back to Nazi Germany. Hitler wasn’t doing so incredible when he invaded the Rhineland, right? And plenty of historians have written the hypotheticals of, “If France just decided to fight, then they would have won the war.”

Do you think the same would have happened if the Germans weren’t confronted militarily but in a more aggressive economic fashion in that time period?

Nick Mulder: Yes, I think it would have caused huge problems for the Nazi regime — and so it is an important counterfactual to ask at certain moments what it would have done.

Some of the vulnerabilities were compensated for: Germany got more raw materials from Southeastern Europe and from Eastern Europe, where they got these preferential trade agreements and were able to bully Balkan states into giving up their resources. But they certainly remained very vulnerable.

The other thing is, just militarily, Czechoslovakia could have decided to fight in 1938, and there’s a very good chance that Germany would have lost. The Czechoslovaks had a larger army than Nazi Germany in the fall of 1938, [but] they are persuaded — and actually forced — by Britain and France to dismantle their border defenses and stand down with their army.

And a role, I think, should be accorded to the just Munich crisis — there, too, you already have backup plans for an economic blockade if there is a war that breaks out, but ultimately, the appeasement argument wins. So it’s not even sanctions — it would have been just basic alliance integrity: if they had just upheld the French, particularly their pact with Czechoslovakia and with the Soviet Union, Germany would have faced a three-front conflict, and it would have been over pretty quickly.

In 1939, if France had invaded on the western front when Hitler invaded Poland — same thing. The German General Staff would have probably refused to implement Hitler’s plan. 

So there are many, many moments where Hitler rolls the dice, and he keeps winning — but every time he does it again, he has to wager everything he has gained up to then. And that’s the story of the radicalization of Nazi Germany.

Heading to Sanctions School

Jordan Schneider: All right. World War II — it started. How do the Allies take what they’ve learned over the course of implementing sanctions in World War I and through the League of Nations, and do the best that they can to try to cut off the Axis from accessing financing and critical raw materials?

Paid subscribers get access to the second half of our conversation. We discuss:

• How the blockade and sanctions regimes of WWI differed from WWII;

• The history behind the construction of the United Nations, and how it was tied to calibrating sanctions;

• The effect of the nuclear age on the relative morals of sanctions and conventional war;

• Parallels between the Cuban Missile Crisis and today’s tech restrictions against China;

• What lessons pro-decouplers should learn from this history of sanctions.

Read more

An anon asks the question: Is China racing to develop AGI?

U.S. leaders increasingly frame artificial intelligence as a future-defining competition. Secretary of the Interior Doug Burgum warns of an “AI arms race,” while calls for a “Manhattan Project for AI” grow louder. Corporate giants echo this urgency: OpenAI claims China is “determined to overtake us by 2030,” while Anthropic’s Dario Amodei dreams of a US-led unipolar world — only if America speeds up and beats China to AGI.

But this entire “race” narrative hinges on a crucial assumption: that China is sprinting toward AGI as well.1 Is it?

This piece, inspired by Dwarkesh’s AI scaling essay, stages a Platonic debate between two voices:

• The Believer, who insists China is racing to beat the US to AGI

• The Skeptic, who doubts China’s focus and urgency

They clash over several key questions: Are Chinese policymakers truly committed to AGI? Do China’s top AI labs see it as feasible and imminent? Are investors pouring serious money into AGI projects? And does DeepSeek mark an inflection point?

Are policymakers AGI-pilled?

Believer:

China’s senior leadership has been all-in on AI for at least a decade. They launched their “big funds” for semiconductors in 2014. In 2017, the State Council set a clear goal: making China the world’s primary AI innovation centre by 2030. A year later, Xi Jinping himself called AI a game-changer, with a "profound impact on economic development, social progress, and the international political and economic landscape." That’s five years before ChatGPT dropped.

Skeptic:

Sure, they’re all in on AI, but that’s not the same as racing toward AGI. Back then, Beijing prioritized facial recognition, surveillance tech, autonomous driving, and industrial automation — narrow, specialized applications. They didn’t exactly throw their chips in the “AGI” basket, spending only $300m USD on their hyped up “AI megaproject”, which ended up just being a mechanism for funding dozens of small research projects, not some giant cluster.

Believer:

Alright, so you’re saying China is into AI but not AGI. But after ChatGPT in spring 2023, the Politburo explicitly stressed "development of artificial general intelligence" and the importance of building an "innovation ecosystem" around it. That’s a shift toward serious long-term AGI ambitions.

Skeptic:

Yes, that was the first time the Chinese top leaders used the term “AGI”, and that should make us think. But the Chinese term here, “通用人工智能,” doesn’t necessarily translate to “AGI” in the sense of “superintelligence” — it can also mean something more mundane like “general-purpose AI.” This was also a few months after ChatGPT exploded onto the scene, so they may have mostly been just thinking of LLMs at the time.

Believer:

But AGI was just about all anyone was talking about then! Even if the term has two meanings, certainly they would have discussed the idea of AI with human-level intelligence if they were bringing up this kind of term in such a major meeting.

Skeptic:

But here’s where it gets interesting. In China’s political system, a big policy announcement like this one gets followed up by clarifications in party-state media. And guess what? A couple of months later, a People’s Daily op-ed directly referenced the Politburo meeting and explained the difference between "specialized AI" and "general-purpose AI" (通用人工智能). This wasn’t about AI that could outthink humans, but more about systems that can do multiple tasks. The article even said general-purpose AI is still in its "early stages". Nobody in Zhongnanhai is feeling the AGI.

Believer:

That was almost two years ago! In February 2025, Gao Wen — head of Pengcheng Lab and the guy who once briefed Xi Jinping on AI — wrote in People’s Daily that we are in a “transitional phase from weak AI to strong AI”, adding that AI is the “strategic commanding height in great power competition”. If that’s not AGI race rhetoric, what is?

Skeptic:

We can also look at how the wider bureaucracy interpreted the Politburo message on AGI. The Ministry of Industry and Information Technology (MIIT) included “AGI” as a focus area in a new funding round. The projects they describe are aiming to set up data centers with just 1,024 GPUs by 2025. That is a rounding error next to U.S. clusters. The MIIT lumped “AGI” with the Metaverse, humanoid robots, and brain-computer interfaces. Yes, the Metaverse.

Believer:

But local governments are jumping in too — Beijing, Anhui, and Guangdong have all announced AGI initiatives, pooling compute and data resources for large model research. They’re discussing roadmaps that go beyond LLMs, exploring brain-inspired AI and causal reasoning models. Doesn’t that show a shift toward AGI?

Skeptic:

A shift? Maybe, in so far as these are the first policies that specifically focus on general-purpose AI. But the fine print still emphasizes relatively boring applications like healthcare, government services, urban management, and autonomous driving — not superintelligence.

Also, just because they want some fancy new approaches to AI to succeed doesn’t mean they will — the field of AI is littered with the bones of research agendas that absolutely would have created AGI, if only they worked.

Policymakers waking up after DeepSeek?

Believer:

You dismissed the 2017 AI plan and the 2023 Politburo meeting. Fine. But DeepSeek has changed the game. Within weeks, CEO Liang Wenfeng met with Premier Li Qiang and even Xi Jinping. That’s serious state recognition.

Skeptic:

Of course these meetings mean something. DeepSeek’s sudden rise injected fresh confidence into China’s economy, and the state would be foolish not to capitalize on that. Getting Liang in the room and state media then saying basically that the Party supports the private sector, and the private sector supports our national goals is a clear signal: the tech crackdown is over.

But that does not necessarily mean Beijing wants to bankroll DeepSeek’s path to AGI. Li Qiang and Xi meet with plenty of tech companies every year–and by the way, the firms who got the most prestigious seats in the meeting were hardware companies like Huawei, Xiaomi, and BYD, not software firms. These are carefully curated PR moments, designed to showcase government priorities and entrepreneurial success stories. DeepSeek certainly checks those boxes. But not every company that gets a handshake from Xi receives a blank check.

Catching Xi’s eye is like drawing Sauron’s gaze: impressive, but rarely ends well. Remember when Premier Li Keqiang cozied up to Jack Ma in 2013? He invited Ma to exclusive symposia, and lauded Alibaba’s contribution to job creation. We all know how that ended — Alibaba got caught in the regulatory crosshairs during the tech crackdown.

Coming back to Liang Wenfeng’s sit-down with Li Qiang. They supposedly discussed the 2025 Government Work Report. And what did Liang walk away with? A mention of the AI+ initiative — something that was already in last year’s report — and a vague nod to "the extensive application of large AI models," plus "AI-enabled phones and computers.” All of this is very application-focused. “AI+ initiative” basically translates into “AI + literally anything but AGI.” Soon we’ll have AI + rice cookers, AI + karaoke machines, and AI + slightly smarter traffic jams.

If Liang went in hoping to convince Li Qiang to push for AGI, it sure looks like he came out empty-handed.

Additionally, the Chinese government remains cautious. Vice Premier Ding Xuexiang, who’s leading China’s new Central Science and Technology Commission, recently emphasized that China “will not blindly follow trends or engage in unrestrained international competition”. That’s not a race mentality.

Believer:

That’s just diplomatic rhetoric, they don’t want to freak out American policymakers even more by saying they’re racing to AGI. Oh, and by the way, Huawei got prime position in that Xi meeting and they’re the one designing China’s AI chips.

Skeptic:

But their regulations speak to this. Remember that Baidu’s ERNIE bot had its first demo version available in March 2023, but had to wait months for the Cyberspace Administration to finalize its genAI regulations and grant Baidu a license before they made it available to the general public. Going through China’s genAI licensing process takes several months for some companies. This suggests a willingness to slow AI deployment for control and social stability. China’s not racing — they’re speed-walking with a leash.

Believer:

Being cautious also doesn’t equal not believing in AGI. Chinese thought leaders have expressed concerns about existential risks from AGI. If you don’t believe in truly transformative AI, there is also no reason to be concerned about x-risk.

Leading AI Labs and their Investors

Believer:

And the AI labs are charging ahead — policymakers or not!

DeepSeek CEO Liang Wenfeng says flat-out, “Our destination is AGI.” Zhipu.AI 智谱AI founder Tang Jie 唐杰 says they are “conducting AGI-related research around superintelligence and superalignment,” and need to “plan for AGI based on large models” with an “aim to lead the world”. Moonshot AI’s CEO Yang Zhilin called AGI “the only meaningful thing to do in the next 10 years”, while Minimax CEO Yan Junjie compared the road to AGI to the Long March.

Skeptic:

Yes, some start-ups are certainly ambitious. But not everyone shares their enthusiasm. Baidu CEO Robin Li, for instance, claims that “today’s most powerful AI is far from AGI” and that we “don’t know how to achieve that level of intelligence yet”. And Li Kaifu’s 01.AI has also backed away from foundational models.

But ambition alone doesn’t sway investors. China’s VC is struggling. DeepSeek’s CEO complained that they want a quick buck for their money, and are hesitant to support those who make true innovation, suggesting he may also struggle receiving the kind of money he wants.

China’s big tech is not helping the most dynamic independent research labs like Microsoft, Amazon, and Google have. Zhipu’s $341 million from Alibaba and Tencent and Moonshot’s $1 billion from Alibaba are peanuts — Microsoft dropped $14 billion on OpenAI. None of the major Chinese start-ups is valued above $3.3 billion. DeepSeek is a bit of a special case because it is entirely funded by its parent High-Flyer and has no external investments, so its value is anyone’s guess. Forbes estimates it at somewhere between $1 billion and $10 billion.

Even if we assume $10 billion — compare that to OpenAI at $300 billion and Anthropic’s $61.5 billion. It’s not a race — it’s a rout.2

Believer:

DeepSeek has proven that China can leapfrog, not just follow. And investors are taking notice.

Just a year ago, Allen Zhu Xiaohu (朱啸虎), a prominent Chinese VC, dismissed AGI advocates as “delusional” and saw “no point” in engaging with China’s genAI startups. Fast forward to February 2025, and he’s changed his tune: "DeepSeek is almost making me believe in AGI."

And it’s not just private VC money pouring in. In January 2025, the Bank of China launched a massive AI support project — 1 trillion RMB ($140 billion) over five years — funding AI through equity, loans, bonds, insurance, and leasing. The launch event was high-profile, co-chaired by major players:

• Ge Haijiao (葛海蛟), Chairman of the Bank of China

• Yang Jie (杨杰), Chairman of China Mobile

• Zhang Peng (张鹏), CEO of Zhipu AI

• Zhou Bowen (周伯文), Director of the Shanghai AI Lab

• Li Meng (李萌), former Vice Minister of Science and Technology

Every major ministry was present: the National Development and Reform Commission, Ministry of Science and Technology, Ministry of Industry and Information Technology, and the State-owned Assets Supervision and Administration Commission. All the big tech firms — Huawei, Tencent, Baidu, Ant Group, iFlytek, China Mobile — were there too.

At the announcement, Ge Haijiao didn’t hold back: "Large AI models are profoundly shaping the global political and economic order” and present “a strategic pillar for ensuring national security." He lines up the country’s AI heavyweights, delivers that message, and then drops the bombshell: “And here is 1 trillion RMB.” If that’s not commitment, what is?

Skeptic:

Let’s break down the Bank of China’s announcement a bit further.

First, the total scale: 1 trillion RMB ($140 billion) over five years translates to $28 billion per year. That’s a serious chunk of money. But compare it to Stargate’s intended $100 billion per year, and it starts looking less impressive.

Second, where’s the money actually going? The new funds will support the entire AI ecosystem — not just AGI research. It covers chips, data, and AI algorithms, but also AI + robotics, AI + the low-altitude economy, AI + biomanufacturing, and AI + new materials. These are all important, potentially transformative fields, but they dilute the focus. Meanwhile, Stargate is laser-focused on super-scaling compute for AGI. It’s a concerted effort with resource centralization. This Bank of China project, in contrast, looks more dispersed, just like that earlier megaproject.

Subscribe now

Let’s be realistic about who’s managing this money: Chinese state banks. These institutions are notoriously risk-averse. They favor safe, mature technologies over speculative bets on moonshot research. No way they’ll push in all their chips to fund AGI breakthroughs, which require Masayoshi-levels of risk tolerance.

Believer:

Yes, there’s uncertainty about how exactly the money will be spent. But the Bank of China head specifically named computing centers in China’s eight designated AI “hubs” as a priority. If anything, that suggests a significant chunk of the funds will go toward centralized compute infrastructure.

Chips and Compute

Skeptic:

Let’s talk compute. No one in China is building GPU clusters at the scale of 100,000 chips like in the US. And despite DeepSeek’s efficiency breakthroughs, developing and deploying superintelligence will demand massive scale — far beyond anything China has today.

Believer:

Again, DeepSeek has changed the game.

Alibaba just committed over 380 billion RMB (~US$52.4 billion) in the next three years to build cloud and AI hardware infrastructure. That’s more than their total capital expenditure from 2015 to 2024 combined. In other words, in just three years, they’ll spend more than the last decade. It’s the largest-ever investment by a Chinese private company in AI infrastructure. And Ali CEO Eddie Wu 吴泳铭 made their ambitions crystal clear: “Alibaba's ultimate goal is to achieve AGI. AGI will be able to perform more than 80% of human capabilities. Since 50% of the global GDP consists of human wages, achieving AGI would create the world's largest industry.”

Policymakers are also signaling increased interest in compute. On a recent visit to China’s top mobile operators — who play a major role in compute infrastructure — Premier Li Qiang emphasized that AI is “bringing profound changes to the world” and urged them to optimize compute resource use.

Skeptic:

You have to do the math: $52.4 billion over three years is $17.5 billion per year. In 2024 alone, Amazon spent $77.8 bn, Microsoft 75.6 bn, and Google 52.6 bn.

Chinese compute investment may be growing, but it still trails far behind what Western investors are pouring into their 100,000-GPU clusters. There’s nowhere near the AGI fever in China that we see in the U.S.

Believer:

If China isn’t building 100,000-GPU-scale clusters yet, it’s not because they don’t want to — it’s because they don’t have the chips. U.S. export controls have slowed their access to cutting-edge semiconductors. If they had the chips, they’d build the clusters.

Skeptic:

China does have chips — over 1 million new AI chips were estimated to be added in 2024 alone. Some estimates suggest even more. In theory, that’s enough to build multiple 100,000-GPU clusters. The problem isn’t just chip supply; it’s that these chips are spread across smaller, fragmented clusters, many of which are underutilized. Their chips are smeared like peanut butter on a bagel — thin, uneven, and oddly unsatisfying

The real bottleneck? Not chips — belief in AGI. No investor is stepping up to bet big on AGI-scale clusters. If there were real conviction that AGI was imminent, we’d see someone pulling these chips into centralized compute hubs. But so far, that hasn’t happened.

Believer:

You really think they’re content with their current chips? Look at the $47 billion third round of the “Big Fund” — they’re doubling down on domestic semiconductor manufacturing. They aren’t just waiting around; they’re going all in.

Skeptic:

Sure, but how do you know AGI is the end goal? $47 billion for chips could also be justified by more narrow economic or military concerns, not a full-blown AGI sprint. And that brings us full circle: China is heavily investing in AI — but that doesn’t mean it’s racing toward the singularity.

Conclusions

So, who wins this debate?

Overall, the Skeptic makes the stronger case — especially when it comes to China’s government policy. There’s no clear evidence that senior policymakers believe in short AGI timelines. The government certainly treats AI as a major priority, but it is one among many technologies they focus on. When they speak about AI, they also more often than not speak about things like industrial automation as opposed to how Dario would define AGI. There’s no moonshot AGI project, no centralized push. And the funding gaps between leading Chinese AI labs and their American counterparts remain enormous.

The Believer’s strongest argument is that the rise of DeepSeek has changed the conversation. We’ve seen more policy signals, high-level meetings, and new investment commitments. These suggest that momentum is building. But it remains unclear how long this momentum can be maintained–and whether it will really translate into AGI moonshots. While Xi talks about “two bombs one satellite”-style mobilzation in the abstract, he hasn’t channeled that idea into any concerted AGI push and there are no signs on any “whole nation” 举国 effort to centralize resources. Rather, the DeepSeek frenzy again is translating into application-focused development, with every product from WeChat to air conditioning now offering DeepSeek integrations.

This debate also exposes a flaw in the question itself: “Is China racing to AGI?” assumes a monolith where none exists. China’s ecosystem is a patchwork — startup founders like Liang Wenfeng and Yang Zhilin dream of AGI while policymakers prioritize practical wins. Investors, meanwhile, waver between skepticism and cautious optimism. The U.S. has its own fractures on how soon AGI is achievable (Altman vs. LeCun), but its private sector’s sheer financial and computational muscle gives the race narrative more bite. In China, the pieces don’t yet align.

This essay deliberately avoids comparing U.S. and Chinese progress or assessing whether Beijing could achieve AGI — whether through innovation, espionage, or other means. Our focus is strictly on intent, not capability: what China seeks to pursue, not what it might accomplish.

A final point of caution: this staged debate covers a snapshot of arguments from early 2025. Things are changing fast in AI, and many of the arguments presented here could change as well. While this piece highlighted that there is currently little evidence that China is racing towards AGI, this may of course change at some point in the future. If the government or major investors started radically centralizing resources while changing their tone in public statements, this could signal change. ChinaTalk will keep an eye out!

ChinaTalk is open to sponsorships. Email jordan@chinatalk.media to start a conversation.

Welcome to part two of our Cold War history series with Sergey Radchenko. Here’s part one.

In today’s epic interview, we discuss…

• Khrushchev’s removal from power and the transition to the Brezhnev era,

• How the USSR and China managed their relationships with Vietnam,

• Sino-Soviet border conflicts, Brezhnev’s negative feelings toward China, and Nixon’s rapprochement,

• Watergate and the inability of China or the USSR to understand American politics,

• Why the Soviets decided to invade Afghanistan,

• Reagan’s approach to negotiations and his relationship with Gorbachev,

• How to manage the containment paradox and unknown adversary motives when competing with China and Russia today.

Co-hosting today is Jon Sine of the Cogitations substack.

Listen now on iTunes, Spotify, YouTube, or your favorite podcast app.

The Fight For Vietnam

Jon Sine: At the end of part one, we were just talking about the Cuban Missile Crisis. Then, there was the transition in 1964, when Khrushchev was unceremoniously deposed. My question is, in your story of status, how much does prestige carry over with each new leader? Brezhnev’s era was described as “Khrushchevism without Khrushchev,” after all.

Sergey Radchenko: The phrase “Khrushchevism without Khrushchev” was coined by the Chinese. They were unhappy with how the Soviets continued to pressure China to adopt policies that the Chinese deemed unacceptable. They believed that although Khrushchev pursued anti-Chinese policies and was removed, the policies remained in place. Thus, they invented the term “Khrushchevism without Khrushchev” (没有赫鲁晓夫的赫鲁晓夫政策).

Regarding what happens to the standing of new leaders when leadership changes — that’s actually a very interesting question. Khrushchev was removed in October 1964, at which point he basically was Soviet foreign policy. He stood head and shoulders above everybody else in the Soviet Politburo. He could make decisions single-handedly without consulting anybody, like sending missiles to Cuba. There were some people like Anastas Mikoyan who spoke up against this, but even he was very careful in his opposition. Khrushchev was unassailable in many ways during his last years in power.

Once Khrushchev was overthrown — or, “retired” as they put it — you have a new set of Soviet leaders, including Leonid Brezhnev as the Party First Secretary, who later becomes the General Secretary. You also have Alexei Kosygin, who is effectively the prime minister, and Nikolai Podgorny. They form a triplet, a “Troika” (Тройка) of leaders that ran the Soviet Union for a period of time.

Starting from 1964, Brezhnev, who is nominally in charge as the First Secretary, feels out of his depth. Brezhnev came to power not entirely confident of himself, particularly in foreign policy. He was consulting with others in a way that Khrushchev never did.

This explains why, in 1965, Alexei Kosygin went to China in a bid to repair relations with Mao Zedong. Brezhnev didn’t go himself, but he could not stop Kosygin from doing so. Kosygin wanted to repair relations with China, and Brezhnev agreed to let him try. Of course, nothing came of it because Mao Zedong told Kosygin that their struggle with the Soviet Union would “last yet another 10 thousand years – less is impossible.”

Meanwhile, Brezhnev, in order to improve his standing and legitimacy as the leader of a communist superpower, extended aid to Vietnam. When we talk about legitimacy and its practical effects, Vietnam is a key case study. Khrushchev didn’t really care about Vietnam, and the Soviet Union was not heavily involved there until 1964. Beginning in 1964, there was escalation with the Gulf of Tonkin incident and increasing American presence, but there also was renewed Soviet commitment to Vietnam.

Why? Because Vietnam was a communist ally in need. Helping Vietnam bolstered Brezhnev’s personal legitimacy and standing as a leader, especially with the Chinese watching. The Chinese were accusing the Soviets of trying to sell out communist movements around the world. Brezhnev, as the leader of this communist superpower, had to help Vietnam then.

That’s why we see the beginning of massive Soviet involvement in the Vietnam War, including advisors on the ground and military equipment. This would not have happened under Khrushchev, but for Brezhnev, there was a deficit of legitimacy, and Vietnam filled this gap.

Jon Sine: Can you explain why you see this change in attitudes towards Vietnam? In your book, you write that the Chinese position under Mao favors conflict, struggle, and violent revolution as something to be promoted, whereas Khrushchev prefers peaceful coexistence. What changes with Brezhnev?

Sergey Radchenko: I struggled with this question in the book because I could not understand why Khrushchev was so committed to supporting Fidel Castro while not caring about the Vietnamese. I have no answer except for perhaps some personal factors. Ho Chi Minh visited Moscow quite a few times in the late ’50s and early ’60s, trying to repair the relationship between the Soviet Union and China. During these visits, Ho Chi Minh contradicted Khrushchev softly and tried to teach him in a way that I think annoyed him.

It’s difficult to read into these dynamics, but somehow Khrushchev just didn’t like Ho Chi Minh and didn’t want to bother with this faraway place that he did not understand. He preferred to focus on another faraway place he didn’t understand — Cuba — for reasons unknown, getting deeply involved there but not in Vietnam.

Was Brezhnev very different? In terms of his knowledge about Vietnam, no. He was on the same page as Khrushchev and knew nothing about Vietnam. But for Brezhnev, it was a matter of demonstrating his commitment to the communist cause of struggle against American imperialism.

Part of the explanation for Brezhnev’s increased commitment is that it coincided with American escalation in Vietnam. If at this point he had done nothing — saying, “I don’t care about Vietnam, let the Chinese handle them” — he would have looked weak and appeared to have abandoned a communist ally. These factors contributed to the increasing Soviet involvement.

Jon Sine: Let’s stay on this discussion of American involvement. The Gulf of Tonkin incident in August 1964 arguably changed a lot in the course of this sort of ménage à trois between the US, the USSR, and China. On the Chinese side, you read into their documents and show that Mao’s Third Front movement (三线建设) was basically inaugurated by the American escalation in Vietnam. That is to say, Mao was able to build the biggest economic change in China since the Great Leap Forward as a result.

In writing this book, what new insights about this did you discover during the research process?

Sergey Radchenko: The most interesting thing for me regarding Vietnam — and there’s a whole chapter on the Vietnam War in the book — is that we in the West typically think about Vietnam as America’s war. We focus on American boots on the ground, how the Tet Offensive and impacted the United States, the bombing campaigns, and so on. We look at the situation through American eyes.

What I was trying to do in this chapter was to understand how Vietnam mattered for the Soviets. What I discovered was that it wasn’t even American involvement in Vietnam that drove the Soviets crazy. The Soviets were really worried about how Vietnam impacted their competition with China.

As you mentioned, the Chinese in the mid-1960s became increasingly radicalized and saw Vietnam as a case study where they could showcase their influence and vision for world revolution. Mao was at the forefront, advising Vietnamese leaders and saying, “The Americans are invading? Well, that’s not a big deal. That’s fine.” At one point he said, quoting this Chinese proverb, “You have green hills in Vietnam — you can go into the green hills and there’s always firewood there (留得青山在，不怕没柴烧).” Reading this, you might wonder what he was talking about, but his attitude was that revolution was on their side, the future was on their side, so they should fight against the Americans and not listen to the Soviets.

The Chinese were really upset about the Vietnamese taking Soviet aid. Meanwhile, the Vietnamese were trying to balance between both sides, telling the Chinese about how much they admired Chairman Mao, how wise he was, what good advice he gave. Meanwhile, they also felt that they needed Soviet aid and weapons to fight against the Americans. They were balancing both sides.

From the Soviet perspective, this was basically a struggle for influence. Who would win in establishing influence in Vietnam, the Soviets or the Chinese? That was their Vietnam War.

Jordan Schneider: You also bring up the aspect of logistics — how the Soviets were getting aid to the Vietnamese. Partly, they had to go through China. The US innovated during that time too — this is when containerized shipping first came into play.

What can you say about how the Chinese were interfering with Soviet aid to Vietnam?

Sergey Radchenko: It was awkward because the Chinese could not tell the Vietnamese, “We hate the Soviets, so therefore we’ll be blocking the aid they try to send you once it crosses our borders.” But in reality, the Soviets did try to slow it down. There was a pileup of train shipments at the border.

The weapons had to go across China, starting from the Northern border and going all the way to Southern China and then onto Vietnam — sometimes those train consignments were looted by the Red Guards. There was something akin to a civil war going on in China in the 1960s. It was chaos.

The Soviets would always raise this issue with the Vietnamese. They would say, “Look at the Chinese. You say they are your friends, but look what they’re doing. They’re not transporting our weapons, which you desperately need.” The Vietnamese response was, “Please, just let us handle it. We have to be very careful, and we have to understand the Chinese are having a difficult time.” This was a real problem for the Vietnamese and certainly a propaganda point for the Soviets in the late 1960s.

Jordan Schneider: Were the Vietnamese actually that unruffled by the Chinese interfering with the shipments?

Sergey Radchenko: They hated it. There are so many things that the Vietnamese hated about what the Chinese were doing in the 1960s. We have to understand, of course, that the Chinese were extending considerable aid themselves to Vietnam, and their aid was also important in terms of light weapons and railroad workers — about 300,000 people at some point over the entire period. There was actually considerable Chinese involvement on the ground, and it was important to the North Vietnamese.

But what they did not like was the Chinese interfering with the Soviet weapon shipments, and their pressure on Vietnam to stop taking Soviet aid. They did not like this at all.

One thing they really hated was when Chinese domestic politics became radicalized during the Cultural Revolution — from mid-1966 onwards — and the Chinese tried to promote a Cultural Revolution in Vietnam. If you imagine being the Vietnamese at this point, fighting this war against the Americans, and the Chinese come with their crazy radical ideas and Red Guards and whatnot, that is not something that’s going to sell well with the Vietnamese Communist Party leadership.

You see already at that point a cooling in the Vietnamese approach towards China. They don’t like what the Chinese are doing, and they’re more willing to listen to Soviet advice. The Soviet advice, by the way, is basically to talk to the Americans to end the war, or at least engage in peace talks, whereas the Chinese told them to keep fighting.

Jordan Schneider: I would recommend folks check out The Dragon in the Jungle: The Chinese Army in the Vietnam War, which is a fun dive into Chinese archives regarding this question in particular.

Greatness over Grain and Unlikely Partnerships 狐假虎威

Jordan Schneider: Anyway, let’s take it to Nixon. Brezhnev all of a sudden develops this incredible love affair with one of America’s arguably least lovable presidents. What changed about Brezhnev’s approach to Vietnam, the US, and China once Nixon came onto the scene?

Sergey Radchenko: Jordan, as a way of introduction, we have to understand where Brezhnev finds himself in the late 1960s. First of all, there was effectively a war going on with China. They fought a border war in March 1969 over this little island, which is actually closer to the Chinese side of the Ussuri River. The Chinese have a reasonable claim, but they fought a war over this island. Then, the Soviets made noises about a potential preemptive nuclear strike on China. This was a nasty situation, and the Soviets were worried about a Chinese invasion.

The Chinese, by the way, were afraid of a Soviet invasion. “Afraid” does not cover it — they were paranoid. They thought the Soviets were ready to do a 1968 Czechoslovakia-style special operation all over again. By 1969, they were really preparing for a Soviet invasion.

The Soviets, by contrast, thought that the Chinese were going to invade Siberia. There’s even a Soviet-era joke about the war between the Soviet Union and China that lasted for only two days. On the first day, war is declared. On the second day, the Chinese surrender, and 100 million Chinese cross over the border as prisoners of war. Then the Soviet Union declares unconditional capitulation.

There’s this sense in the Soviet Union that China is an existential threat on their border in the Far East, and that drove Brezhnev nuts. He really hated the Chinese so deeply. You can see that in his various commentary about China and the Chinese — how they’re unreliable, how you can never trust them. There are a lot of orientalist tropes there.

One of the things that I was able to do in my book was track where Brezhnev got his ideas about China. It turns out that he got them from 19th-century Russian orientalist literature.

Jordan Schneider: Let’s stay on that weird racist arc for a second. Do you think these ideas precipitated the Sino-Soviet split, or do you think the breakdown in relations emerged from other factors and then Brezhnev just used orientalism as a justification?

Sergey Radchenko: It’s both. You also see these tendencies under Khrushchev, even when the relationship with China was pretty close. Khrushchev never trusted Mao and thought that Mao was this sketchy character, dictatorial in his ways, and so on.

Once the relationship really started to deteriorate in the 1960s, it became so much more pronounced, both for Khrushchev and especially for Brezhnev, who basically just went full-blown racist on China, saying really nasty things about the Chinese and contrasting them very negatively with the Europeans and the Americans whom he thought he could have a good relationship with.

We have to, of course, put this in context. What was happening in China in 1966-67 was crazy. The Soviet Embassy was literally under siege by Red Guards who erected a scaffold, a platform to hang the ambassador. This was canceled at the last moment because Zhou Enlai came out and talked to what I think was a 14-year-old girl who was in charge of this Red Guard contingent trying to storm the Soviet Embassy.

Zhou Enlai said that hanging the ambassador would interfere with diplomatic relations, so they canceled the operation. But from the point of view of Soviet diplomats, they were about to be lynched, and this was all relayed to Moscow in the form of reports. The Soviets were reading this saying, “What is going on? These people are crazy.”

This contributed to this quite racist thinking about the Chinese and complete failure to understand what was going on there. The Soviets were not alone — nobody could understand what was happening in China. The Cultural Revolution was madness in so many ways, but it contributed to Brezhnev’s thinking that Europe was the better place to build bridges.

Already in the late 1960s, even in the mid-’60s, he began his engagement with de Gaulle, then Pompidou, and later with Willy Brandt after the German Social Democrats came to power in September 1969. Brezhnev felt that Willy Brandt was the way to go, and Brandt, of course, had his Ostpolitik and the promise of economic cooperation with the Soviets.

Brezhnev started developing this European détente before he even turned to the United States. At this point, he was saying nasty things about Nixon. Nixon, of course, was known to Soviet leaders — he was in Moscow in 1959 having the famous, or rather infamous, Kitchen Debate with Khrushchev. They knew who Nixon was, and they didn’t like him. Brezhnev also said things about Nixon that were not very complimentary, but then things changed.

What changed was, first and foremost, Henry Kissinger visited Beijing in the summer of 1971, and Nixon’s visit to Beijing was announced. Brezhnev saw that as an “Oh my God” moment. I don’t know if he believed in God, but he was thinking, “The Chinese are our enemies. The Americans are going there — this is really bad — so we have to get Nixon to come to Moscow.”

He really invested himself into this summit, which ultimately happened in May 1972, and he developed a fairly friendly relationship with the American president. After that, I think it really built from there. Somehow, he thought that he had Nixon’s trust, and he then went to Washington to visit Nixon and went all the way to Nixon’s Western White House, which is not far from LA.

Jordan Schneider: Brezhnev was pushing for détente with Nixon. You emphasize that when Khrushchev was envisioning himself on the world stage, he thought he could spread his wings and play some big nuclear weapons-backed games. But Brezhnev, as you say, had a different vision of the US and Soviet relationship — he wanted to run the world together. How did he come to want such a world, and how did that manifest in his policies?

Sergey Radchenko: Jordan, to explain that, let’s go back to Khrushchev. Khrushchev was a really optimistic character. He thought that the Soviet Union was surging ahead and doing so well economically, and in terms of science and technology. The launching of Sputnik made him really optimistic — and for good reason. That’s why he proclaimed that in 20 years’ time they would “establish communism."

As the Soviet joke goes, they decided to hold Olympic Games instead because it was never going to be realized. But in 1960, it seemed like this was possible. Then after that, things went downhill pretty quickly.

The Soviet economic situation wasn’t turning out so well. In 1963, the Soviet Union was already importing grain and spending gold reserves. How can you build communism if you cannot even feed your own people? That was a major problem.

By the late 1960s, it was totally obvious that this promise of communism was not being realized. The Soviet Union was not coming closer to that goal of building abundance and joy for everyone that was part and parcel of that initial promise Khrushchev made.

In 1968, there is this memorandum from Andropov to Brezhnev which basically said, “Look, we’re losing the Cold War. We are losing to the Americans because we’re not investing enough in R&D, in education. Our labor productivity is low,” and so forth.

I was able to access the Soviet Politburo Discussion from 1966, which includes hundreds of pages of discussions about the state of the Soviet economy, and they all knew that things were not going well.

They tried to implement reforms, the Kosygin reforms, introducing incentives, basically making the country more capitalist. But it wasn’t working. They tried to tinker with it here and there, but the whole system was just garbage. It was just not delivering.

Because of that, the ideological underpinnings of the Soviet system started to fall apart. People were not buying anymore because the material abundance was not there. Brezhnev understood that. Brezhnev was looking for a new idea, and that is where this external aspect of greatness comes into play. America would recognize the Soviet Union as an equal superpower, and that could be sold to the Soviet people as Brezhnev’s achievement.

Together with this came the idea of peace, and Brezhnev saw himself as a peacemaker leader of the USSR. In his conversations with Nixon, he would often refer to the fact that the Soviet Union and the United States together had enough nuclear weapons to destroy the world six, seven, eight times over, and so they had a responsibility to fix the world — to resolve the nuclear problem, to stabilize China.

The title To Run the World came from the moment in the spring of 1973 when Henry Kissinger went to Moscow, and Brezhnev took him hunting wild boars. It was outside of Moscow in this dacha, and they were in this hunting tower, just Brezhnev, Kissinger, and the interpreter, waiting for the wild pigs to come and feed so they could shoot them.

Kissinger recounted that moment later in a conversation with Nixon, he said, “Look, Brezhnev told me, ‘Don’t take any notes, but I’ll tell you this. What we want to do is we want to run the world together with the United States.’” That struck me as a very interesting proposition from a leader of a communist superpower, working together with the United States. How do you even explain this from a Marxist-Leninist perspective? That doesn’t make any sense.

True, it doesn’t make any sense! But it does make sense from the perspective of selling Soviet greatness to the Soviet people.

Jon Sine: Nowadays, there are strategists talking about how the US could peel Russia away from China. It’s interesting that, at various points during the Cold War, both the Soviets and the Chinese were seeking to align with the US against the other.

What historical lessons from that dynamic do people forget today?

Sergey Radchenko: This is my favorite example of how ideology can be just cast aside. We can talk about China being this revolutionary power that complained endlessly about the Soviet Union betraying the global revolution, but then we get to the late 1960s, and they basically come around and embrace the United States.

An interesting moment discussed in the book was, I think in the fall of 1970, when the American Mao biographer Edgar Snow turned up in China, as he did on occasion. The Chinese leaders thought Snow was a CIA spy, but he was actually just a leftist journalist.

Anyway, Edgar Snow turned up in China and had a conversation with Mao Zedong in which Mao said, “We think that Nixon is a good fellow.” Snow essentially said, “Well, how can you say that? You don’t mean that, right?” Mao Zedong repeats himself, “We think he’s the best fellow in the world.”

Mao, of course, hoped that Edgar Snow would carry this to the Nixon administration — remember, in his mind, Snow was a CIA agent. Later, the summary of this conversation was circulated to the party committees around China to introduce the Chinese people to the idea that China was changing course and turning toward the United States.

Do you know how party committees would have kind of fake debates? They had debates and questions were collected, and those questions were reported back to the center. Questions included things like, “If Nixon is the number one best fellow in the world, why are we having a quarrel with the Soviet Union? Can’t we repair relations with them as well?” The Chinese party officials were confused about this, but Mao would not have any of that. He felt that those people didn’t understand strategy, and that China had to turn to the United States because the Soviet Union was the enemy.

You see, it doesn’t matter that turning to the United States entails turning to “imperialism.” That’s not what matters.

Going back to your question — at that point, both the Soviet Union and China were willing and able to set ideology aside and turn to the United States and try to improve relations with them on the basis of geopolitical great power competition with each other.

Jon Sine: That’s exactly what I was thinking, because normally the story is you have brilliant strategists and Kissinger making that flight to Pakistan and then secretly flying off to China for this engagement. But some have argued, and I think your findings support this, that Mao in some ways, insofar as anyone deserves credit, might be the one making this move.

Reading back through Mao’s various writings, when he talks about the Japanese, for instance, he will sometimes half-jokingly, half-seriously say, “The Japanese are the people that we have to thank the most, because without them, we would never have come into power.” Are you sure there’s no element of speaking tongue in cheek here when he’s saying these things about Nixon being the number one good fellow?

Sergey Radchenko: You always have to be careful with his pronouncements. Mao does say that about the Japanese consistently. What he basically means is that because of the Japanese invasion, the nationalists were weakened — the Kuomintang Party was weakened — which provided the space for the Chinese communists to establish their power.

In a sense, it does make sense what he says about the Japanese. Of course, he’s being sarcastic to a certain extent.

With regard to Nixon, when Nixon and Mao met in February 1972, Nixon tried to engage him in conversation, and Mao kind of brushed it all aside and said, “Oh, you can talk to Zhou Enlai about practical matters. I don’t care about practical matters, I want to talk about philosophical matters.” He says something like, “We like rightists, and we voted for you in the 1968 election."

I don’t think Nixon quite gets that. Nixon is like, “Okay, well, listen...” But Mao is trying to say that he finds the Republicans more trustworthy because although they’re reactionary, as far as Mao is concerned, at least you always know where they stand. Whereas with the Soviets, you don’t. The Soviets would say one thing, and they would cover their actions with leftist phraseology, but in reality they do something completely different.

The same goes for Social Democrats, and he’s full of disdain for Social Democrats like the Europeans — Willy Brandt and all those people in Europe who are basically trying to engage with the Soviet Union, which would allow the Soviet Union to deal with China. He’s accusing the Western Europeans of trying to orchestrate another Munich, where they would sell out China.

Those are the kind of issues that Mao brings up, and I think he’s quite honest about it. It’s not a sarcastic comment when he says that he likes Nixon.

Jon Sine: The Soviets also really liked Nixon. This is probably my favorite part of your book — certainly the funniest. I was actually laughing. They find out about the Watergate scandal and see Nixon about to be removed from power, and Brezhnev is flipping out, thinking that the whole political system in the US is specifically trying to undermine détente. Then he sends a message, I believe it was to Andropov, essentially saying, “You’ve got to help Nixon. You’ve got to find some dirt on his opponents to help him.”

Share

Sergey Radchenko: It is funny. It shows what kind of great material you can find in the Russian archives. In this particular episode, Andropov of the KGB is there, and you have Brezhnev’s aide asking, “Do we have some compromising material on Nixon’s opponents that we can use to help Nixon?”

It is hilarious, but the Soviets never could understand Watergate. Mao also did not understand Watergate. They could not comprehend how you could remove such a wonderful president who was creating détente and bringing the Cold War to an end and who had just won a big election. How? Clearly it’s some kind of conspiracy. At one point they say, “You see? They killed Kennedy, and now the same people are bringing Nixon down.” That was Brezhnev’s take on this. They never get it; they never understand what this is about.

Jon Sine: That was such a crazy time in American politics, from that decade on — the president literally having his brains blown out on TV, to then having a president actually impeached and removed from office. We forget today how things have been very crazy in the past.

The last analogy I was thinking of is when Nixon was about to become president. Some people say he engaged in what might technically be called a treasonous act by being in contact with the Vietnamese and trying to discourage them from agreeing to a peace deal. His rhetoric might sound familiar to people today who pay attention to news about Ukraine. He came in saying that the war in Vietnam was a complete disaster and we needed to get out of it immediately. But what ended up happening was an escalation beyond anything previously seen. I don’t know how much of a warning that would be for today, but I did think it was interesting.

To be fair, I heard Stephen Kotkin bring this up, so I thought it was an interesting analogy, though obviously there are many disanalogies — things that don’t map equally.

Sergey Radchenko: Niall Ferguson also raised this in a recent article comparing Trump to Nixon. The difference between Ukraine and Vietnam was that in the late 1960s, American troops were in Vietnam, which had a direct impact on American society and politics. There were anti-war protests going on in the United States. Today, American troops are not in Ukraine, so you don’t have the same kind of impact — no protests.

Both Nixon and Trump see these theaters as peripheral to America’s core interests. There is a parallel there.

Nixon’s way to get out of Vietnam was to try to coerce Vietnam, including by intensifying the bombing, dropping threats of a nuclear attack on Vietnam (which did not really work), and also working with the Soviets. From his perspective, that was a big part of the whole engagement with the USSR — finding an exit for the United States, “peace with honor,” getting out of Vietnam, and getting the Soviets to facilitate this.

The problem was that the Soviets were not facilitating any of this. From the Soviet perspective, they loved the fact that the Americans were engaging with them. They loved the honor of being a co-equal superpower. But if you’re a co-equal superpower, aren’t you supposed to have clients? You’re not supposed to betray your clients or force them to surrender to the United States. For them, this linkage never worked. They thought they could both help Vietnam and have a good relationship with the United States. That did not really work for the Americans.

Jordan Schneider: I love this line in your conclusion. This is talking about Khrushchev, but it applies to Brezhnev as well. The core question is, “Would he be willing to moderate Soviet foreign policy in return for being accepted as America’s equal? The proposition never worked because of the very fact that being accepted as America’s equal meant rejecting external constraints on foreign policy behavior.” What sort of equality would you talk about if you couldn’t have proxy wars or missiles in Cuba?

“Soviet engagement in the Third World, and indeed American acceptance of this engagement, were part and parcel of what it meant to be a superpower.”

This gets at the other core question as we come to the end of Brezhnev’s effectiveness — if he had stayed healthy and if Nixon wasn’t impeached, they could have kept the good thing going. But once his health deteriorated and Nixon departed, you argue that bureaucratic interests took over, and everything started ramping up again.

What are your thoughts on the transition?

Sergey Radchenko: It’s hugely important. We’re talking about health and power. It’s a big question for the late USSR, and we’re also familiar with this question in the West.

Brezhnev was very charismatic, very active, and very engaged until about 1974, and then he declined rapidly. He developed all kinds of ailments and basically became a figurehead. By the late 1970s, he didn’t really decide anything.

If you look at his summit with Carter in Vienna, he was just reading from pieces of paper. He wasn’t even thinking about what he was reading. When Carter responded, he would turn to his aides and ask, “What should I say now?” They would give him another piece of paper and he would read from that.

The Soviet deep state took hold in the late 1970s in a major way. The bureaucratic interests took over, and the Ministry of Defense became really important. For them, promoting various geopolitical schemes in the Third World was a key issue. They also resisted nuclear disarmament. They thought it was a bad idea. They wanted more investments from the state, so increasingly the Soviet economy became more militarized.

You have the interests of the International Department of the Soviet Communist Party, whose role is to promote revolutions in the Third World. It’s in their job description — promote revolutions. Previously, Brezhnev would have pushed them away because it’s beyond their pay grade to define policy. But by the mid to late 1970s, they come to influence policy in a major way, and we see that in the increasing Soviet involvement in Africa.

Then you had the Foreign Ministry and the KGB, and they all had their own distinct interests. Those bureaucratic interests increasingly came to dominate Soviet policymaking, and it drifted as a result. Policymaking became more conservative overall because there was no single figure who could break the ice and take charge the way Brezhnev did with the Nixon Summit in 1972. The bureaucracy was against it, but Brezhnev did it anyway.

Well, by the late 1970s, he couldn’t do that because he was no longer mentally alert. It is part of the story of Soviet decline, and it also shows how having an active leader at the top could actually have good results. Not always, because sometimes you could have an active dictator who will do terrible things precisely because he’s not constrained by the bureaucracy. But in the Soviet case in the late 1970s, power at the top was missing and the bureaucracy took over.

Boredom and the Graveyard of Empires

Jon Sine: The last important thing that happens under Brezhnev, though he’s not really conscious of it, is the Soviet invasion of Afghanistan. To your point, you have Gromyko, Andropov, and Ustinov — the KGB, defense, the Troika. They go into Afghanistan, and on the US side, my understanding is Carter and maybe Brzezinski assumed this was a play to get all the way to the Persian Gulf.

What was the motivation to invade? If you were to identify some continuity of motivations between regimes, what did you find that was new about what was driving them?

Sergey Radchenko: I found some interesting things. We’ve had many people write about the Soviet invasion of Afghanistan, from Arne Westad to Rodric Braithwaite and many others. It was difficult to find anything new, but I did manage to discover some interesting materials from the fall of 1979.

At that time, there was already a conflict between Nur Muhammad Taraki and Hafizullah Amin in Kabul, but Amin had not yet killed Taraki. Taraki came to Moscow, and Brezhnev had a conversation with him — though “conversation” is in quotation marks because Brezhnev was in no position to have any productive dialogue. He does offer a warning to Taraki, “Look, when you go back, be careful because there’s a problem in your senior party ranks.” Taraki says, “Leonid Illyich, don’t worry about it. Everything is fine."

He goes back, and of course, he’s arrested by Amin’s men and ultimately put in prison and murdered on Amin’s orders.

What we see then is Soviet leadership thinking about how to respond. Their first instinct is actually to work with Amin despite the fact that they consider Taraki’s death an absolute slap in the face, a betrayal of the USSR. They feel they could work with Amin because he’s surrounded by “pro-Soviet people,” people who had studied in the USSR. Brezhnev writes about it in some exchange of memoranda at the senior leadership level, which I discuss in the book. This suggests they might continue working with Amin.

But by December 1979, they decided that they had to remove him. I ask in the book why this happened. There are different possible explanations for Soviet involvement — perhaps the Soviets were just adrift with nobody making foreign policy anymore, but I don’t think that fully explains it.

A more interesting explanation, which I highlight in the book, is that they worried Amin would “do a Sadat” on them. They had lost President Sadat in Egypt, who was supposed to be a pro-Soviet client, but they dumped him because he decided to align with Kissinger and Nixon. The Soviets got outplayed in Egypt.

After losing Egypt, the Soviets viewed almost every country in the Middle East as potentially another Egypt, and Afghanistan fell into that category. They grew suspicious of Amin, thinking he had ties with Americans. They received information that he was in contact with Americans and concluded he was potentially pro-American and could sell them out. They feared Americans would then establish a presence in Afghanistan, creating a strategic problem for the Soviets. They decided to act primarily because they didn’t want Amin to become another Sadat.

Jon Sine: Let’s transition to Gorbachev — and start with the Gorbachev-Trump analogy.

DOGE is American perestroika, at least in the charged anti-bureaucratic approach. Under Gorbachev, the Soviets decreased employment in the ministries and party personnel by something like 30 to 40%, which the current administration would certainly aspire to with the civil service in the US.

You have Gorbachev willingly giving up the Warsaw Pact, which some might analogize to Trump’s stance on NATO. Ultimately, as Axios reported, you have Trump’s desire for a Nobel Prize and the prestige and legitimation of doing something great for a foreign audience. People criticized Gorbachev for similar reasons when he wrote his 1987 book, Perestroika and New Thinking. He had it translated immediately into English, and it was written for a US publisher.

Sergey Radchenko: The book was also published in the USSR. Historical analogies are limited to some extent, but I see some interesting parallels.

One counterpoint I would offer is that the Soviet economy was basically going to hell in the 1980s, and they knew it. They knew they were losing the Cold War and that the promise of deliverance for the Soviet people was just a fake promise. The American economy, if you look at productivity, investment in technology, R&D, and so forth, is far beyond anything else anybody in the world can offer.

The question is one of necessity. I understand some people will say, “America has to carry out reforms because of the national debt” or other issues. In the Soviet case, perestroika was absolutely necessary, and they knew they had to do it. They knew it in the ’60s, but they didn’t act then because they craved stability under Brezhnev and were trying to avoid political upheavals.

They were also, to a certain extent, bailed out by the price of oil and the discovery of oil in Western Siberia, which they could sell for hard cash. This helped feed the Soviet people because they could import grain and some technologies. But they knew the system wasn’t working. When Gorbachev came to power, he had to begin doing something because doing nothing was not an option.

Subscribe now

Jordan Schneider: You write:

Gorbachev had everything in 1985 — an empire (however decrepit), an ideology (however stale), and above all, an office with truly awesome power. What he did not have was greatness, as he chose to understand it, greatness before history. He pursued that fleeting dream for himself and his country all the way to the famous Pizza Hut ad.

When you compare what Gorbachev and these other historical figures define as greatness, and what Trump defines as greatness, you see that the Trump definition is smaller, more personal — “People respect me, and speak to me nicely in the Oval Office.” All these other leaders had a vision of being grand historical figures who achieved something monumental.

There are parts of Trumpism that claim to revive America, defeat wokeness, and bring back ideals. Every once in a while, he’ll mention something about manufacturing, but “Make America Great Again” is much more about Trump personally than it is about a national vision of prestige and greatness.

Sergey Radchenko: It’s related to a certain extent. Soviet leaders desired greatness for themselves and for their country. This book is applicable to almost any other would-be great power or superpower. To a certain extent, it can be read as an allegory of the United States. This is how John Lewis Gaddis saw it in his review in Foreign Affairs — he thought that between the lines, the United States was clearly lurking there, really telling the American story, not just the Soviet story.

The question of legacy is super interesting and important because, as you say, Jordan, Gorbachev is at the helm of a superpower. He has all the power concentrated in his hands, and yet he wants something more. It’s similar to Mao Zedong in 1965. He had all the power, he could do anything. Even after he got rid of people he thought were conspiring against him by 1966, he continued the Cultural Revolution. For what? For legacy, for greatness before history.

If you think about Putin and his invasion of Ukraine, he has all the power, and yet he’s invading Ukraine. For what?

For greatness before history, the way he understands greatness.

If we return to Gorbachev once again and ask, “Fundamentally, why did he do it?” The answer is because he wanted something greater than what he had. He wanted to be remembered as a person who would bring the Soviet idea to life for the first time because he thought his predecessors never made it work. He would make it work, make it globally applicable, end the Cold War — this was his mission. That, I believe, is very important.

Jordan Schneider: We talked in the first episode about national ambition — is it a gas or is it a solid? Will it expand until it hits another obstacle that contains it, or does it have some natural limits? It’s interesting to analogize that not at a national level, but at a personal level.

When you’re in power for 15 years, you get bored. You’re probably already a bit of a gambler if you were able to make it all the way up the system. You just want more. This is why term limits are important — leaders tend to go a little cuckoo after too long. Either they go senile like Brezhnev — who thankfully didn’t start World War III because he was worried about his bladder — or they do something like Putin in Ukraine. It’s a scary thing to contemplate.

Sergey Radchenko: I agree. It’s fascinating to consider how these leaders, perhaps out of boredom, perhaps because they have nothing else to do, take dramatic actions. That might sound ridiculous, but speaking of boredom, Putin was once asked about Ukraine in one of his recent press conferences, and he essentially said, “We were kind of bored, and we decided to do it.” I’m paraphrasing, but he literally mentioned the word “boredom” in his comments on why he invaded Ukraine.

That sounds crazy, but if you’re a leader with all the power in your hands, you need something else. Schopenhauer in the early 19th century reflected that one of the major problems human beings face is generally the problem of boredom. They don’t know what to do with themselves. Now multiply that with absolute power, and you get people like Gorbachev who think, “Why not do something so great that everybody will remember me?” And well, he certainly succeeded.

Jon Sine: There’s a hedonic treadmill effect when it comes to status. You’ve achieved so much and yet, when you look back once you’ve achieved it, you realize you could achieve more, and the last thing you accomplished no longer seems as good. Now you’re wondering, “How can I really leave a legacy that my children and my children’s children will remember?”

What better than changing your borders, acquiring Greenland, getting back Panama?

Sergey Radchenko: Exactly. Or launching Perestroika. That’s why Gorbachev’s book which you referred to, the one published in the West, was actually subtitled “Perestroika For Our Country and For The World.” He was trying to restructure the entire world, not just his own country. That’s the extent of his ambition.

Humiliation and Containment

Jordan Schneider: Let’s talk about the dance that the Europeans and the Americans did with Gorbachev. The goal was to get the Soviets to understand that their relative power was decreasing, and that America and its allies had won the economic, science, and technology competitions. But you also want to allow a leader like Gorbachev to pursue his vision without risking something like a Stalinist revanchism, which wasn’t totally out of the cards.

Could you talk about how the West managed Gorbachev’s moment of defining national greatness as Perestroika rather than, say, conquering Germany?

Sergey Radchenko: First of all, the Americans were quite worried about Gorbachev to begin with, and that’s a well-known story. They were concerned because they thought he could actually succeed in what he claimed to be doing — reinventing the Soviet idea and making the Soviet Union a much more serious strategic competitor to the United States.

Some thought Gorbachev wasn’t for real and his reforms would ultimately be undone. Then, a moment came when American leaders thought they could reach out to Gorbachev in the name of international peace and reach some agreements.

Reagan was the key person here because he also believed in the importance of avoiding nuclear war and felt the great responsibility that was on his shoulders. Many people in the United States like to talk about Reagan “winning the Cold War” and how tough he was on the USSR. For me, the real Reagan was the one who believed that nuclear war had to be avoided and that we needed to talk to the Soviets.

As Reagan said at one point, Soviet general secretaries kept dying on him, but then he finally had a partner in Gorbachev with whom he could talk. They first met in Geneva in 1985, then in Reykjavik. There came a remarkable moment in Reykjavik where they discussed abolishing nuclear weapons altogether. It was “almost decided” to the horror of Reagan’s advisors. Here was an interesting moment when an American leader thought, “What if we actually try to play along and see how far we can get? Maybe we can actually change the course of history.” Reykjavik represented that possibility.

Jordan Schneider: Going all the way back to Stalin and Truman, you have this quote from Henry Kissinger who wrote in a 1957 book, “The powers which represent legitimacy and the status quo cannot know that their antagonist is not amenable to reason until he has demonstrated it. And he will not have demonstrated it until the international system is already overturned.”

We’ve got our answer with Putin, but we don’t quite have our answer yet with Xi’s China. This presents a very difficult choice — on the one hand, if Stalin had turned out to be reasonable, we could have had a different timeline. But if not, the price for running that experiment would have potentially been the Soviet conquest of Europe or even the world.

How are we supposed to think about that dilemma sitting here in 2025, imagining the future of US-China relations?

Sergey Radchenko: Jordan, that’s a great question. The fundamental problem is we don’t know what the other side is really thinking. The other side may not even know what they ultimately want. This is the case with Stalin. Even with the passage of all these years, we don’t know what Stalin’s ultimate ambitions were. Was he limited in his appetites? Would he stop, or unless he met with counterforce, would he keep pushing? Would he keep going until he conquered Europe, and from Europe jump over to conquer Asia and the world?

It sounds fantastic, but “appetite comes during eating” as the French say (“l'appétit vient en mangeant”). If you’re an American policymaker faced with this situation, you’re thinking, “How do we stop that? What is the most reasonable policy we can adopt?” It seems the most reasonable approach is the policy ultimately adopted — containment. You push back on expansionist desires, and the other side has to take this into account. They are deterred and limit their ambitions, but as a result, a struggle unfolds that becomes a prolonged confrontation like the Cold War.

It’s a sad situation, but maybe inevitable and unavoidable. It’s like choosing between two evils. On one hand, containment and counter-containment lead to a Cold War and could potentially lead to a hot war. On the other hand, doing nothing could expose you to a situation where you’ll basically have to accept that the other side dominates everything.

It’s the same with China. The fundamental problem we face today, even with Russia as well, is that we simply don’t know what the other side wants. Does Xi Jinping want to overturn the existing order, or is he just trying to change China’s position within this global order? That question has been debated for 20 years, even before Xi Jinping.

We don’t know the answer. Some people say, “We know Xi Jinping is trying to overturn the world order because here’s the evidence.” I would say, “I’m looking at this evidence, and I’m not 100% convinced.” We simply don’t know whether Xi Jinping himself knows what he wants to do.

Under those circumstances, what’s the best policy to pursue? The policy is probably a combination of containment, firmness, plus clear signaling — “We see what you’re doing. This will be our response.” No jumping around and doing impulsive things, which American foreign policy sometimes tends to do. We’ve seen that with Trump, but also with previous administrations.

With China, consider Nancy Pelosi’s ill-advised visit to Taiwan. What did it achieve? It was very provocative and completely useless in many ways. As a result, we had a breakdown of communication at senior levels between Chinese and American militaries. Did it benefit anyone? No. What was the point? There was no point.

We have to be firm, maintain dialogue, and signal to the other side that we see what they’re doing and are preparing certain countermeasures. But if they back off at the right moment, we will also not proceed with those countermeasures. I think that’s the way to mitigate great power confrontation in this potentially new Cold War that is unfolding before our eyes even today.

Jordan Schneider: I want to apply the lessons of prestige to the US-China relationship, and Putin as well. Are leaders satisfied by the rate of change of prestige or by the absolute value of prestige? Are there ways to give prestige that don’t fundamentally compromise core national interests and power? It seems that prestige is often in the eye of the beholder, and America can give prestige in many different ways — some costly, some much less so.

Sergey Radchenko: Core interests for prestige certainly exist. In my book, I discuss Poland and how it was central to Stalin’s security concerns in Europe. No matter how much prestige he would have received from the allies, he was determined to do what he wanted in Poland because he considered it a fundamental red line core to his security interests. You might say this also applies to Putin or Xi Jinping today.

On the other hand, we sometimes underestimate the importance of recognition, respect, and acceptance. We also tend to underestimate how our actions can humiliate the other side and provoke adverse reactions we would rather have avoided.

Consider President Obama’s rhetoric about Putin being “the kid in the back of the classroom,” which reportedly outraged him. You might ask whether Putin would have still invaded Ukraine if Obama hadn’t made those remarks. Perhaps he would have because Ukraine is central to his neo-imperialist vision of Russia, or perhaps not. It’s a counterfactual — we simply don’t know.

The question is whether he feels he has standing and respectability. I believe it ultimately does matter. The same likely applies to the Chinese leadership.

It’s important to note that, despite our strategic rivalry with China and difficult relationship with Russia, these countries are ruled by autocrats who sometimes make decisions impulsively, for no particular reason, and occasionally in ways that clearly contradict their country’s national interests. They do this because they feel insulted or humiliated. We should not underestimate those sentiments. In dealing with these countries, we should be careful — respectful but firm.

Jordan Schneider: Perhaps even more famously, Barack Obama at the 2011 White House Correspondents’ Dinner made fun of Donald Trump for the birther conspiracy, and some trace Trump’s desire to run for president back to being humiliated on national television.

Sergey Radchenko: That’s a great counterexample.

Jordan Schneider: Maybe this is the 4D chess with Trump — we’re going to stop calling Putin a dictator and say he’s the best thing since sliced bread. Will that make him pull out of Ukraine?

Sergey Radchenko: Maybe not, but it’s a cumulative effect. Over time, these things matter. It’s not as if we suddenly say, “We love you, come back to the G7, let’s have a good relationship,” and he’ll respond, “Yes, I’ll pull out from Ukraine.” We’ve already turned that corner, and it’s very difficult to go back. But on the road to confrontation, these factors do matter, in my view — perhaps not decisively and not always, but they do matter.

Radiooooo: A Cosmopolitan Dreamland

More and more of my listening this year has shifted away from Spotify. As big a fan I am of AI, I’m sick of Spotify degrading artists and albums in favor of mixes and playlists by ghost musicians. In its place for me has come the magical website and app Radiooooo.

Radiooooo takes human-submitted tracks, categorizes them by country and decade, and lets you explore. On their whimsical global map, you can either let their algorithm take you on a world tour, or you can decide that today’s the day to spend an hour with the tunes of pre-revolutionary Iran.

Most of the world maps I look at reflect geopolitical competition, but Radiooooo’s map flips that on its head, using music across time to show how cross-pollinated we are. The selection is biased towards selections that show global influence, and it is so cool seeing that global influence! Japan had excellent do-wop in the 1970s! A Swedish singer in 1959 made a song about Puerto Rico! A calypso song from Hong Kong in 1960! Soviet crooners from the early Brezhnev era! Radiooooo is a life-affirming reminder that joy and creativity persist even in decades of dictators and war.

Beethoven provided my other favorite listening experience of the year. Norman Lebrecht’s Why Beethoven: A Phenomenon in One Hundred Pieces is a brilliantly written collection of short essays, which introduce, in some creative oblique way, a Beethoven composition in just a few pages.

Chinese TV

再见爱人 See You Lover — China’s “Road Rules for Divorce” show — is a reality show where three celebrity couples all married for at least ten years and on the verge of divorces take an 18-day road trip together. It is some of the most gripping content I’ve ever seen. Discussed here and with on the podcast episode below. It’s the year’s only must-watch piece of Chinese content.

Over the course of the show, you get to watch modern Chinese women wake each other up to the fact that they can demand more from their neglectful and abusive husbands through an interplay of independent earning power and global feminist ideas (filtered primarily through Japanese writers). Score: 9.8/10

欢乐家长群 Growing Together — this light series is about four couples in Beijing raising young children, and the drama that takes place through a parent’s WeChat group. Contemporary marital drama is much more engaging in 再见爱人, but this show is a sanitized window into how rich parents try to raise their kids in a post-cram school era where prioritizing self-actualization holds some weight relative to academic achievement. Score: 7.5/10

I got four episodes into 大明王朝1566 before getting overwhelmed. It’s quality content that requires real attention, which paternity leave did not allow for. Will try again this year!

山花烂漫时 She and Her Girls [they should just let me name these shows, this is awful] is a dramatized true story of a teacher who founded an all-girls high school in a poor rural village.

It’s a high quality drama and illustrates why top flight Chinese dramas have a hard time finding footing internationally. In one episode, the teachers start jumping ship because of bad working conditions and low pay. The only ones who stick around are the Party members — who use the power of red songs to get their kids to study. It’s a lot to stomach. Chinese audiences understand that elements like these are non-negotiable if you want to make a show showcasing the reality of modern poverty, but it causes global audiences to tune out.

“In the war against the Japanese, as long as there is one Party member, there’s no way that group would lose ground. Today I learned we have six teachers who are Party members. Could it be that we lose ground for the Party at this women’s high school? The long march was so long, I know this semester will be hard, but not that hard! This semester let’s all review again the Party Membership Oath. Before we were colleagues, now we are comrades. 之前我们是同事，现在我们是同志!”

Ep 5 is the standout episode for me where the teachers go out into the countryside to recruit students only to encounter desperate poverty preventing these girls from studying. It’s a refreshing departure from the normal poverty porn you see in food shows like Bite of China.

Other themes include:

• A morally complex protagonist! Usually in a propaganda-boosted storyline like this, with someone sacrificing their personal life for poor people, the main character gets turned into Mother Teresa. In this show the teacher is proud — and resentful of the students who still rebel when given the opportunity.

• The real importance of the gaokao — ep 19 opens with a dream sequence of everyone getting super high test scores (which doubles as satire of traditional gaokao shows), and closes with a speech by the lead teacher about why the gaokao isn’t absolutely everything. The show illustrates how deeply that exam pressure has traumatized these girls, and by the next episode the head teacher is back to stressing everyone out about test scores. College is the only social mobility path for these rural youth, but if you don’t get accepted anywhere, at least you didn’t have a child at 14!

• Alcohol — being raised by alcoholic parents contributes to a lot of the girls’ misery, but the founder of the school still uses booze to bond with teachers and schmooze donors and officials.

• Bureaucracy — we run into unhelpful lower level officials, only to of course find an understanding higher-up who swoops in to save the day once the protagonist remonstrates and proves their worthiness.

Other TV

Culinary Class Wars — Koreans with nice “Great British Bake Off” vibes cooking food that’s novel to me. Funny how the “Chinese-style” Korean chefs all have a very flat interpretation of Chinese cuisine. Most remarkable is the star judge, Korea’s first three star chef. He has a crazy backstory:

Anh was born in South Korea but moved to the United States in 1993 at the age of 12. He grew up in California where his parents ran a Chinese restaurant. He enlisted in the US Army after high school, and after the September 11 attacks and the start of the Iraq War, asked to serve in Iraq. Anh served as a mechanic, where he helped fuel combat vehicles like helicopters and tanks.After his Army service, Anh was set to attend mechanic school to become a Porsche mechanic. After seeing a group of culinary school students walk by in chef coats, however, Anh changed course and enrolled in culinary school.

Score: 8.3/10

Industry Season 3 — this show is fun, but it’s no Succession. Score: 7.8/10

Sports

Most of the ‘make you gasp’ moments I have following professional sports are surprises about how awful things can get. On a weekly basis for years now, Josh Allen, QB of the Buffalo Bills, has delivered these moments. Watching him is balletic — his invincibility lets you forget all the crushed bodies left in the NFL’s wake.

Dipping into independent fan podcasts after big games or trades is one of my great pleasures. The best sports drama arc of 2024 were the Mets gloating at the Soto signing.

In an increasing number of sports, women’s competition offers the superior viewing experience. This was especially apparent watching the Olympics this year. For men’s indoor volleyball, the jump serves result in too many out serves and dead time, but the women’s game has more action and dramatic points. The same goes for women’s tennis, where Caitlin Clark is a highlight. However, the WNBA is still a little too slow for my taste.

I started playing soccer in adult leagues in NYC this year! It would be fun to get a ChinaTalk team going if there’s any interest…

Movies

We’ve since learned that ChinaTalk doesn’t only get people promoted, it also gets them cast in movies! Phil spent this fall as one of Adam Sandler’s four sons in Happy Gilmore 2. He’s also the lead in an excellent play currently on in Atlanta from March 29 to May 4th.

Billy Madison — the teacher deciding to go out with Billy Madison after he gropes her on the bus is a little bit of a stretch…

Happy Gilmore — I had more fun with this one than Billy Madison. Do they not make short dumb well-made movies anymore, or do I just not watch the ones that come out?

抓娃娃 ‘The Successor’ (2024) — this is one of China’s top grossing movies of the year. The Successor is a ‘Truman Show’ where rich parents decide to raise their child in poverty after spoiling their first kid while training him to take over the family’s business empire. It was interesting to see the backlash to tiger parenting across this movie and the parents WeChat group show. You can stream this movie on Iq.com with English subtitles! Score: 8.8/10

The Divorcee, 1930 — I went on a 1920s kick, which led me to Ursula Parrot’s epic 1929 novel Ex-Wife and its 7.1/10 movie adaptation. The best discovery was this labor of love website, pre-code.com, that I’m excited to dive deeper into.

Anora — I fell asleep in second act. Score: 7.6/10

Video Games

Elden Ring (Levels 60-85)

Parenting an infant considerably shrinks your physical universe — before I became a parent, I was traveling twice a month. Returning to Elden Ring helped me feel less claustrophobic as I made the transition to life within a 5-block radius.

This game is the perfect experience for the pre-VR/AI world, and a marvel of human creativity.

One day I went to the Met Museum after a few hours of playtime, and thought, “Elden Ring’s world is more beautiful and striking than half of these landscapes.” There are few experiences in content consumption quite as satisfying as beating an Elden Ring boss on the 7th try. There were a few times in the past year, in real-life contexts, that I opted to beat “a boss” creatively instead of beating my head against a wall trying the same thing over and over again. Perhaps you could credit my “Elden Ring brain” with helping me find lateral solutions to these problems.

Fromsoft’s triumph is a testament to the magic of maintaining a singular focus on honing a craft for decades. Sure, hardware could not handle something as ambitious as Elden Ring in the past, but the gameplay and design heights of Elden Ring would not be possible had the developer not already made six games in the ‘Soulslike’ genre.

Even if you’re not a big video game person, I’d really encourage you to try to experience the game for yourself. If there’s no chance you will, consider watching this video:

Magic the Gathering: Bloomburrow Quick Drafts

It’s dark magic. Doing drafts gives you slot machine randomness with just enough brain-engagement to trick you into thinking it’s an intellectual activity. Deck-building is genuinely interesting, but then you’re back to casino-like brainstem stimulation during actual gameplay. Generally, there are only one or two moments in each game where you’re rewarded for thinking. The beauty of playing on chess.com is that the apps have a direct feedback loop to teach you how to improve after each mistake you make, which doesn’t exist in MTG Arena.

Von Neumann has a great quote comparing chess and poker: “Chess is not a game. Chess is a well-defined form of computation. You may not be able to work out the answers, but in theory there must be a solution, a right procedure in any position. Now, real games are not like that at all. Real life is not like that. Real life consists of bluffing, of little tactics of deception, of asking yourself what the other man is going to think I mean to do. And that is what games are about in my theory.”

Maybe in-person magic has that in-person bluffing aspect coupled with the feedback of learning by chatting after the games — but playing online has left me hollow. After spending down $20 in gems, I blocked the game.

Black Myth: Wukong.

If you’re looking to appreciate the artistry of the game without having to git gud, check out this 4k bilibili playthrough.

Asian food in nyc superlatives

Read more

Part 2 of our interview with Shashank Joshi, defense editor at the Economist, and Mike Horowitz, professor at Penn who served as Biden’s US DAS of Defense for Force Development and Emerging Capabilities. Here’s part 1.

In this installment, we discuss…

• AI as a general-purpose technology with both direct and indirect impacts on national power,

• How AGI might drive breakthroughs in military innovation,

• The military applications of AI already unfolding in Ukraine, including drone capabilies and “precise mass” more broadly,

• Whether AGI development increases the probability of a preemptive strike on the US.

Listen to the podcast on iTunes, Spotify, YouTube, or your favorite podcast app.

I’m hoping to expand on this show with an interview series exploring AI’s impact on national security. Too often today, debates center on “superweapons” lazily pattern-matched to the nuclear era or go in circles on cyber offense vs defense. The goal instead is to repeat the exercise Dario did for biotech in Machines of Loving Grace: deeply explore the bottlenecks and potential futures across domains like autonomy, decision-support, stealth, electronic warfare, robotics, and missile defense. Guests will be engineers and technologists who can also explore second-order operational and strategic impacts.

But this needs a sponsor in order to happen! If you work at an AI firm, defense tech, VC, university or think tank and want to help facilitate the best conversations about the future of warfare, please reach out to jordan@chinatalk.media.

Military AI and the Drone Revolution

Jordan Schneider: Let’s talk about the future of war. There is this fascinating tension that is playing out in the newly national security-curious community in Silicon Valley where corporate leaders like Dario Amodei and Alex Wang, both esteemed former ChinaTalk guests, talk about AGI as this Manhattan Project-type moment where war will never be the same after one nation achieves it. What’s your take on that, Mike?

Michael Horowitz: There’s a lot of uncertainty about how advances in frontier AI will shape national power in the future of war. I’ve been, historically, extremely bullish on AI from a national defense perspective. I remember when Paul Scharre and I were in a small conference room at CNAS with basically everybody in Washington D.C. who cared about this. Now it’s obviously attracting much more attention.

But I think the notion that AGI will inherently transform power in the future of war makes a couple of questionable assumptions. The first is that AGI is binary and immediately causes a jump in capabilities, which essentially means that you can then solve all sorts of problems you have been trying to solve that you couldn’t solve before.

That might be true. It also might be true that you have continuing growth in AI capabilities that may or may not constitute AGI. In that case, you never have one specific moment, yet you still have ever-increasing frontier AI capabilities that militaries can then potentially adopt.

This other assumption is that technical breakthroughs are the same thing as government adoption, which the history of military innovation suggests is incorrect.

I worry that US companies will lead the world in AI breakthroughs, but the US Government will lag in effective adoption due to legacy bureaucracy, budgeting systems, and the relationship between the executive and Congress.

Maybe the PRC will get there later, but adopt the upside faster.

Jordan Schneider: What is the AI eval that would convince Mike Horowitz that this is the next stealth bomber, or the next nuclear weapon?

Michael Horowitz: I just think that’s the wrong way to think about it. AI is a general purpose technology, not a specific widget. If what we are looking for is the AI nuclear weapon or the AI stealth bomber, I think we’re missing the point.

When general purpose technologies impact national power, they do so directly and indirectly. They do so directly through, “All right, we have electricity, now we can do X,” or, “We have the combustion engine now we can do X.” They do so indirectly through the economic returns that you get, which then fuel your ability to invest in the military and how advanced your economy is. AI is likely to have both of those characteristics.

The thing that’s not entirely clear yet is whether there’s essentially a linear relationship between how advanced your AI is and what the national security returns look like.

Jordan Schneider: Just to stay on defining terms, the direct applications we’re talking about, like the AI for science, are still in the very early phases. There’s a really fun book by Michael O’Hanlon called Technological Change and the Future of Warfare, that includes this cool table of all of the different vectors on which technology can get better over a 20-year horizon.

Show me some crazy material science breakthroughs that you can put in weapon systems, and I will be convinced that this stuff is really real and going to matter on a near-term horizon on the battlefield.

But I think the Dario Amodei framework strikes me as really not grappling with the challenges both on the scientific side as well as on the adoption side. Maybe Shashank, before we do adoption, anything on what this can potentially unlock that we would want to?

Shashank Joshi: We could split the applications of that general-purpose technology up a million different ways. The way I have tended to do it in my head is thinking about insight, autonomy, and decision support.

Insight is the intelligence application. Can you churn your way through satellite images? Can you use AI to spot all the Russian tanks?

Autonomy is, can you navigate from A to B? Can this platform do something itself with less or no human supervision or intervention? The paradigmatic case today, which is highly impactful, is terminal guidance using AI object recognition to circumvent electronic warfare in Ukraine.

The third interesting thing is decision support. This includes things that nobody really understands in the normal world, like command and control. It’s the ability of AI to organize, coordinate, and synchronize the business of warfare, whether that’s a kind of sensor-shooter network at the tactical level for a company or a battalion, or whether it’s a full theater-scale system of the kind that European Command, 18th Corps, and EUCOM has been assisting Ukraine with for the last three years.

This involves looking across the battlescape, fusing Russian phone records, overhead, radio frequency, satellite, IM satellite returns, synthetic aperture radar images, and all kinds of other things into a coherent picture that’s then used to guide commanders to act more quickly and effectively than the other side. That’s difficult to define. But if we’re talking about transformative applications, that is really where we need to be looking carefully.

Michael Horowitz: I agree. General Donahue is a visionary when it comes to what AI application can look like for the military today, and in trying to at least experimentally integrate more cutting-edge capabilities.

To the Dario point though, it’s all a question of timeframe and use cases. If you imagine AGI as instantly having access to 10,000 Einsteins who don’t get tired, then that’s going to lead to lots of breakthroughs that will generate specific use cases.

These could lead to new material science breakthroughs that decrease radar cross-sections, new advances in batteries that finally mean the dream of directed energy becomes more of a reality, or advances in sensing in the oceans that create new ways of countering submarines. It could lead to all sorts of different kinds of things. The challenge is it’s difficult in some ways, ex-ante, to know exactly which you’ll get when.

Jordan Schneider: Perhaps Dario would say that your framework, Shashank, is weak sauce and he’s talking about an entirely new paradigm.

Which of these applications are currently being developed in Ukraine?

Shashank Joshi: The autonomy piece is super interesting because of the pace of change. To sum this up, when I was talking to Ukrainian first-person view (FPV) strike operators, they were saying that if you are a member of an elite unit with loads of training, you can get a hit rate of up to 70-75%. But if you’re an average strike pilot, this is not easy. Sticking those goggles on, navigating this thing — you don’t know when the jamming is going to kick in and cut the signal. You have to get it just right, and you’re getting like 15-20% hit rates maybe.

What I am seeing with the companies and entities building these AI-guided systems for the final 100 meters, 200 meters, 300 meters, and increasingly up to 2 kilometers in some cases, is that the engagement range is going up. You can hone in on the target beyond the range of any plausible local jamming device. That’s a huge deal. More importantly, the hit rate you’re getting is 80% plus. That’s phenomenal. That changes the economics, the cost per kill — that changes the economics of this from an attrition basis.

There are all these interesting ripple effects. You can achieve this with like 30 minutes of training. Think about what that unlocks for a force, particularly sitting here in Europe where we have these shrunken armies with no reserves, with the manpower requirements as well as the training times to bring new people in when you have attrition in a war in the first round.

Subscribe now

This little tactical innovation — terminal guidance, AI-enabled — looks very narrow, but it has all these super interesting and consequential ripple effects on the economics of attrition, the cost per kill, lethality levels, the effectiveness of jamming, and on manpower and labor requirements. That’s why it’s so important to get into the weeds and look at these changes.

Michael Horowitz: We’re seeing at scale something that we kind of thought might happen, but it just always been theoretical rather than something real. The argument for why you would need autonomy to overcome electronic warfare has been obvious for decades. When they were questioning if the technology is there or if we want to do it this way, there were different kinds of approaches.

What we’ve seen is that when you are fighting conventional war at scale, if you want to increase your hit rate and overcome jamming when facing electronic warfare, you can update software to try to counter the jamming. You can try to harden against jamming, and although it increases costs, you can use different concepts of operation to try to get around it, to sort of fool the local jammers.

But to the extent that autonomy becomes a hack that lets you train and operate systems more effectively in much less time — that’s a game changer, and it’s not one that we should expect to be confined there. Imagine all of the Shaheds that Russia fires at Ukraine with similar autonomous terminal guidance out to a couple of kilometers. Imagine all sorts of weapon systems with those kinds of capabilities. We’re seeing this at scale in Ukraine in a way we just had imagined.

To be clear, it’s not just in the air. Let me now just give my 15-second rant on the term “drone.” We are currently using the term “drone” to refer to a combination of cruise missiles, loitering munitions, ISR platforms, and uncrewed aerial systems that themselves launch munitions. All of those are getting called drones right now, even though we actually have correct names for them. It might be helpful if we used those names since we’re talking about different capabilities, essentially. But yeah, plus one to everything.

Shashank Joshi: Mike, do you want me to start using terms like UAS, UUVs? I’m going to get sacked if I start using all those acronyms.

I’d like to ask you about something since you’ve raised the issue of different domains. One of the questions I often get asked by readers is: where are all the drone swarms? Where are the swarms that we were promised? Maybe this is a kind of ungrateful thing because we bank a bit of technology and we are desensitized to it and then we forget everything else.

What interests me is, when writing about the undersea domain a while ago and submarine hunting, I was struck by how difficult it is — this is obvious physics — to communicate and send radio frequency underwater. Radio waves don’t penetrate water very much, if at all. Acoustic modems and things like that are very clunky. So the technologies we have relied on for things like control signals, navigational signals, oversight in the air domain operate very differently in places where signals don’t travel as much — the curvature of the earth or in the water. Do you think that uncrewed technology and autonomy operates in some kind of fundamentally different way in those domains or will it be less capable in those places?

Michael Horowitz: That is a great question. Let me give you a broad answer and then the answer to the specific undersea question.

We’ve essentially entered the era of precise mass in war, where advances in AI and autonomy and advances in manufacturing and the diffusion of the basics of precision guidance mean that everybody essentially can now do precision and do it at lower cost. This applies in every domain — it applies in space, in air for surveillance, in air for strike, to ground vehicles, and can apply underwater and on the surface now. The specific way that it plays out will depend on the specifics of the domain and on what is most militarily useful.

If the question is “Where are the swarms we were promised?” and what we end up with is a world where one person is overseeing maybe 50 strike weapons that are autonomously piloting the last two kilometers toward a target, there may be actually military reasons why we don’t want them to communicate with each other. If they communicated with each other, that would be a signal that could be hacked or jammed, which then gets you back into the EW issue that you’re trying to avoid. There’s an interaction in some ways between the “swarms we were promised” and some of the ways that you might want to use autonomy to try to get around the electronic warfare challenge.

This points to the huge importance of cybersecurity in delivering essentially any of this. Part of the issue is that swarms potentially are vulnerable to some of the same issues that face FPV drones and other kinds of systems in different kinds of strike situations. It wouldn’t be surprising to me if you then see a move more towards precise mass in the context of autonomy without swarming in a world where you think that you’re getting jammed.

Now underwater, you absolutely have the physics issue that you’ve pointed out — communication underwater is just more difficult. To the extent that something like swarming requires real-time coordination, that becomes even more difficult the further away things are from each other. It wouldn’t be surprising then that the underwater domain would be challenging here.

To take it back to the AGI conversation we were having, two notes are relevant here. One, the “where were the swarms we were promised” question reminds me that how we define artificial general intelligence often is a moving target. We’re constantly shifting the goalposts because once AI can do things, we call it programming. Artificial general intelligence is always the thing that’s over the horizon.

Going back to my belief that there might not be one AGI moment, it reflects that way that the definition of AI has tended to be a moving target. But specifically, if you’re in the “AI will transform everything” paradigm, one of the things you would probably try to use AGI to do that would have transformative impact would be to solve some of these communication issues in the undersea domain that can potentially limit the utility of uncrewed systems in mass undersea. That’s an example essentially of a science problem that then maybe major advances in AI help you address when you have paradigm-shifting AI.

Debating China War Scenarios

Jordan Schneider: I want to talk about two odd theories for why a US-China-Taiwan war could kick off. One is China’s dependence on TSMC, and the other is this idea that if one side is close to AGI, then the other would do a preemptive strike to stop their adversary. What do you think about these scenarios?

Michael Horowitz: Those are great questions. A lot of what we know from the theory and reality of the history of international relations and military conflicts suggests that war in either case would be pretty unlikely.

Let me start with the Taiwan scenario. I am extremely nervous, to be clear, about the prospect of a potential conflict between China and Taiwan. There is real risk there. But the notion of China essentially starting a war with Taiwan over TSMC would be kind of without precedent. Put another way, there are lots of paths through which we could end up with a war between China and Taiwan. The one that keeps me up at night is not an attack on TSMC.

Jordan Schneider: Ben Thompson just wrote a piece that defines China’s reliance on Taiwanese fabs as an important independent variable in Beijing’s calculation of whether or not to invade. How do you think about that line of argument in a broader historical context?

Michael Horowitz: The best version of the argument, if you wanted to make it, is probably that if China views itself as economically dependent on Taiwan, it would then seek to figure out ways to get access to the technology that it needs from Taiwan. You could imagine that effort happening in a couple of different ways.

One is to mimic what TSMC does, which obviously they’re already attempting to do. Another would be attempting to coerce Taiwan to get better access to TSMC. But starting a war with Taiwan where tens of thousands, if not hundreds of thousands of people are likely to die, and which could trigger a general war with the United States and other countries in the Indo-Pacific over the fabs — I think that’s relatively unlikely since China would have lots of other ways to try to potentially get access to chips that they need.

Jordan Schneider: You keep coming back to the straw man of going into TSMC to take the chips. But there’s another line of argument that as long as China still needs TSMC and is able to buy NVIDIA 5-nanometer or 3-nanometer NVIDIA chips and needs the output from those TSMC fabs to run its economy in a normal, modern way, then that would drive down the likelihood of China wanting to start a conflict.

Michael Horowitz: There’s certainly an argument in that direction, which is to say that economic dependence could generate incentives to not start a conflict as well. My belief tends to be that the probability of war between China and Taiwan will be driven by broader sociopolitical factors.

Jordan Schneider: I would also agree. Putting on my analyst hat for a second, I can think of several factors that are orders of magnitude more important to China than TSMC in deciding whether to invade — domestic Chinese political dynamics, domestic Taiwanese politics, the perception of America’s willingness or Japan’s willingness to fight for Taiwan.

How about the preemptive strike over AGI?

Michael Horowitz: The argument is insufficiently specified, and I will say my views on this could change. For a situation in which, say, China would attack the United States because it feared the United States was about to reach AGI, that presumes three things.

1. It presumes that AGI essentially is a finish line in a race — that it’s binary and once you get there, there’s a step change in capabilities.

2. It presumes that there’s no advantage to being second place, and that step change in capabilities would immediately negate everything that everybody else has.

3. It assumes that advances are transparent, such that the attacker would know both what to hit and when to hit it to have maximum impact.

There are a lot of reasons to believe that all of those assumptions are potentially incorrect.

It’s not clear, despite enormous advances in AI that are transforming our society and will continue, that there will be one magical moment where we have artificial general intelligence. Frankly, the history of AI suggests that 15 years from now we could still be arguing about it because we tend to move the goalposts of what counts as AI. Since anything we definitively have figured out how to do, we tend to call programming and then say that it’s in support of humans.

It’s also not clear that these advances would be transparent and that countries would have timely intelligence. You need to be not just really confident, but almost absolutely certain that if somebody got to AGI first, you’re just done, that you can’t be a fast follower, and probably that it negates your nuclear deterrent.

If you believe that AGI is binary, that if you get it, it negates everything else, and that there’ll be perfect transparency — in that case, maybe there would be some incentive for a strike. Except that military history suggests that these are super unlikely.

What we’re talking about in this context is a bolt from the blue — not the US and China are in crisis and on the verge of escalation and then there’s some kind of strike against some facilities. We’re talking about literally being in steady state and somebody starts a war. That’s actually pretty unprecedented from a military history perspective.

Leaders tend to want to find other ways around these kinds of situations. If you even doubted a little bit that AGI would completely negate everything you have, then you might want to wait and see if you can catch up rather than start a war — and start a war with a nuclear-armed power with second strike capabilities. It’s so dangerous.

Jordan Schneider: I am sold by arguments one and three. If the story of DeepSeek tells you anything, it’s not even fast following like three years with the hydrogen bomb in the Cold War; it’s fast following like three months with a model you can distill.

If it’s not a zero-to-one thing, then maybe the more relevant data point is Iran and Israel in the 2000s and 2010s. You don’t literally have missiles being fired and airstrikes, but you have this increasingly nasty world of targeted assassinations and Stuxnet-like hacking of facilities.

What is now a happy-go-lucky world in San Francisco could become a lot more dark and messy. Mike, what could trigger that potential timeline?

Michael Horowitz: Let me be more dire and ask, what’s the difference between that and the status quo?

Obviously there’s a difference if we’re talking about assassination attempts and those kinds of things. But every AI company around the world, including PRC AI companies, is probably under cyber siege on a daily basis from varieties of malicious actors, some of them potentially backed by states trying to steal their various secrets.

To me, this falls into a couple of categories. One is cyber attacks to steal things — hacking essentially for the purposes of theft. A second would be cyber attacks for the purposes of sabotage, like a Stuxnet-like situation. A third would be external to a network, but physical actions short of war — espionage-ish activities to disrupt a development community.

On the cyber attack aspect, there is a tendency sometimes to overestimate the extent to which there are magic cyber weapons that let you instantly intrude on whatever network you want. Are there zero days? Yes. Are cyber capabilities real? Yes. Many governments, including the United States government, have talked about that, but I’m not sure it’s as easy to say “break into a network” that is, to be clear, pretty hardened against attack, and just flip a switch like, “oh, today we’re going to launch our cyber attacks."

There are effectiveness questions about some of those things. But also those networks are constantly being tested.

Stuxnet was really hard to achieve. Stuxnet is probably the most successful cyber-to-kinetic cyber attack arguably in known history. It’s this enormous operational success for Israel against Iran.

Jordan Schneider: But the difference with Stuxnet versus what we’re talking about, is that a data center in Virginia or Austin is much more connected to the world. They hire janitors. It’s not like in a bunker somewhere.

Michael Horowitz: Those are more accessible, but there are also more data centers. Targeting any one data center in particular is not likely to grind all AI efforts to a halt.

Frankly, if there is one AI data center that is widely regarded as doing work that will be decisive for the future of global power, that’s going to be locked down. The company will have incentives to lock it down, just like defense primes have incentives to lock their systems down, even if we’re not talking about defense companies. Companies like Microsoft and Google have incentives to lock down non-AI capabilities as well.

My point isn’t that there won’t be attempts and even that some of those attempts won’t succeed, but there’s sometimes a tendency to exaggerate the ease of attack and its structural impact. In a world where we’re talking about hitting a very accessible facility in Virginia, that means there’s probably similar accessible facilities in other places that also can potentially do the job.

Now the toughest scenario is the espionage one where you’re talking about essentially covert operations targeting companies. It wouldn’t surprise me if some of those companies are intelligence targets for foreign governments. The challenge analytically is that these arguments quickly enter the realm of non-falsifiability.

If I tell you that I think this kind of espionage or that kind of espionage wouldn’t be that likely, you could say, “Well what about this?” We’re not going to be able to resolve it with facts. Non-falsifiable threat arguments make me nervous analytically. Maybe this is the academic in me that makes me want to push back a little bit because I feel like if an argument is legitimate, we should be able to specify it in a clearly falsifiable way.

Jordan Schneider: Like what? Give me the good straw man of that.

Michael Horowitz: The best straw man argument would be if you could basically demonstrate that the PRC is not really trying to target for collection varieties of AI companies and that it would be relatively easy for them to do so. That would raise the question of why they’re not doing that now. Then we get back to the question about what’s the point at which they would start those kinds of activities.

They would need to have enough information that they believe some company or set of companies is getting close to AGI, but not enough that they would have done something previously — assuming they’ve got good capabilities on the shelf that they would pull off if they have to.

The non-falsifiable part is about to what extent they could ramp up attacks, to what extent there would be defenses against those attacks, and to what extent those non-kinetic strikes would actually meaningfully delay the development of a technology. Another way of saying this — my prior is that there’s lots of espionage happening all the time. I want to see more specificity in this argument about what exactly folks mean when they talk about escalation.

Jordan Schneider: One of the things that has been remarkable about China, at least how it deals with foreigners, is that you haven’t seen what Russia did with all these targeted assassinations. The sharpest we’ve gotten, at least with dealing with white people, has been the handful of Canadians who were grabbed and ultimately let go after a few years in captivity following the Meng Wanzhou arrest.

People are very focused on China starting World War III out of the blue. But there are also world states in which China becomes much more unpleasant while not necessarily kicking off World War III.

Michael Horowitz: I’m a definitive skeptic on the “China starts World War III over AGI” point. I buy the notion that China could become more unpleasant as we approach some sort of AGI scenario — including non-kinetic activities, espionage, etc. I tend to maybe not view those as decisive as some others do potentially.

You’re right that they certainly could become a lot more unpleasant. If the question is why they haven’t already, the answer is probably twofold. One, there’s an attribution question — suppose Chinese espionage involved doing physical harm to AI researchers or something similar. If they were caught doing that, they’ve now potentially started a war with the United States, and they’re back to the reason why they wouldn’t launch a military strike in the first place.

If it’s non-attributable, then the question is exactly how much are they going to be able to do? I wonder whether there is something about their broader economic ties with the United States that maybe makes some of the worst kinds of these activities less likely in a way that is less troubling to Russia.

Jordan Schneider: This is a decent transition to precise mass in the China-Taiwan context. What can and can’t we infer about military technological innovation in Ukraine to what a war would look like over the next few years?

Michael Horowitz: It’s not necessarily the specific technologies, but it is the vibes. By that, what I mean is the advances in AI and autonomy, advances in manufacturing, the push for mass on the battlefield that we see already in publicly available documents and reporting on how the PRC is thinking about Taiwan. We see that already in the US in the context of Admiral Paparo and Indo-Pacom and the Hellscape concept or something like the Replicator initiative in the Biden DoD — and full disclosure, I helped drive that, so I certainly have my biases.

We see that if you look at some of the systems that Taiwan has been acquiring over the last couple of years. You essentially have a growing recognition that more autonomous mass, or what I’d call precise mass, will be helpful in the Indo-Pacific. It’s unlikely to be the exact same systems that are on the battlefield in Ukraine, but variants of those scoped to the vastness of the Indo-Pacific.

Shashank Joshi: I have a few thoughts on this. One way to think about what Mike is saying is for any given capability, you can have more intelligence that is defined however you like, whether that’s in terms of autonomy or capacity to do the task on the edge at a lower price point.

That capability could be a short 15-kilometer range small warhead strike system in the anti-infantry role. It could be a 100-kilometer system to take out armor with bigger warheads, or it could be significantly longer range systems that have to be able to defeat complex defensive threats. Obviously, the third of those things is always going to be more expensive than the first.

What that revolution in precise mass, if it is a revolution (we can debate if that’s what it is), does is push you down. The capability per dollar is going up and up. That is the essential point.

Michael Horowitz: Just to be clear, that’s the reverse of what we saw for 40 years, where in the context of the precision revolution, you were paying more and more for each capability, whereas now we are seeing the inverse of that in the era of precision mass.

Shashank Joshi: Is the transformative effect comparable across each level of sophistication or capability or range? Are there specific things to FPV-type systems because they, for instance, rely on consumer electronics, consumer airframes, and quadcopters — they can draw upon a defense industrial base or an industrial base that has existed for commercial drones? Is it easier to have that capability revolution for intelligent precision mass at one end of the spectrum relative to building a jet-powered system that has to travel significantly further, has to defeat defense mechanisms, may have to have IR thermal imaging, etc? Is a revolution comparable at each end?

Michael Horowitz: I was with you up until the end about defeating all of those systems, because the thing that’s so challenging about this for a military like the United States is it’s a different way of thinking about fighting. You’re talking about firing salvos and firing at mass as opposed to “we’re going to fire one thing and it’s going to evade all the adversary air defenses and hit the target."

Look at Iran’s Shahed 136. That’s a system that can go, depending on the variant, a thousand-plus kilometers. It can carry a reasonably-sized warhead that, in theory, could have greater or lesser levels of autonomy depending on the brain essentially that you plug into it. That’s not going to be as sophisticated a system as an advanced American cruise missile that costs $3 million or something. But it doesn’t have to be because the idea is that these are complements where you’re firing en masse to attrit enemy air defenses. Your more sophisticated weapon then has an easier path to get through. It’s just a different way of thinking about operating, and that creates all sorts of challenges beyond just developing the system or buying the system.

Shashank Joshi: That’s a really interesting point. This gets us to a phrase, Mike, that is very popular in our world and you and I have talked about this, which is the mix of forces that you have and specifically the concept of a high-low mix. It’s not just that small drones will replace everything. You have a high-low mix where you will have some, albeit fewer, very expensive, high-end capabilities that can perform extremely exquisite, difficult tasks or operate at exceptionally high ranges. Then you will have a lot more in quantity terms of lower-end systems that are cheaper, more numerous, and that will not be as capable — they can’t do things that a Storm Shadow cruise missile or an ATACMS missile can do, but they can do it at a scale the Storm Shadow can’t do or the ATACMS can’t do.

The most difficult concept when I’m writing about this for ordinary people who are maybe not into defense is explaining the mix, the interaction of those two ends of the spectrum. Here’s the difficult bit — pinning down what is the right mix. Do we know it yet? Will we know it? How will we know? Does it differ for countries? That’s where I’m struggling to understand all of this.

Michael Horowitz: It probably differs for countries and even within countries differs by the contingency. For example, if you are fighting, if you’re back in a forever war kind of situation and you’re the United States, then you might want a different mix of forces than if you’re very focused on the Indo-Pacific and on China in particular. What comprises your high-low mix probably changes.

The way that I’ve tended to think about this is you have essentially trucks, which are the things that get you there. Then you have brains, which is the software that we’re plugging in. And then you have either the sensor or the weapon or the payload piece. In some ways, what I think we’re learning from Ukraine that is applicable in the Taiwan context is that sometimes it matters a lot less what the truck is than what the brain is.

Shashank Joshi: The other thing that I struggle to get across is the relationship between the “precise” bit and the “mass” bit, particularly the role of legacy capabilities in this. The great conflict I see today is in the artillery domain. Do strike drones replace or supplant artillery? Strike drones now inflict the majority of casualties in Ukraine, not artillery, as was the case at the early phase of the conflict.

There is this really interesting line in the British Army Review published in 2023 — “There is a danger that the enemy will be able to generate more combat echelons than we have sensors or high-end long-range weaponry to service.”

You can have these remarkable AI kill chains that can spot soldiers moving and feed that data back to your weapons — but if you don’t have the firepower to prosecute those targets and then keep prosecuting them week after week in a protracted conflict, you don’t have deterrence.

That is what we are waking up to now.

Michael Horowitz: The important thing here is the notion that you didn’t need a high-low mix, that you could just go high, presumes short wars where you can just use your high-end assets and sort of shock and awe the adversary into submission. Whether we’re talking about a forever war situation or in the Indo-Pacific, if you’re fighting in a world of protraction, then you need much deeper magazines in all ways, including in your platforms frankly.

Maybe in that world AI is actually helping you with what you’re manufacturing and how you’re manufacturing and can deliver a bunch of other benefits on the battlefield. A big challenge here is that I don’t think these capabilities necessarily mean there’s no role for traditional artillery. Although if they can do the same job better at lower cost, then they will eventually displace those capabilities, or militaries that don’t adopt them will fall behind. We’re more at a complement than a substitute stage right now for those capabilities. But things could change.

The challenge right now for a military like the US is you have all these legacy capabilities, and maybe you wish to invest less in them to be able to invest more in precise mass capabilities, which is something I advocate. But then the question becomes what are you doing with those legacy capabilities and when across what timeframes?

This does tell you some things that are really important from a force planning perspective. For example, the “one ring to rule them all” approach to air combat that led to the F-35, where you’re just going to have one fighter that will operate forever and presumably be useful for every single military contingency. Turns out that means it’s optimized for none of them, and that generates a bunch of risk. One of the things that the new administration is probably going to be doing is figuring out how to address those risks. I don’t want to hate on the F-35 and its stubby little wings. Sorry, I’ll stop now.

Shashank Joshi: I’m really glad you raised F-35 because this gets to another one of my points of thinking. You said, “If it can do the same job,” right? What are the jobs it can and can’t do?

When you look at a simple mission, let’s take an anti-tank guided missile — it’s looking pretty clear to me that a kind of mid-range, low-cost strike system, one-way attack strike system is looking like it can do the job of an ATGM very effectively at significantly lower cost. Therefore, unless the ATGMs also get transformed, you’re going to see a supplanting of roles.

However, I’m also acutely conscious that we are thinking about some battlefields that are in some ways uncluttered. We’re looking at a stretch of the Donbas in which anything that moves is going to be the target that you want to hit — it’s going to be a Russian military vehicle. You’re not going to accidentally hit a school bus full of children. In the Taiwan Strait, you are using your object recognition algorithms to target shipping. You’re not accidentally going to hit something in the context of an amphibious invasion of Taiwan.

Michael Horowitz: Yeah, if the balloon goes up, it’s not like there’s lots of commercial fishing just chilling in the Strait.

Shashank Joshi: Well, and if it is, it’s probably MSS operatives. You’re fine. But in the air power situation too, right? You may be doing stuff like this. However, urban warfare is not going away. I’m imagining fights over Tallinn, over Taipei, over thin, cluttered, complex, multi-layered subterranean environments like Gaza, like Beirut, like places like that. I worry a lot more about the timeline over which autonomy will suffice.

To end on one last point before I spin off the RAF, the Royal Air Force believes that an autonomous fighter aircraft will not be viable prior to 2040. Now maybe that’s ultra conservative, but that’s based on some of the assumptions about the tasks they think it will need to do. I know you have your debate over NGAD and long-run capabilities. So are there limits to this process?

Michael Horowitz: The limits are how good the technology is. Frankly, I’ve argued this for years in the context of autonomous weapon systems. The last place that you would ever see autonomous weapon systems is in urban warfare. Not only whatever ethical moral issues that might surround that, but that’s just way harder to do than figuring out whether something is an adversary ship or an adversary plane or an adversary tanker in a relatively uncluttered battlefield.

Shashank Joshi: The second part was to do with the air picture, right? Countries are now having to decide what our air power is going to look like in 2040, and how much can we rely on the technology being good enough by then? You’re right, that is the question — will it be good enough? But you have to make the bet now. You have to make the judgment now because of the timelines of building these things.

Michael Horowitz: There are two questions there. One is, how good is the AI technology? Frankly, even if Dario is overestimating how quickly we get to something universally recognized as artificial general intelligence — and just to be clear, dude’s way smarter than me and I’m not saying he’s necessarily wrong, just that there’s uncertainty here. He’s super smart and thoughtful.

Side note, the fact that CEOs of today’s leading companies post their thoughts on the internet and come on shows like this is super useful. Now that I’ve taken off my Defense Department hat and I’m back in academia, it’s excellent for understanding how a lot of the people designing cutting-edge technology are thinking about it and interacting with government policy and militaries.

If you’re talking about what the future of a next-gen aircraft airframe would look like or what a collaborative combat aircraft could do, my bet is that the militaries are underestimating how quickly AI will advance and the ability to do that. What they might be accurately assessing is, given the way the process of designing new capabilities works today and given today’s manufacturing capabilities, how long it would take to actually design and roll out a new system.

From a parochial American perspective, shortening that timeline is way more ambitious than what we were attempting to do with Replicator. I expect the new team actually will continue to push forward a lot of those things, whether they call it Replicator or not. You could have a scenario in which your AI is at a level that you believe you could have more autonomous operations of a collaborative combat aircraft, and you have some advances in manufacturing that mean you could produce maybe more of them at a slightly lower price point, but still be unable to do so before 2040 for various bureaucratic and budgetary reasons.

Jordan Schneider: Looking back over the past four years, where are the places that the Biden administration made progress in defense innovation?

Michael Horowitz: We accomplished a lot in getting the ball moving, specifically toward greater investments in lots of important next-generation capabilities like collaborative combat aircraft and varieties of precise mass through the Replicator initiative. But there is a lot more to be done.

It was a journey in two ways. One was bringing everybody along on that defense innovation journey to get to the point where folks bought into the importance of some of these emerging capabilities for the future of warfare, specifically in the Indo-Pacific. Then there’s what we could accomplish before the clock ran out.

To start with the first piece, taking everybody on the journey — nothing happens in the Pentagon without getting lots of people on board. It took a little while for there to be consensus on the state of the defense industrial base. If you look at varieties of think tank reports about what DoD should do differently, there are often all these suggestions like, “DoD needs to scale more of this system, scale more of that system, build more of that.” All of that is great, except that even if you fully fund really exquisite munitions like the Long Range Anti-Ship Missile or JASSM or something like that, you’re going to be really capacity constrained because those facilities just have limits. You can change those limits, but you can’t change those limits quickly.

What can you then do to scale capability relevant for the Indo-Pacific in the short term? The answer is precise mass capabilities — more attritable systems, more AI-enabled and autonomous systems, systems that maybe sometimes, but not always, are built by non-traditional companies. There’s a coalition of the willing that pushed through all of those in a bunch of different contexts, including Replicator, which DIU I think did a brilliant job of leading implementation of, and that generated some growing momentum.

But it also highlighted some real limits. Reprogramming 0.05% of the defense budget to fund multiple thousands of attritable autonomous systems for the Indo-Pacific under the first bet of the Replicator initiative required over 40 briefings to Congress, including a ton by the Deputy Secretary of Defense who’s really busy. Congressional oversight is really important, but the degree of effort required to reprogram essentially less than a billion dollars demonstrates a budget system unable to operate at the speed and scale necessary given the rate of technological change and given the threat that the US is under from Chinese military advances in the Indo-Pacific.

We did some good things. We moved the ball forward. I’m proud of some of the things that we did, but there’s a lot more work to be done.

Jordan Schneider: Ray Dalio recently said that “America can’t produce things, any manufactured goods, cost-effectively.” Shashank, if Ukraine can figure out how to build a million and a half drones a year, couldn’t the US do this too if we were more determined?

Shashank Joshi: Let’s unpack this a little bit. First of all, Ukraine is doing this under wartime conditions. It’s ripping up rules — not all of them, but a lot of them. It doesn’t have to worry about pesky things like health and safety standards, like “Can I build this explosive warhead facility next to this town?” You try doing that in the US or the UK or in Europe. The Ukrainians can get around that because they’re at war, it’s fine.

Secondly, what are they building these things with? The Ukrainian supply chain for UAVs and for lots of other things, including electronic warfare systems, is still full of Chinese stuff. They haven’t got it all out. Yes, there are companies beginning to find alternative supply chains and finding stuff from Taiwan and other countries, but the US government can’t do that. There’s the little pesky matter of the NDAA which prohibits you from just sticking Chinese components into all of your drones and building them out. So you have to get supply chains right. That’s problem number two.

Problem number three is to what standard are you building these things? Are you demanding that they can cope with this level of electronic warfare, radiation hardening, cybersecurity standards? Those are all the requirements being placed on UAS in many Western governments. I won’t speak directly to the US exactly, because Mike knows it better than I do, but certainly in my country I’m aware of this. The Ukrainians can build things to a satisfactory standard that would never stand up to the scrutiny of an accounting or auditing official in a Western country to the same way.

The final point on all of this is the level of mobilization for the defense industrial base is quite different. Ukraine has nationalized its IT sector. It had a brilliant IT sector before the war. It had some great technicians, tech-minded people, software-minded people. That is just not the case in most of our countries. We can’t nationalize our tech industry to go work to build software-defined weapons at a low cost very effectively and quickly. Those are some of the reasons I can see for the discrepancy. Mike, do you disagree?

Michael Horowitz: I think that is all correct. I’m probably more optimistic than you about the ability to build at low cost. Although what exactly you define as low cost? If we think about attritable as replaceable kinds of systems, and what’s attritable probably varies depending on how wealthy you are and how big your defense budget is. What’s attritable for the United States might be different than what is attritable for Ukraine.

There’s a lot of possibility to get lower cost systems. One example is last year there was an Air Force DIU solicitation for a low-cost long-range cruise missile that would be about $150,000 to $300,000 a pop. That’s way more expensive than an FPV drone in Ukraine, but that’s a lot more capability than an FPV drone in Ukraine. Given that existing systems, some of those existing systems cost a million dollars or more — if you can deliver that at what is a fraction of the cost, you’re buying back a lot of mass in a useful way.

The thing that I will be really interested to see, shifting actually to Europe for a second — if you look at what Task Force KINDRED has done for Ukraine, my question is, when is the UK going to buy those capabilities for the UK? If these are militarily useful capabilities for fighting Russia, that seems like something that might be useful for the United Kingdom’s military.

Or say the German company Helsing — they produced 6,000 drones that were going to get sold to Ukraine. Obviously that’s in the context of millions being produced, but presumably those 6,000 are pretty good from a quality perspective. Why isn’t Germany buying those drones?

Shashank Joshi: Helsing is a really good example. In Ukraine, distributed in Ukrainian manufacturing facilities, they are building these UAVs where the hardware is not completely standardized. The software has to adapt to the different airframes being built by different producers. But they’re good enough, they’re doing a good job, they’re making a difference, they’re producing Lancet-like capability at considerably lower cost as far as I understand it.

But Helsing is also building drones for NATO countries in southern Germany in its own factories. The advantage is that it controls the supply chains. It can standardize the process, it can build software-defined weapons in ways where the hardware is optimized to take that on — in the way that Tesla once built other people’s cars and put its code in them. But since that initial phase, it’s built its own hardware because it’s easier to build a software-defined car if you do it yourself. But it’s going to be more expensive and it’s not going to be as cheap and quick and easy as the Ukrainian manufacturing. I believe there are some trade-offs that you see even within the same company.

Michael Horowitz: Absolutely. That also is why you need more open architecture all around and why in some ways the government needs to own more of the IP. If you think about what I said before about trucks and brains — if you’re buying a truck that only can have a brain from the same company, then you’re locked into a manufacturing relationship that’s almost necessarily going to generate higher costs over time than if you can swap out the brain over time with something that might be more advanced. Frankly, whether it’s lower cost or not, it’s probably a better idea to be able to swap it out.

This reflects differences in not just the defense industrial base, but in how the US and western militaries have thought about requirements for capabilities over time in ways that now require challenge.

Shashank Joshi: There was a really interesting speech. The Chief of Defence Staff in the UK, the head of our armed forces, in this case Admiral Sir Tony Radakin, gives a speech every year at Christmas at the Royal United Services Institute, the think tank in London.

Michael Horowitz: Is there port?

Shashank Joshi: There is wine after the lecture, but we’re not drinking port during the lecture.

Michael Horowitz: But I’m imagining like brandy and cigars and port and like a wood-paneled room.

Shashank Joshi: You’re not far off on the room itself.

Jordan Schneider: Are there oil portraits? Do we have like Hague and stuff?

Shashank Joshi: There are oil portraits. I must declare I’m on the advisory board at RUSI, so I’m very fond of it. But anyway, I’ve gone down this rabbit hole. The reason I brought it up is because I wanted to quote a line from that speech he gave at Christmas where he said:

[W]e have only been able to demonstrate pockets of innovation rather than the wholesale transformation we need.

Where we have got it right is because we used an entirely different set of permissions which elevated speed and embraced risk so that we could help Ukraine.

But when we try to bring this into the mainstream our system tends to suffocate the opportunities.

He then proposed a “duality of systems… Whereby major projects and core capabilities are still delivered in a way that is ‘fail-safe’ – clearly the case for nuclear; but an increasing proportion of projects are delivered under a different system which is ‘safe to fail’”.

Mike, this is pretty much what you told me when we spoke a few weeks ago, right? A willingness to embrace failure — not for your Ohio-class SSBNs or your bombers, but for your smaller systems where the cost of failure is not terrible and you need to fail to innovate. That is so much of what it’s going to take for us to be able to be more Ukrainian in our own systems.

Jordan Schneider: Let’s contrast that with a quote on the US side of the pond from Bill LaPlante, who is the DoD’s top acquisition executive during the Biden administration:

“The Tech Bros are not helping us much… If somebody gives you a really cool liquored-up story about a DIU or OTA, ask them when it’s going into production, ask them how many numbers, ask them unit costs, all those questions, because that’s what matters… Don’t tell me it’s got AI and Quantum in it. I don’t care.”

Michael Horowitz: Bill LaPlante thinks that it’s still 1995. The thing that was remarkable about that quote is even at the time it was so profoundly incorrect in the way that it described the ability to scale emerging capabilities. Shortly before we left office there was a speech or maybe it was from under a question or something where he said that he had learned from the Houthis and what they’d done in the Red Sea that it was possible to produce low-cost munitions. Unclear why what was happening in the US had not triggered that revelation. But he got there.

That is the mindset that requires challenge. If you view the only things worth using as an extremely small number of exquisite platforms, then that takes you down a road where emerging capabilities, even those you can scale, might seem less useful because they might require using force differently — if you’re operating at mass rather than just operating those exquisite capabilities.

The bigger challenge though, is every single major defense acquisition program in the US military is either behind timeline, over budget, or both. Most are both. What that suggests is that the current system, which is designed to buy down risk and produce these great capabilities — and it does, but just slower than it’s supposed to, with higher prices than it’s supposed to — in a way that suggests that the current system is not succeeding.

Risk is the right way to think about this. The scale and scope of the challenge posed by the Chinese military is unlike anything I have seen in my lifetime from an American perspective. That means that the assumption that undergirded the 90s, frankly, about the inevitability of American conventional military superiority, is just no longer the case. It’s not just that we can’t sit on our laurels, which is something that I think I wrote maybe a decade and a half ago. It’s that we are being actively pushed and challenged across almost all domains.

What that requires is accepting more risk in the capability development process, which I feel comfortable doing not only because I’m generally bullish on the ability of emerging technologies to deliver, but also because the status quo system just isn’t working.

Shashank Joshi: What we can’t ignore then is what is stopping us — or you in the US case — from taking that risk. Often it’s the politics. You talked about how shifting 0.05% of the budget requires this Herculean bureaucratic political effort on the Hill to plead with Senators and Congresspeople, “Please let me move this $15 million here and there.” That’s not sustainable if you’re trying to make a systemic effect.

You have to have appropriators who are willing to say, “Actually I trust you with this money, and I trust you to be able to spend it in a way that’s flexible and won’t lock you into a spending path for the next six months without wasting it. And let’s test you on that in six months,” but not micromanaging everything.

I don’t know how you’re going to be allowed to have the failure that you need to have the innovation you need if Congress doesn’t trust people to innovate at scale, not just in these little pockets of innovation as Radakin called it.

Michael Horowitz: Every single rule that the appropriations committees have exists because of something that happened in the Department of Defense in the past. To be clear, we are in a different era now with a different set of risks and a China challenge that’s unlike anything we have faced before. We need a new bargain in some ways with the appropriations committee to be able to innovate at the speed and scale we need.

Keep in mind the Pentagon’s budgeting process was invented by Robert McNamara during the Vietnam era and has not changed since then. In the best of times, it is a two-year cycle between when one of the military services decides it wants to invest in a technology and when it gets the money to invest in that technology.

We’ve spent several years of the last decade and a half in continuing resolutions, which means Congress can’t pass and appropriate a budget. This means you can’t start new programs, which then delays adopting new capabilities even further. Something has to give there.

Jordan Schneider: I want to recommend the podcast series Programmed to Fail: The Rise of Central Planning in Defense Acquisition by Eric Lofgren, who’s now working in the Senate. He used to run Acquisition Talk and do shows with me about this stuff. You know, it wasn’t just McNamara — it was McNamara trying to learn from the Soviets.

Subscribe now

Michael Horowitz: The assumption was we’ve got the technology we need. We think that our basic tech development system works. What we need to be able to do is produce this stuff and produce good stuff, and then we’ll beat the Soviets. It worked, but we’re in a different period now.

Shashank Joshi: There’s an interesting book I wrote a review essay on for Foreign Affairs about a year ago by Edward Luttwak and Eitan Shamir, head of an Israeli think tank, called “How Israel Fights.” I don’t find all of it persuasive, but it raises the question of how this country in the 1960s — this agrarian society that is poorer than many parts of Southern Europe in GDP per capita — produced these anti-ship missiles that are able to defeat the Soviet weapons being carried by the Egyptians of that era and the Syrians.

What did they do right? What have they done right? There are many things that they’ve done wrong, and there are many cases in which tech innovation did not help them strategically or even contributed to complacency. But there’s something about that innovation, including innovation under conditions of peacetime or semi-peacetime, that I think we should be thinking about.

Jordan Schneider: I have a five-hour Ed Luttwak episode in the tank that I’ve been dreading editing. But we did get into it, and there is something about this topic. It comes back to some of the Ukraine stuff — Israel is a semi-mobilized society. It’s playing at a smaller scale.

There’s this great anecdote where someone walks into an office and says, “You should arrange the tank this way instead of that way.” Then they do it because somebody thought it was a good idea. You take all his stories with a grain of salt, but still conceptually, the fact that this is all among friends in this small network of, by the way, the best minds in the country.

Michael Horowitz: Whereas in our system, fourteen different people can say no and stop a capability, but no one person can say yes and move it forward.

Jordan Schneider: One of the many shames of the Trump imperial presidency is that despite having enough control of Congress to do this well, getting Pete Hegseth to be the one to lead it is just one of these unfortunate timelines we’re in because the President couldn’t give two shits about this stuff. Maybe there are enough tech people around him though.

Michael Horowitz: Let me muster a point of optimism here, frankly, on this. In the brilliant article that Shashank wrote in The Economist on some of these questions, I sounded a similar note. If you look at Hegseth’s testimony, his discussion of defense innovation is very coherent. He makes points that are not structurally dissimilar to the ones that we have been making for whatever the last period of time has been.

If you look at Stephen Feinberg’s testimony yesterday to be Deputy Secretary of Defense, he actually makes some very similar points, and you hear some of those echoed by various tech sector folks that look to be entering either the White House or the Defense Department.

There is a potential opportunity here for the Trump administration to push harder and faster on precise mass capabilities, on AI integration, and frankly, on acquisition reform in the defense sector, because the president right now seems to have a strong hand with regard to Congress. Whether the president’s willing to use political capital for those purposes is not clear. How the politics of that will play out is unclear. But if the Trump administration does all the things that it says it wants to do from a defense innovation perspective, that may not be a bad thing. There are a lot of things they want to do that are very consistent with things that many of us have advocated for over the years.

Shashank Joshi: I really admire your dispassionate assessment of that and the willingness to think about it apart from the politics. My concern is that you have people who are good at radicalizing and disrupting many businesses and sectors and fields of life. But the skills required to do that are different from the skills needed in a bureaucracy like this.

Just because you were able to navigate the car sector and the rocket sector doesn’t mean you know how to cajole, persuade, and massage the ego of a know-nothing congressman from — I’m not going to name a state because that’ll end up being rude — who knows nothing about this and who simply cares that you build the attritable mass in his state, however stupid an idea that is, and who wants you to sign off on the $20 million.

I worry that they will either break everything — and what I’m seeing Doge do right now with a level of recklessness and abandon is worrying to me as an ally of the United States from a country that is an ally — but also that they will just not have the political mouse to navigate these things to make it happen. Just because Trump controls Congress and has sway over Congress doesn’t mean that the pork barrel politics of this at the granular level fundamentally change. You need operatives, Congressional political operatives, and a tech pro may have many virtues and skills, but that isn’t necessarily one of them.

Michael Horowitz: No argument. There’s a huge gap between being willing to lean further forward on defense innovation and transformation and the ability to bureaucratize, essentially, as a friend of mine is fond of saying, and be able to get the job done delivering. The Pentagon is the world’s largest bureaucracy and it will continue to be the world’s largest bureaucracy even with whatever is happening. That requires a lot of bureaucratic political acumen to be able to deliver results. It is a very open question whether this administration will be able to deliver on that. Frankly, there are early signs that are concerning. But again, it’s still early days.

Jordan Schneider: I want to reflect a little bit about the role of inside knowledge and outside knowledge when it comes to understanding what’s happening in Ukraine as well as the future of war. What does stuff like Shashank’s reporting get? What can it not get? How does all of the open source analysis that’s happening today about the war in Ukraine filter into discussions about budgets in Congress and R&D?

Michael Horowitz: Systematically drawing insights from open source material, both Shashank’s and others, in ways that inform what we do in the Defense Department is important. In the context of Ukraine lessons learned, there are actually a number of different efforts, both classified and unclassified, that try to dive into those things. In the case of Ukraine specifically, there’s heroic effort both inside and outside the context of just the Ukraine desk at the Pentagon or on the Joint Staff to do that.

The challenge sometimes is making some of those insights more visible and then connecting them to the change that you wish to see. Part of the issue is that folks are really busy in ways that are sometimes even difficult to comprehend on the outside. Your read time, even to read things that you really want to read, is just extremely limited.

The role of an influential columnist like Shashank that lots of people trust is invaluable because for a lot of senior folks, that might be the only outside thing they read about defense in a given week. This points to the importance of networks. I think about this a lot from the perspective of how academics should bridge the gap between academic research and policy — networks play a huge role there.

In the case of Ukraine, I think there actually has been really good pickup inside, at least in the US, on what lessons learned look like because of a lot of the great reporting out there. Sometimes people would say, “Why haven’t you purchased this drone that Ukraine uses?” It’s a great question, but it’s really challenging to consume all of the information out there that you should consume. A lot of it ends up getting mediated through staffs.

Jordan Schneider: Here’s a deep cut for you guys. Ian Hamilton, who ran the Gallipoli campaign, was a journalist who covered the war in Manchuria between Imperial Russia and the Japanese and saw the future of war and wrote about it really clearly. But even though this guy was there and then was on the battlefield running it in Turkey, he was not able to instantiate the lessons that he saw firsthand into the way he ended up killing a few hundred thousand people who probably didn’t need to die if he had made smarter decisions. None of this shit is easy.

Michael Horowitz: As part of research for a future book, I was at the National Archives in College Park last week looking for information on US military procurement decisions in the early 20th century surrounding General Purpose Technologies. I found some really interesting back and forth between the War Department and the Wright brothers about the airplane that sounded a lot like modern debates. The Wright brothers are saying, “Well, send us the cash and we’ll send you the airplane.” And the War Department’s responding, “Prove it works and meets these metrics, then we’ll pay you and then you deliver the airplane.” It was like, “Oh dear God, maybe nothing has changed” in some ways in some of these debates.

Jordan Schneider: Shashank, any reflections on the role of popular writing in all this?

Shashank Joshi: I’m amazed by the cut-through we can sometimes get. People will say, “I can’t take my classified system on a plane to read, but I can take a copy of the Economist.” So you suddenly have this responsibility. I’ve had deep experts on something like armored warfare and tanks say, “I’ve been screaming this message into the ether for years, but it was only when you quoted me that this general read me."

We’re sometimes in the strange position of being — I don’t want to say conduits because we would never wish to be uncritical conduits for anything — ways that can short-circuit these networks and cut across them in strange and amusing ways. I have the grave responsibility of not only telling my readers about the big stuff, like what’s Trump going to do next or where’s Ukraine headed, but — if this is not too condescending — feed them their vegetables, make them think about Mike’s essay on precise mass.

Maybe I have to bury it in a piece that’s about the future of drones, but I can make them think about budgeting. Mike tells me you have to understand budgets to understand innovation. Then I think, “Okay, now there’s a challenge. My editors may not like me talking about budgets for a page, but this is my job to get it across and to make people read it and listen to it."

I’m fortunate that I have access to expertise like that of Mike and others to be able to translate that. Fundamentally, Jordan, I see my job as not giving people the answers. It’s just giving them a sense of the debates that the knowledgeable people are having. That’s not to say one person’s right, one person’s wrong, but to say, “Here’s the lay of the land. Here are the arguments on each side. Here are the debates,” and give them a flavor. Let them peer through that window into the world of the conversation that Mike may be having with his colleague on what they disagree on.

Jordan Schneider: What you do, what Mike Kaufman does, what Mick Ryan does, what Mike Lee can do — which I imagine would be harder if you are sitting on the Ukraine desk and your job is to cover what’s happening in electronic warfare — is to think about this all synthetically and across the different domains. In your case, Shashank, even across regions.

This is what I feel like I do in some sense with ChinaTalk as well. Shashank, you have an editor who it seems like you can just bowl through at this point, and I’m very glad you can basically write budget articles. Picking what you want your readers to read and think about is the game. Doing that in a really thoughtful way when there is so much new happening, so many battles occurring in every moment, and so many tactical innovations and counter-responses is essential. This is particularly important for the public at large and also for senior leaders who only have about 3% of their time to really sit down and absorb this stuff — or God forbid, the president.

Shashank Joshi: Or the vice president who has lots of it.

Jordan Schneider: Mike, continuing on, why outside writing matters.

Michael Horowitz: Now that I’ve left the government again, when I think about how, as an outsider, as an academic, to try to influence policy, one way to think about this is: if you want the US Government and the national security arena to be doing something and they’re not doing it, there’s usually one of two reasons.

First, you might be wrong. There could be classified information or some other information you don’t have access to that shows you’re wrong.

Second, you’re right, but your bureaucratic allies are losing. It is hubris, given the size of the national security agencies, including the Pentagon, to think that you have some idea that literally nobody in the entirety of the Pentagon, the intelligence community, and the State Department has thought of. Generally, somebody wants the same thing that you want, but they’re losing. The question is how you give them ammunition to help make the case to move that policy forward.

In the kind of writing that involves advocating for policies or making arguments for policies, I think it makes sense to think about this in terms of how you’re providing support to your bureaucratic allies, even if you don’t know who they are and even if they don’t know you until they see something you write show up in the Early Bird in the morning at the Defense Department or through some sort of press clippings. That’s how I think about the role of outside writing and how you can try to influence policy.

Jordan Schneider: It’s always weird for me when I write something in ChinaTalk and get an email from someone I’ve never met saying, “Thanks for this. This was helpful.” Just putting your arguments with some good, thoughtful analysis out into the ether sometimes works in mysterious ways.

Michael Horowitz: There’s this fiction that you write an op-ed and it somehow ends up on the desk of the president, and then all of a sudden US foreign policy changes. That’s just not how the real world generally works, especially if what you’re trying to influence is policy within a bureaucracy.

I’m encouraged to hear people within the government should be listening to ChinaTalk and getting insights from it. That’s terrific. That’s evidence for this idea that you have allies and you’re trying to help give them ammunition to make the case in whatever fora they’re engaged in.

Jordan Schneider: I got one last question for you guys, if you don’t mind. All of this, compared to reading and writing and doing podcasts about the PLA itself, is just so much more interesting because I’ve read a lot of PLA books and I’ve really thought about doing more shows about it. It’s so difficult to talk in hypotheticals when you’re reading doctrine stuff and doing the OSINT and whatever. It’s just so hard to actually learn things and talk about them in an interesting way. I’m curious if you guys have any advice for me about what better PLA coverage in outlets like ChinaTalk or just more broadly could look like to get people thinking more seriously about all this stuff.

Michael Horowitz: This is super hard. We think about this a lot actually in the context of PhD students, junior faculty, and what the academic China-watching community is doing in this space. Especially given the way that Xi has consolidated control in China, there’s still a lot available, but there’s so much less available frankly than there was 20 years ago or even 10 years ago. That creates analytical challenges because what you can get raises questions like, “Well, why could I get this?” There are essentially, to be really nerdy, selection effects that govern what you’re able to access.

The truth is there’s no military in the world where we probably have a greater uncertainty parameter about its potential performance in a conflict than China’s military — the PLA — because it’s just been decades since it fought. We know what weapons they have. We know what their doctrine says. The ability to put all that together, as we know from the Russia-Ukraine context or any war in history, is very different than what it looks like on paper.

I do not envy the task. All you can do in some ways is acknowledge that irreducible uncertainty and do your best to give folks the information that is available.

Shashank Joshi: I would just add to that: let’s learn from our analytical errors in the past and think about how they might apply. I’ve really enjoyed some of the writing done by people like Sam Bresnick at the Center for Security and Emerging Technology at Georgetown on the way the PLA thinks about AI. He doesn’t say, “Oh, they’re miles behind, that’s useless.” But he does give a flavor of Chinese debate, saying they worry about many of the same things that we do. They worry about explainability, about control, command, oversight. They even worry about ethics — that’s not completely absent. They’ve got issues around compute capacity and all these other things.

It’s just helpful to be reminded that what they say on the page about intelligentized warfare and this and that — they’re grappling with some of the same challenges that we all are. I admire the work of people like Sam and others and his colleagues and many others who think about these from a very fieldwork-based or empirical perspective, getting their hands dirty, reading the stuff, talking to people, looking through journals. That’s great work.

Jordan Schneider: Well, as Sam’s peer advisor in high school when I was a senior and he was a sophomore, I am going to take complete credit for all the brilliant work that he’s done both in Beijing and now in Washington. All right, last thing. Each of you give one book for everyone to read.

Michael Horowitz: My junior colleague at the University of Pennsylvania, Fiona Cunningham, is the most talented academic scholar of the Chinese military of her generation. She has a book that just came out with Princeton University Press titled Under the Nuclear Shadow: China’s Information-Age Weapons in International Security. I would highly recommend checking out Fiona’s book. [We recorded a pod already!]

Shashank Joshi: The orthodox choice is someone who came up earlier — Paul Scharre’s book, Army of None, which I still think is fantastic on the issue of autonomous weapons. It’s brilliant on the history and thinks about it historically — fantastic book for anyone still thinking about autonomous weapons.

But the slightly left-field choice I want to put out there is The Billion Dollar Spy by David Hoffman, which is the story of one of the CIA’s most difficult operations in Moscow running Adolf Tolkachev, a Soviet engineer. The reason it’s relevant to this conversation is it’s about the application of technology to operations — in this case, intelligence operations, running an agent in Moscow, communications technology, miniaturization, the way that the emerging plastics industry and transistor industry affects the CIA’s choices in the ’50s, the way that changes with satellites. I love the idea of thinking about this in a completely different field, intelligence, espionage, and the parallels and ideas that may spark for us thinking about the defense world.

This continues a series examining China’s role in the global biotech and pharmaceutical industry, and focuses on China’s drug markets. Our first piece explored the role of AI in China’s biotech ecosystem. Future articles will cover the Chinese biopharmaceutical research, innovation, and supply chains. If you have tips on any of these topics, please reach out.

As China prospers, so do its citizens’ expectations for high-quality healthcare. That means expensive drugs – but how can it make them affordable?

The government is transforming the landscape of China’s pharmaceutical markets, cutting drug costs, and pushing domestic firms to become more competitive. But even as policy advances, demand is evolving just as quickly.

As incomes rise, so does China’s disease burden. The country is seeing more of the chronic and complex illnesses common in wealthier nations — cancer, cardiovascular disease, and aging-related ailments — that often require expensive, innovative treatments. Forecast to grow to USD $264.5 billion in 2026, China’s pharmaceutical market is the second largest in the world behind the United States. The US is still by far the world’s largest pharma market, at over US$600 billion today.

China’s healthcare ecosystem remains challenged by corruption and other inefficiencies. Though domestic firms are improving, they are never going to fully meet the growing need for novel biotech and other expensive therapies. And though China’s new policy mechanisms have cut the price of many medications, they can’t keep patients, physicians, and companies happy all at once.

That means China continues to rely on imports for some of the most advanced treatments — those with high price tags that even aggressive government policy can’t cut away. For the sake of its citizens and broader social stability, China will have to navigate the difficult balance between affordability, access, and market power in the global pharmaceutical landscape.

The need for reform

In the 1990s, financial pressure to offset the losses from government funding cuts drove public hospitals to inflate prices and over-prescribe, causing widespread problems. The Chinese Food and Drug Administration (now known as the National Medical Products Administration, or NMPA) suffered from incompetence, corruption, and a large backlog of drug approvals. Pharmaceutical companies marked up drug prices and leveraged kickbacks1 to generate sales. In fact, generic drugs2 — which made up 95% of China’s drug approvals at the time — sold at gross profit margins of 80-90%.

At the same time, between 2009 and 2017, China’s total health expenditure on pharmaceuticals more than doubled, from 754.38 billion RMB to 1,820.3 billion RMB (or about $US110.45 billion to $268.76 billion).3

These trends were unsustainable for a growing China.

The Healthy China 2030 plan, unveiled in 2016, recognized the need for reform in the “three spheres” 三医 of healthcare, medical insurance, and pharmaceuticals.

To improve the quality of pharmaceuticals, the NMPA implemented Generic Quality Consistency Evaluations4 in 2015 to assess generic drugs relative to their original brand-name counterparts. In 2017, China joined the International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use (ICH) and started adhering to the ICH-GCP (Good Clinical Practice) guidelines. Additional reforms streamlined and strengthened the drug approval regulatory process.

To improve the affordability of pharmaceuticals, China has begun to leverage two systems: the National Reimbursement Drug List (NRDL) and centralized volume-based drug procurement (CVBP).

National Reimbursement Drug List (NRDL)

Established in 2018, the National Healthcare Security Association 国家医保局 (NHSA) determines which patented drugs are included on the National Reimbursement Drug List (NRDL) and therefore covered under public health insurance schemes, which cover over 95% of China’s population to some degree.

How the NRDL works:

• The NHSA assesses novel drugs based on safety, efficacy, affordability, and clinical value.

• Drugs that demonstrate both high clinical value and low cost are categorized as Class A on the List and fully reimbursed by public insurance. Class B Drugs are clinically effective but higher cost, and thus only partially reimbursed.

• Innovative drugs, orphan drugs, and drugs for rare diseases receive priority.

• Nevertheless, innovative drugs that exceed a certain price threshold – including the most groundbreaking new cancer treatments – remain excluded.

• The current 2024 NRDL includes 3,159 drugs, including 91 new additions.

Centralized Volume-Based Drug Procurement (CVBP)

The National Reimbursement Drug List improves access to clinically important drugs, especially new innovations.

China takes a more aggressive cost-cutting approach to generic medications. Piloted in 2018 and rolled out nationwide in 2019,5 the national centralized volume-based drug procurement system 国家组织药品集中采 combines the buying power of all public hospitals and leverages it to negotiate the lowest possible prices from suppliers.6

How CVBP works:

• Every few months, the government7 invites suppliers to engage in a competitive bidding process for a selection of drugs.

• Drugs submitted for bidding must first pass Generic Quality Consistency Evaluations, and then win by offering the lowest price.

• Winning suppliers gain the right to sell a certain volume of medications — usually up to 70% of the previous year’s total consumption — to public hospitals. Domestic firms overwhelmingly win procurement rounds due to their ownership of raw-material manufacturing, large production capacities, and overall economics of scale.

• The National Health Commission monitors hospitals and physicians to ensure that they prescribe bid-winning drugs.

Since the two pilot programs, the government has held 10 centralized volume-based procurement rounds and 8 iterations of the National Drug Reimbursement List and continues to fine tune both programs.8

Pills without the price pain

So, have the government’s programs worked?

If price reductions were the goal, then yes.

The National Drug Reimbursement List’s first round achieved price reductions of over 50% on treatments for chronic hepatitis B and lung cancer. Subsequent rounds achieved similar reductions in the 50-60% range, as the following chart indicates.

The pilot round of centralized volume-based procurement had similar results, with an average price cut of 52% for 25 drugs. A hepatitis B drug experienced the highest price cut — a whopping 96%, for a final bidding price of US$0.09 per tablet. That same drug would cost US$10.94 in the United States and US$15.84 in the United Kingdom.

The CVBP program has allowed China to access some of the lowest prices for generic drugs in the world.

Between 2018 and 2023, generic drug manufacturers cut prices to win contracts, on average, by over 50%, and sometimes over 90%. Studies of the “4+7” pilot show evidence that after a procurement round, consumption of quality-assured bid-winning drugs increases and overall drug spending decreases. Here are some notable examples:

• Insulin: median price reduction of around 42% for a 2021 round applied to 42 insulin products — improving affordability9 and saving an estimated total of US$2.85 billion in the first year of contracts with the winning suppliers.

• Lung-cancer treatment: price reduction by 83%, from 108 RMB to 18 RMB, saving patients an estimated 8,100 RMB per treatment cycle.

• Stents to treat coronary heart disease: price reduction by 90%, from over 10,000 RMB (US$1,500) to around 1,000 RMB (US$150). As a result, the number of stents supplied to patients grew, and more medical institutions — including 500 additional second-tier hospitals10 — carried out stent implantation surgeries.

What do lower prices mean?

Government statistics estimated that as of 2022, CVBP created national savings of over 260 billion RMB (approximately $36.3 billion USD). The National Healthcare Security Administration (NHSA), which oversees public health insurance, reallocates over half of its savings from generic drug procurement to cover innovative medicines through the National Drug Reimbursement List.

In terms of affordability, one analysis of the pilot program’s impact found that the proportion of affordable drugs increased from 33% to 67%, and the mean affordability improved from 8.2 days’ wages to 2.8 days’ wages. Urban residents benefited more, a reflection of larger urban-rural healthcare disparities. In theory, improved affordability means patients are more likely and better able to adhere to their prescribed treatments.

Still, impacts on patient health outcomes have yet to be comprehensively evaluated. And the depth of the latest cuts has caused concern among both patients and physicians.

Saving cents, losing sense

In December 2024, the largest procurement round thus far resulted in the steepest price cuts yet, with some products discounted by over 90%.11 Not everyone was happy about this.

“You get what you pay for” is a common sentiment for consumers. So when generic aspirin tablets drop to an astonishing 0.03 RMB (US$0.0041) per pill, people start wondering what was sacrificed in pursuit of lower costs. On Chinese social media, patients have vocalized concerns that the generic drugs are less effective or cause more side effects.

One particular phrase by Zheng Minhua 郑民华 — surgeon-director of Shanghai Ruijin Hospital and member of the Chinese People’s Political Consultative Conference — went viral recently: “blood pressure doesn’t drop, anesthetics don’t bring sleep, and laxatives don’t release shit” 血压不降、麻药不睡、泻药不泻.

The phrase is part of a proposal submitted early this year by Zheng and his colleagues outlining concerns about the efficacy, reliability, and flexibility of bulk-procured generic drugs. Another doctor’s article — which has since been taken down — questions the validity of data published from evaluation trials of procured drugs.12 Some industry participants report that companies are able to replace excipients (non-active ingredients) of drugs after passing consistency evaluations without re-doing testing.

The government has since responded. However, these controversies come at a moment of broader frustration with China’s healthcare system, which stringent COVID-19 lockdown policies and sluggish post-pandemic economic recovery has only made worse.

Can you ever cure corruption?

Even with the government’s attempts to keep drug prices down, per capita medical costs have more than doubled in the last decade. While rising costs likely stem from a variety of systemic and economic factors, in China, corruption is a standout issue that the government can’t ignore.

In 2023, the National Health Commission initiated a crackdown on corruption, including bribery, misuse of insurance funds, rent-seeking by administrative officials, and unethical conduct. Over 160 hospital chiefs were detained.

Widely publicized, this anti-corruption campaign – which rewarded reports of corruption and the imposed strict monitoring of doctors – fueled distrust between the public and physicians.

Importantly, a majority of such medical corruption involves commercial bribery, in which pharmaceutical and medical suppliers give kickbacks to healthcare providers, leading to overprescriptions and inflated costs for patients. Legally, drug companies wield similar influence through heavy marketing, with one report showing that Chinese pharma companies had aggregate sales expenses of about 2.6 times that of R&D. Heavy workloads and low pay make doctors more receptive to drug suppliers’ incentives.

That pharmaceutical firms still turn to bribery and marketing to sway doctors suggests that government efforts to shape drug availability based on quality and cost have yet to fully succeed.

Despite the government’s efforts, domestic policy and industry still today falls short of the Chinese public’s desire for innovative drugs.

Imported drugs still matter

Accessing imported brand-name drugs can be a challenge — even if patients are willing to pay out-of-pocket. One reason may be Beijing’s focus on boosting Chinese drug manufacturers, actively cultivating their capabilities and favoring them during CVBP and NRDL selection processes.

This tension played out strikingly through Pfizer’s Paxlovoid, a treatment for severe COVID-19. Paxlovoid received conditional regulatory approval in February 2022 – so it was authorized to sell in China – but despite high demand, the drug never got included on the National Reimbursement Drug List.

So what happened during NRDL negotiations?

Officials cited Pfizer’s high asking price as the reason for no deal. Meanwhile, two COVID-19 treatments did make the list – the traditional Chinese medicine Qingfei Paidu and the domestically-produced antiviral pill Azvudine – reinforcing evidence of China’s preference for cheaper homegrown treatments.

Pfizer defended its price and rejected the notion of a future deal for domestic manufacturers to distribute a generic version of Paxlovoid in China, a common practice for low- and middle- income countries. “They are the second highest economy in the world and I don't think that they should pay less than El Salvador,” said CEO Albert Bourla at the time.

The Chinese public was left to deal with the fallout. Facing extreme shortages, people turned to the black market and to unproven Indian generic versions that sold for as much as 50,000 RMB, over twenty times the original price.

The case of Paxlovoid highlights both the extent of China’s progress and the gaps that still exist.

Yes, the domestic pharmaceutical industry is becoming increasingly formidable. But some important innovative drugs still remain beyond China’s borders, leaving it dependent on imports.

The Chinese government has made substantial – and in some cases, excessive – strides in ensuring medicine is affordable for its citizens. But novel pharmaceuticals come with high price tags that can’t be easily negotiated.

In 2024, the size of China’s innovative drug market13 reached a milestone of over 100 billion RMB (about USD $13.89 billion). Out-of-pocket payments or commercial health insurance covered over half of that market. It’s clear that China’s national medical insurance programs aren’t alone enough to meet China’s medical needs – especially if the goal is to provide the same level of care to its whole population that is expected in a G7 country.

Industry projections expect China’s combined market for innovative drugs and medical devices to exceed 1 trillion RMB (USD $137 billion) by 2035 – a full 30% of the global pharmaceutical market.

Is this a market opportunity for multinational companies? Or, will China be able to achieve its ideal world: a thriving domestic pharmaceutical ecosystem, managed by rigorous national insurance and bulk procurement programs, so that its citizens get the medicine they need.

Taiwan and China on the Leak

Lily Ottinger and Nicholas Welch report:

The Trump administration leaked its plans to bomb the Houthis by accidentally adding The Atlantic’s editor-in-chief to a Signal group chat. This situation calls for humor, so we curated a mini-roundup of reactions from the Sinosphere. Enjoy!

From the Taiwanese platform PTT:

(Pushed) “I wonder what the American ancestors would think when they see these unworthy descendants.”

(Pushed) “I shouldn’t, but I really want to see the Indo-Pacific war plan leaked…”

(Pushed) “This is too ridiculous. Could it be because the reporter just happened to have the same name as the person they wanted?”

Indeed, such an error would be more difficult to make in Chinese given the huge variety of characters used in Chinese names. From Weibo:

American names are too simple. … 😂

This is truly an epic level of face-losing disgrace. (这是史诗级别的丢人现眼)

The following comments come from another Weibo post about Trump’s response to the leak:

“I don’t know, that didn’t happen” has hoodlum energy (無賴嘴臉)

Shameless people are invincible (人不要脸，天下无敌)

Another commenter in that thread invoked the phrase, “Not listening! Not listening! The turtle is chanting buddhist scriptures!” (不听不听王八念经). This meme is originally from a nursery rhyme satirizing people who shut down when confronted with an opposing viewpoint.

Some other relevant Chinese idioms:

• 泄露天机 (xiè lòu tiān jī) — “To divulge the will of heaven”

• 掩耳盗铃 (yǎn ěr dào líng) — “Covering one's ears while stealing a bell”

• 事后诸葛亮 (shì hòu zhū gě liàng) — “To become Zhuge Liang after the fact” — Zhuge Liang was a military strategist widely regarded as a genius thanks to his portrayal in Romance of the Three Kingdoms. This is a bit like saying, “Everyone’s an Einstein in hindsight.”

Emergent Ventures Grants for Taiwan

Jordan Schneider reports:

Tyler Cowen’s Emergent Ventures fellowship is a special institution. It is a no-overhead fellowship where Tyler reads short essays and gives grants to individuals to support personal projects and career development. Emergent Ventures recently received a donation to start finding talent in Taiwan and is looking for applicants. If you are or know anyone who could use a grant to get a personal project off the ground, do consider applying and select Taiwan on the regional dropdown menu!

For some inspiration, see here for an article on Tyler’s selection critera and a site that has compiled brief descriptions of all the past winners who’ve done things like use AI to decipher ancient text, compose Bach-style fugues, cover local politics, and a thousand other things.

Emergent Ventures helped get ChinaTalk off the ground, giving me a grant in 2018 to buy microphones and Chinese lessons. I’ve since attended three conferences (which I count as some of the most exciting weekends of my life), and met up with winners from around the world. What I’ve taken most from this community is a renewed optimism in what the future can bring and an ambition that I can continue to improve and broaden what I do and really make a dent.

Mark Carney — China Nerd?

Irene Zhang reports:

It’s official: Mark Carney is Canada’s 24th Prime Minister. The prominent economist and political novice, who led the Bank of Canada through the 2008 financial crisis and headed the Bank of England from 2013 to 2020, won the incumbent Liberal Party’s leadership contest on March 9 and was sworn in as Prime Minister on March 14. (In Canada, the leader of the largest party or coalition in Parliament becomes Prime Minister, making Mark Carney an oddity as a PM who doesn’t represent a constituency of his own.)

Britain voted to leave the European Union in 2016, three years after Carney became the first foreign head of the Bank of England. In the ensuing years, Carney worked to open the UK to new trading opportunities — including China. He traveled to Beijing with Britain’s then-finance minister in late 2017 to seek a $1.34 billion trade deal, though a UK-China free trade agreement never materialized. Along the way, he’s had to wade through choppy waters. In August 2019, as Chinese paramilitary troops amassed near the Shenzhen-Hong Kong border and protests in Hong Kong reached a boiling point, Carney awkwardly pulled out of a high-profile dinner in London with China’s ambassador to the UK. He faced criticism from the British political arena for bending over backwards to promote Chinese investment at a time when China was repeatedly violating human rights in Hong Kong.

Subscribe now

Carney’s successful banking career comes with a special level of insight into China’s economic situation. He was bullish on the Chinese economy as of 2012, calling on Canadian firms to adopt China strategies and “reorient” toward China. But at least as early as 2018, he became deeply concerned about China’s domestic financial situation. During a BBC interview, he warned that China was one of the top risks to the global economy because its financial sector “has many of the same assumptions that were made in the run-up to the last financial crisis.” He elaborated on this in February 2019, telling a Financial Times audience that the possibility of a Chinese economic slowdown the second-most-important risk to global growth:

“While China’s economic miracle over the past three decades has been extraordinary, its post-crisis performance has relied increasingly on one of the largest and longest running credit booms ever, with an associated explosion of shadow banking. … The Bank of England estimates that a 3% drop in Chinese GDP would knock one per cent off global activity, including half a per cent off each of UK, US, and euro area GDP, through trade, commodities and financial market channels. A harder landing would have significantly larger effects, as these channels would likely be accompanied by negative spillovers to global confidence.”

Later that year he called the world’s reliance on the USD as the reserve currency “risky”, but said the Chinese RMB was far from ready to step in as a replacement.

Carney arrives in Ottawa with some knowledge of the top man himself — he met Xi Jinping twice while heading the Bank of England: once in 2017 and once in 2019. Last year, when rumors about him running to replace Trudeau first began to swirl, Carney met Xi Jinping again at the China Development Forum in March as head of Bloomberg’s board. These previous run-ins may not count for much in the short term, as Canada-China relations remain on ice. Just one day before Carney was elected to lead the Liberals — and by extension Canada, for now — China slapped 100% tariffs on Canadian rapeseed oil, oil cakes, and pea imports, as well as a 25% tariff on seafood. These are in retaliation to the 100% EV tariffs and 25% steel and aluminum tariffs Canada imposed on China last October, a coordinated move with the US at a time when relations between Washington and Ottawa looked very different.

What do you do when you get into a tariff war with your second-biggest trading partner to help your biggest trading partner, but now your biggest trading partner keeps talking about annexing you while trying to decimate your economy? You elect a central banker, of course.

Carney’s technocratic appeal was already compelling to voters as Canada struggled with productivity stagnation — now, the trade war is only adding to his political edge. A federal election is required to happen in Canada before October 20 this year. Before Trump made anti-Canadianism his newest fixation, the incumbent Liberals were set for massive losses while opposition Conservatives hoped for a majority in Parliament. But Conservative leader Pierre Poilievre’s previous amiability towards the American president has practically doomed him to quisling status only two months later, and now some polls predict Carney’s Liberals could even win a majority if the election were called today. The events have been quite jaw-dropping, and the polls still contain much volatility, but if Carney manages to cement his mandate, he will lead Canada through unprecedented headwinds on both American and Chinese frontiers.

Mood Music — Canadian Disco:

Why China Doesn’t (Really) Frack

Caleb Harding reports:

China’s natural resource endowments have been succinctly summed up as “Rich coal, poor oil, small gas” (“富煤、贫油、少气”). Despite claiming “small gas,” however, China actually has the world’s largest reserves of shale gas, an unconventional type of methane deposit that must be tapped with hydraulic fracturing. China is one of only four countries in the world with commercial fracking operations (the others being the USA, Canada, and Argentina), which began after President Obama shared fracking technology with the PRC in 2009.

Chinese media acknowledges that the shale gas revolution was a game changer for U.S. energy supply and security, but China has yet to realize its own revolution with its immense reserves — why? The simple answer is geology.

China’s shale reserves are deeper, more scattered, and in more mountainous terrain than those in the United States. Guo Tonglou 郭彤楼 is a chief engineer at Sinopec, one of just two companies involved in commercial scale extraction of shale gas in China. Commenting on China’s situation, he said that, “Although we are a country rich in shale gas resources, shale gas extraction in our country is difficult and costly. This means that it is impossible for us to adopt the same development approach as the United States" (虽然我们是页岩气资源大国，但我国页岩气开采难度大，成本高。这就意味着，我们不可能采取美国那样的开发方式).

Shale wells in Sichuan (home of China’s main shale plays) and Texas. Cherry-picked, yes, anecdotal, yes, but makes the geological difference a bit more visceral.

Other barriers exist, including a lack of natural gas infrastructure and imperfect legal/policy frameworks, and China has been working to address those issues.1 But the fundamental issue remains the economics and geology at the well — if it was cheaper to extract, China could rapidly expand their natural gas infrastructure, just as the U.S. did in the decade after 2005. But that seems unlikely anytime soon.

But that hasn’t stopped Beijing from announcing ambitious targets for fracking. In 2018, the Chinese Ministry of Finance and the State Administration of Taxation introduced a preferential tax policy to reduce the resource tax on shale gas production to 4.2% from 6.0%, which in 2023 was extended through December 2027. Following the release of China’s 14th Five-Year Plan in 2021, policy directives continued to support the development of unconventional natural gas resources, including adjustments made to the pipeline tariffs last year, and a new subsidy policy released just this month that incentivizes higher production.

Rather than hoping for a shale revolution, Guo Tonglou’s aspirations for Chinese shale gas are modest — he’s optimistic that, with more technical breakthroughs, shale gas could account for ⅓ to ½ of Chinese natural gas production. That would be an increase of their current production from shale from 25 bcm to 100-200 bcm annually, or from ~0.6% to ~ 2-4%2 of China’s overall energy. In contrast, the U.S. produced 836 bcm3 of natural gas from shale formations last year, which could provide around 30% of the U.S. overall energy.

But is the juice really worth the squeeze? With a comparative advantage in renewables and coal, why invest in such a difficult domestic energy source? As one CRS report estimates, China’s demand for gas through 2030 can be met under existing contracts, and demand through 2040 can be met without having to increase trade with Western suppliers.

Economically, it doesn’t seem to make much sense. But in the age of great power competition, economics are no longer the primary concern — security is. And as the world’s largest importer of LNG, China has valid concerns. Shale will likely never reach the scale in China that it did in the U.S. But until China feels totally secure in its energy supply, its development will continue.

Chinese is Hard, Even for Natives

Moly from Chinese Doom Scroll reports:

Question: “Why does the 8th line on the Chengdu subway have an announcement reminding people to not bite fellow passengers?” [撕咬 — sīyǎo, to bite.]

A response from Chengdu Subway’s official account: “Uh! Sweetie. The subway is saying, ‘Do not broadcast noise from electronic devices to annoy fellow passengers.’ [滋扰 — zīrǎo, to annoy. In local dialect, it might be pronounced very similarly, I have no idea.] Create a civilised city, be a civilised citizen. The subway invites you to travel in a civilised way.”

“Not only are you not allowed to bite fellow passengers, you also have to be careful of the spy between the train and the platform.” [“Spy” “奸细” jiānxi here sounds like “gap” “间隙” jiànxì]

“I’ve also wondered forever why Chengdu’s East Station bans Cadillacs. [凯迪拉克— Kǎi dí lā kè] Does it actually mean something?”

“Are you sure it’s not, ‘No solicitations?’” [喊客拉客—hǎn kè lā kè]

“Shanghai buses play announcements, ‘Please dance when the door opens.’ It’s very confusing.”

“‘Please be careful as the doors open.’ They should also broadcast the same thing in Shanghaiese.”

“I came to Shenzhen for uni and heard the subway broadcast, ‘Watch out for kasayas,’ [袈裟 — jiā shā] And thought it meant to watch out for scammer monks or something.”

“So what was it actually?”

“To avoid getting pinched? [夹伤 — jiā shāng] Probably?”

“It’s okay, at least it’s better than the Shanghai subway. The announcement there is, ‘No bathing, performing, social, San Francisco.’”

“Huh? So what the hell is it talking about?”

“No begging, performing, selling, or passing out ads.”

“So what does, ‘Please go through the fried chicken passage.’ mean?” [炸鸡 — zhájī]

“Go through the turnstiles.” [闸机 — zhájī]

“There’s a minnan dialect announcement that just goes Lou Cha Lou Cha endlessly…”

“Lou Cha means depart hahahahahaha”

“Our maths teacher: Do not bite the classroom.” [干咬 — gānyǎo. I’m pretty sure it was supposed to be 干扰 — gānrǎo, disrupt the classroom.]

“The Beijing subway told me to not use yellow corpses.”

“Then I figured out it meant, ‘Do not use child leashes.’”

“One time, I was passing by an intersection and heard someone calling out, ‘Crazy~ Bread rolls~ Crazy~ Bread rolls~’ And I was like, ‘Just how crazy can bread rolls get? I gotta check it out!’ So I turned the corner to see the sign on the car, ‘Honey bread rolls.’ Someone was just calling out ‘Honey~ Bread rolls~’ with a dialect.”

“I’m the only one here with a straight answer. Biting is too cruel. We don’t engage in such inhumane methods. We usually prefer to swallow whole.”

Sergey Radchenko’s book, To Run the World: The Kremlin’s Bid for Global Power, is a masterwork! In my mind, it’s in pole position for best book of 2025. Sergey takes you into the mind of Soviet and Chinese leaders as they wrestle for global power and recognition, leaving you amused, inspired, and horrified by the small-mindedness of the people who had the power to start World War III.

We get amazing vignettes like Liu Shaoqi making fun of the Americans for eating ice cream in trenches, Khrushchev pinning red stars on Eisenhower’s grandkids, and Brezhnev and Andropov offering to dig up dirt on senators to help save Nixon from Watergate.

Sergey earns your trust in this book, acknowledging what we can and can’t know. He leaves you with a new lens to understand the Cold War and the new US-China rivalry — namely, the overwhelming preoccupation with global prestige by Cold War leaders.

In this interview, we discuss…

• Why legitimacy matters in international politics,

• Stalin’s colonial ambitions and Truman’s strategy of containment,

• Sino-Soviet relations during the Stalin era and beyond,

• The history of nuclear blackmail, starting with the 1956 Suez crisis,

• Why Khrushchev’s vision of abundance couldn’t save the Soviet economy.

Co-hosting today is Jon Sine of the Cogitations substack.

Listen now on iTunes, Spotify, YouTube, or your favorite podcast app.

Why Authoritarians Need Legitimacy

Jordan Schneider: Let’s start with prestige. How do you define it, and is it anything bigger than just schoolyard dynamics?

Sergey Radchenko: That’s a deep question, and that’s where the Soviet experience is not actually unique. If you think about the Soviets pursuing prestige, what’s so unique about that? Everybody does that. People strive for status, people strive for recognition, and states do the same thing.

That is why in the introduction to my book I try to zoom out a little bit — I’m trying to say is that the striving for recognition is not something unique to Soviet leaders. In fact, I show that even during the Cold War, the Soviets were not unique. The Chinese also had similar motivations and sometimes also the Americans.

We’re starting to throw around difficult terminology — recognition, prestige — what do those things mean? To add to the complexity, what the Soviets really wanted was legitimacy. Everyone wants legitimacy, but the Soviets had a particular need for legitimacy perhaps because of their lack of internal legitimacy. They had a project on their hands. They wanted to pursue a communist revolution. They were trying to sell it to the Soviet people saying, “Look, soon we will be living under communism. Everything will be so fine. We’ll all be so wealthy.” The abundance, material abundance was supposed to come soon — and it didn’t.

It didn’t arrive because the Soviet project had already started losing steam by the late 1950s. It became increasingly clear that this was happening, creating a deficit of internal legitimacy. Because of that, they wanted external legitimation. They wanted others to say, “You’re great, you’re wonderful, you’re a superpower,” because then they could turn to their own people and say, “Look, we are so strong, we’re a superpower, we’re great because they say so.“

Who are they? Of course, the ones in a position to recognize — the Americans. Others mattered too. If Mongolia or Papua New Guinea said, “This is the great Soviet Union,” that was important. But if the Americans said that, it was all the more important because they were themselves a superpower, so their recognition was absolutely crucial for the Soviets’ sense of greatness. That is where those things connect — prestige, recognition, external recognition, and legitimacy.

Jon Sine: I’m trying to understand what legitimacy from an international perspective gets you. I’m reminded of the phrase, “Foreign policy is domestic policy by other means.” How is a foreign policy of recognition — this external validation — important from a domestic standpoint? Is this what is driving it, or is there some other source driving this need for recognition and legitimacy?

Sergey Radchenko: There may be an acceptable definition — legitimacy is legality and justice put together somehow. It’s a term that reflects legality — a legal position — and justice, meaning you deserve that position.

The Soviets really wanted to be seen to occupy a place they deserved. They tried to sell this to their own people, saying, “Look, we are in this position, we’re great, we deserve to be great, they recognize us as great, we are legitimately great.” This is to be distinguished from raw power.

I show in the beginning of the book how Stalin sometimes actually wanted legitimacy instead of raw power. In some cases where he controlled certain territories, he would back off because he wanted legitimation of his control — legitimate control, not just absolute control by sheer power.

I can give you a couple of examples. One example would be the separatist rebellion in Xinjiang. Stalin supported the rebellion in Xinjiang but then backed out. He basically sold the rebels down the river and engaged in a relationship with the central government in China. Although that meant he lost immediate control of Xinjiang, he acquired a certain sense of legitimacy for his other claims in China.

That is where power and legitimacy were in a state of interplay, and sometimes — not always — but sometimes Stalin would prioritize legitimacy over power. That’s not to say he would always do that. There was a certain bottom line where he would say, “No, I refuse to give up that particular thing because it’s absolutely essential for my security or what I feel is necessary for the USSR.” But sometimes he would say, “I will give this up if you recognize that whatever else I hold here is legitimate,” and that is the essence of the Yalta framework that was constructed in 1945, which Stalin was very fond of and wanted to preserve.

Jon Sine: Staying on the Yalta agreement, one of the things that quite surprised me in the book is how interested Stalin was in a concert of powers and having his view of the USSR’s proper sphere legitimated. Could introduce the percentages agreement?

Sergey Radchenko: The percentages agreement is a well-known episode in the early history of the Cold War. Churchill went to Moscow and met with Stalin in October 1944. Churchill in his memoirs recounts it by saying basically, “I wrote down percentages of influence that the Soviet Union and Great Britain would have in southeastern Europe, and then I gave the paper over to Stalin. Stalin made a tick on it, handed it back to me, and I told Stalin, ‘Maybe we should destroy this piece of paper, because isn’t that very cynical that we basically decided the fate of millions of people in this way?’ And Stalin said, ‘No, you keep it.’”

What’s interesting — this story, of course, is decades old. We’ve all known about it from Churchill’s memoirs. But this is where writing international history becomes really interesting — now we have the British version of the record of conversation, we also have the Russian version of the record of conversation. We have the actual piece of paper in the Churchill archives, the percentages agreement, and crucially, something that has not really been considered before: we have extensive discussions that followed that agreement between the Soviets and the British about specific percentages.

When I was reading those discussions, I thought, “This is so strange, the Soviets really took this stuff seriously.” They argued about an extra 5% in Hungary, this kind of stuff. You think, “Really? What does it even mean?” It sounds bizarre, it sounds absurd, but that shows that Stalin really took that seriously. He felt that he had a legitimate sphere of influence in Eastern Europe. He was very cynical about it, but he basically expected the British and the Americans to defer to his desires in Eastern Europe, and in turn, he would accept the British and American sphere somewhere out there.

He wanted his sphere recognized externally. The problem was that what Stalin considered a legitimate sphere of interests or sphere of influence was not seen as legitimate by the Americans, in particular. That explains a lot about the way the Cold War unfolded.

Jon Sine: Throughout the rest of the book, it’s not clear if any of this is really worth it. But in this particular instance, when you talk about America pushing back on Stalin, you do seem to think that it was very necessary for America to have done that. Basically the Cold War, in a sense, was not inevitable from an ideological perspective, but rather happened because Stalin was going to push until he found resistance. Could you elaborate on your views on that?

Sergey Radchenko: It’s a difficult question. Historians have looked at the early Cold War and have tried to understand whether Stalin was perhaps open to compromise. Some have said that he was looking for great power compromise, and I’m also inclined towards this view. Others would say, “No, we had to stop Stalin’s expansionism, and Stalin could never be trusted,” and so on.

The problem from the US perspective is we’re all wise in retrospect — we can look at the Soviet archives and hopefully understand what Stalin actually was thinking. But the problem at the time is you are facing this difficult person called Joseph Stalin. Nobody knows what he’s thinking. Nobody knows what his ambitions are. Does he want to take over the world or not? Does he want to take over Europe? I show in the book that Soviet post-war planning entailed Soviet control over much of continental Europe, as far as Sweden. They had great appetite.

There’s great uncertainty. You’re in the present moment, trying to deal with the situation, you don’t know what the other guy thinks, so what is the safe policy to pursue? I cannot imagine a safer policy than that which was actually pursued in terms of what George Kennan formulated in The Long Telegram, the policy of containment, the policy of pushing back.

Did that help bring about the growing distrust with the USSR and in some ways precipitate the Cold War? Maybe. Of course, if you’re pushing back on a guy, he’ll think, “Look, they’re aggressive, they’re pushing back on us.” But if you don’t push back on the guy and it turns out he’s actually determined to take over the world, you’ve got a problem.

In the end, what happened — containment — is, I guess, the best worst solution in the situation that we had in the early Cold War. It’s a tragic situation. It’s a situation that results from the lack of trust and from inability to understand or know in real time what the other side is thinking. We can project this onto our own days and ask similar questions, but maybe we can do it later.

Jordan Schneider: It’s fascinating how you see these leaders ending up resorting to the lowest common denominator of understanding the other side. Stalin basically says, “The West only understands force. When you have an army, they talk to you differently. They will recognize and learn to love you because everybody loves force.” Truman basically did the same thing. He was like, “I got a bomb, they gotta listen to me. I can bully him however I want.“

There are moments over the course of the Cold War where leaders start to feel their way to slightly different dynamics, where it’s not just this whole idea of prestige and national power as this gas that inevitably expands until you hit another great power that’s going to push back on you. Perhaps there are different ways in which you can look at those points of contention and try to devise different operating systems.

It’s remarkable and almost speaks to the argument of prestige being this gaseous force that these two countries who had just fought a world war together were not able to figure out how to get their interests to align in a way that wouldn’t lead them down to what the world ended up having to live with for the next 50 years.

Sergey Radchenko: It’s partly that, Jordan, and partly just a different conception of the world. We can get attached to the person of Stalin and say that he was paranoid and crazy, but if you look at Soviet post-war planning, it was based on a kind of 19th century imperial model. The Soviets get to control a lot of territory like the Russians when the Russian empire was expanding. Stalin was thinking in 19th century terms, very unremarkable given that particular historical period.

Churchill was like that, by the way. The Americans were not thinking in those terms. They were coming from a different perspective, and they did not want to allow the Soviets to basically put Eastern Europe into their pocket and walk away. From that, you already see a clash of visions for the post-war order, multiplied by the lack of trust, multiplied by security concerns. You can see how you would have the slide towards confrontation

There were also other factors. One point I raise in my book is that early on — 1945, 1946 in the early post-war world — Stalin was hopeful that the communists would come to power in many places unaided. People would just vote for them because they were so popular and wonderful. Then it turned out that in Eastern Europe nobody thought that, so he had to falsify elections in Eastern Europe. This raises a question. It becomes much more difficult to pretend you can have some kind of cooperation or a coalition of governments.

The same thing happens in Western Europe where the Italian Communist Party and the French Communist Party were initially very popular but then ultimately had to effectively leave the government. Stalin realizes that he will not be able to obtain more pro-Soviet governments or Soviet leaning governments through coalition partnerships and resorts to brute force. That’s another factor I wanted to highlight.

Jordan Schneider: If Stalin is expecting 70% of Europe and the West is expecting 70% of Europe, then was there ever really a way to find some happy medium if their desires of what they wanted the world to look like after 1945 were fundamentally incompatible?

You zoom in very directly to 1947, where Stalin wakes up to the fact that unless he starts playing hardball, he might be losing on a lot more fronts than he was hoping to. But in the hypothetical where communism happens to be more popular and win elections, is this something the CIA even stands for in the first place? It’s hard to imagine, if that is Stalin’s base case, that you really could have ended up anywhere else.

Sergey Radchenko: Exactly. That’s why I’m so cautious. I’m not a revisionist historian saying, “If only the Americans behaved themselves better and understood Stalin or accepted Stalin’s sphere of influence, then it would all have turned out fine.” Maybe it would have, but another thing we need to keep in mind is that Stalin measured his appetites in part as a result from the policy of containment.

If the Americans were not pushing back, who knows whether his appetites would have stayed where they are or would have extended even further. That is why I’m saying we don’t know, and he himself may not have known at that time. That’s why containment was the safe policy.

It is a tragic outcome. Yes, it does contribute to the Cold War, but who could come up with a better policy than containment when you don’t know who you’re dealing with on the other side? Frankly, looking at Stalin’s paranoia and his dark view of the world and his obsession with conflict, it would have been a very unwise policy to try to indulge Stalin too much. I’m saying that as somebody who’s read a lot of Stalin. I think containment was a wise strategy.

Jordan Schneider: I love this little riff you go on in your book, which is a historical wrinkle I wasn’t aware of, where Stalin was like, “Let’s see if we can get some colonies in Libya.” Can you explain that?

Sergey Radchenko: It’s remarkable, isn’t it? Absolutely remarkable. Here again, 19th century European imperialism and colonialism — that’s how Stalin was thinking. “Oh, Africa? Of course. Those other guys have colonies in Africa. Why shouldn’t we have a colony in Africa? We can try our hand.” I think Molotov said something along the lines of, “We can try our hand at colonial administration.”

They, by the way, early on felt that the Americans had promised them something along those lines. Then when James Byrnes became Secretary of State, it turned out that he didn’t want to go in that direction, and Stalin felt his expectations were betrayed.

Why did he want this colony in Africa? Strategic reasons maybe. More important are the issues of prestige that we talked about, because if you’re a great power, of course you want to be in Africa. That’s where all the great powers made their imprint. That’s why he wanted to go there.

Jordan Schneider: Let’s turn to Asia. What happened with Hokkaido?

Sergey Radchenko: The Hokkaido story is fascinating. Basically, Stalin got this idea that the Soviets should help in the liberation of Japan and proposed effectively landing on the island of Hokkaido.

Remember, in 1945, the Japanese controlled not just Hokkaido, but also what they called Karafuto, or southern Sakhalin Island — that’s the place where I grew up — and the Kuril Islands as well. The Soviets liberated those islands as part of the Yalta agreement, which was stipulated in the agreement.

Stalin then said, “We can also land on Hokkaido. There’s a wonderful island off Hokkaido. Let’s liberate Hokkaido as well.” He proposed that and sent a telegram to Truman. Truman basically said, “There’s no way this is going to happen,” and remarkably, Stalin backed off.

You have to ask the question, why did he back off? The obvious answer — we’re talking about mid-August 1945 — would be that he was afraid of the atomic bomb. That’s one possible answer. Is it credible? There’s something to it.

Another possible answer is that he was interested in preserving some sort of cooperative relationship with the United States. The Yalta agreement was in place, the Americans respected it, and he wanted that recognition and that legitimacy offered by Yalta.

In fact, he continued for some years, until 1949, almost 1950, to abide by the principles of Yalta because he wanted American recognition of the legitimacy of his gains. When Truman said, “No, sorry, you cannot. That’s just too far. You were never there, do not even come close,” Stalin backed off.

Subscribe now

It’s interesting — it wasn’t just fear of the nuclear bomb, it was also concern about maintaining a cooperative relationship with the United States. It could be both. History is multi-causal.

Socialist International Relations（唇亡齿寒？）

Jordan Schneider: Another example of Stalin not going all in right after 1945 was the way he handled Mao and the nationalists. Can you talk about how Stalin handled China throughout the civil war?

Sergey Radchenko: That is a very interesting aspect of Soviet foreign policy which I think is understudied. If you look at histories of the Cold War, they all focus on Europe, on the German question more or less, and few people actually examine China.

What you see in China’s case is something quite remarkable. Soviet involvement with China predates the Second World War and goes back to the 1920s. The Soviets were helping both the Kuomintang and at one point were crucial to the establishment of the Chinese Communist Party. Then you have this period of prolonged confrontation between the Kuomintang and the Chinese communists.

The Chinese communists, following the Long March, retreated towards Northwestern China and were headquartered in a place called Yan’an. That is where the end of the Second World War finds Mao. He receives a telegram from Stalin in August 1945 saying, “Go negotiate, go to Chongqing,” which is the military capital of China, “Negotiate with Chiang Kai-shek for a coalition government.“

You can imagine how Mao would react to this. “What? Hang on just a second. We are not friends. We have had a very difficult relationship.” Yes, there was a period of united front and during the war against the Japanese they even cooperated in some ways, but Mao was not keen to have this sort of relationship with Chiang Kai-shek or to enter into a conversation with him.

Mao was basically forced by Stalin to do that because Stalin had concluded the Treaty of Alliance with Chiang Kai-shek. That was a result of very painful negotiations in Moscow. Chiang Kai-shek sent T. V. Soong, who was the head of the Executive Yuan, to Moscow for those negotiations.

What Stalin was able to wrestle from Chiang Kai-shek was not only concessions in China — there’s a discussion of the railroad that the Soviets sold to Manchukuo but now wanted back — but also, crucially, he got Mongolia. The Soviets had been controlling Mongolia for a long time, but de jure it remained a part of China. Stalin wanted Chiang Kai-shek’s government to say, “Okay, we give up on Mongolia.” That was arranged in Moscow in 1945. These were massive concessions.

What did Chiang Kai-shek want in return? I mentioned one of those demands: the end of Soviet support for Xinjiang separatists, and Stalin agreed. Stalin claimed, “We don’t know what’s happening in Xinjiang. This is not us.” Although the whole war was being fought with Soviet weapons and instructors. He said, “No, no, no, that’s not us.” Basically, he gave up and betrayed those separatists in Xinjiang, saying, “Okay, we’re done with you.“

Then the question of Mao Zedong came up, and Chiang Kai-shek said, “Look, you’re supporting Chinese communists.” Stalin replied, “Well, Mao is not really a communist, he’s just a nice guy but he’s not a communist.” As a result, Stalin got his gains in China, he got Mongolia, and he basically told Mao to go negotiate with Chiang Kai-shek.

Stalin had to understand what this meant. If Mao Zedong negotiated with Chiang Kai-shek for a coalition government — we’ve been there, we’ve done that already before. Mao did not have a superior force, and Stalin understood this. Stalin thought that the Kuomintang was a much stronger force in China in 1945. This basically meant surrendering to the Kuomintang. I think Mao also understood that.

There’s this very interesting period from 1945 until the Chinese Civil War broke out again where Stalin was uncertain who he would support or what his position in China should be. He was interested in a solid relationship with the Kuomintang because they had guaranteed his control of Mongolia and those parts of his imperial interests, ratified by the Yalta agreement. That was much more important to him than Mao’s cause and communism.

Later he changed his mind, but that was because the Cold War started to unfold in Europe and he thought that maybe he should have stuck with Mao Zedong after all. But that took place over a period of time, which is why Mao later said on numerous occasions, “Stalin blocked our revolution and regarded me as a half-hearted Tito.” Mao had every right to say that, because Stalin did not believe in the victory of the communist revolution. When it happened, it was a great surprise for him.

Jon Sine: America also sent George Marshall (there’s a great book on this called Marshall’s Mission to China) to try to ensure this compromise between the communists and the nationalists. You’re telling us we have Stalin pushing on the communists to make this work. So what goes wrong?

Sergey Radchenko: What doesn’t work is that Mao does not want to have anything like that in place. Stalin has limited leverage over him — he cannot force him to agree to a coalition government. He can only advise him to have a coalition government.

They have long discussions in the fall of 1945 when they meet in Chongqing. There’s extensive discussion there about what future China might look like. The Chinese communists actually propose effectively dividing China along a north south scenario where the north is controlled by the communists and the south is controlled by the Kuomintang.

Chiang Kai-shek cannot accept this because he is a nationalist, a patriot. He does not want a division of China like that, so he turns this down. They never could agree, and then ultimately you’ve got George Marshall’s intervention in the period of quiet.

Later, Chiang Kai-shek blamed George Marshall for interfering too much, claiming that if his hands weren’t tied, he would have been able to wage war much more effectively against Mao Zedong and would have defeated him.

That’s a whole different story as to why nothing works out and why the civil war is won by the communists. Many historians have written about this — it’s not part of my book — but there were serious problems with the Kuomintang government and the state of the Kuomintang army. That’s the bottom line.

Jon Sine: Let’s fast-forward to the late ‘40s, at which point the communists are moving into Manchuria. You have this piece of evidence — I think it’s Stalin writing to a general operating in Manchuria on behalf of the Soviets — and he says something to the effect of, “If Mao and the communists move in and try to take any of the material, open fire or use force to prevent this.” This really speaks to your narrative. Could you explain what that source is and what Stalin was doing there?

Sergey Radchenko: That comes from one of Stalin’s telegrams. I think it was to Rodion Malinovsky, who later became the Soviet Minister of Defense if I recall correctly.

The situation was that Chinese communists were trying to get into Manchuria because that’s a very important base to have, and they’re trying to occupy Manchurian cities. The Kuomintang government was obviously determined to prevent them from doing that.

The Soviets were actually in control at that time in Manchuria before they ultimately withdrew. Stalin instructed his forces on the ground to deny Chinese communists the ability to enter the cities. This is remarkable if you think about it because, aren’t they supposed to help Chinese communists? They’re the Soviet Union, a communist country — aren’t they supposed to help communists?

The communists say, “We want to get into the cities, we want to control Manchuria,” but Stalin sends a telegram saying, “Keep them out. Open fire if they try to do that.” I can’t remember the exact term that Stalin used — it’s like “these people” or something like that. It’s as if he’s saying, “I don’t trust these guys.” He says, “They are trying to get us into conflict with the United States.“

Is he right? Of course he’s right. The Chinese communists would have been very happy at this point if the Soviets and the Americans came to blows. But that’s still too soon right after the end of the Second World War, and Stalin is of the mind that he should: A) avoid a conflict with the United States, especially over China, and B) work with the Kuomintang government.

He is pursuing a policy of great duplicity. He is tolerating the Chinese communists in the countryside, but he’s keeping them out of the cities. The Soviets were helping communists by providing them supplies over the border. I don’t want to say that he completely rejected the Chinese communists — he was basically playing both sides.

It’s very clear that he was in fact trying to keep communists from taking over cities in Manchuria, and I think that is really a new piece of evidence that contributes to our understanding of what his China policy was from 1945 until 1946

Jon Sine: Let’s zoom out and talk about Stalin’s China policy from a macro perspective. I was reading Stephen Kotkin’s book, the first volume of his Stalin biography, and he basically pillories Stalin’s China policy. He looks into what the Soviets were themselves talking about back in the ’20s when Stalin was suggesting that the communists should ally with the nationalists, which ended in an absolute massacre of the communists by the nationalists.

Kotkin’s view is that Stalin’s China policy has been a string of really atrocious decisions about how to operate there. How does this ultimately reverberate with Mao, thinking about his position vis-a-vis Stalin, someone who has really screwed him in a number of ways, and who then has to deal with Mao coming into power in ‘49? Mao has to maintain a sort of subservient position to Stalin as the long-term leader of the communist camp. How does that all play into this?

Sergey Radchenko: Clearly Stalin and Mao did not really like each other, and Mao had very few good things to say about Stalin. However, he understood that Soviet support was really important for the long-term survival of the communist regime in China and for recognition of the communist regime. Here I mean literal recognition because once the Chinese take over, they need recognition from the Soviets.

The Soviets were very keen to offer that right away, although not without some side stories. For example, when the Chinese communists take over Nanjing, the Soviets follow with the retreating Kuomintang to Guangzhou. The Chinese communists are asking, “What are you doing? Even the Americans are staying in Nanjing!” The Soviets respond, “You don’t understand strategy.”

For Mao, it was essential to get Stalin’s support, Stalin’s diplomatic recognition of the Chinese communists as a legitimate regime, but also aid — Mao counted on Stalin’s aid in the reconstruction of China and on Soviet protection.

There are all kinds of good reasons why, despite his not particularly good feelings towards “Comrade Main Master,” as Mao called Stalin, he still went along and respected Stalin. I think I say in the book that it’s almost in the way that, in a large family, perhaps a son would respect his father in a Confucian way, although Mao always liked to emphasize that he was not a Confucian. It’s almost like he’s paying respect without really liking Stalin very much at all.

I don’t know if you saw that in the book, but I dug out this absolutely hilarious document which I quoted from. Mao sent his wife, Jiang Qing, to Moscow, and she was there in 1949 undergoing medical treatment. She wrote letters to her “beloved chairman” and these letters are in the Stalin archive in Moscow, which obviously shows that they were being read and delivered to Stalin.

The funniest thing is, Jiang Qing surely knew that they were being read, so she would start the letter with, “Oh, my dear beloved chairman, I miss you so much, loves and kisses.” And then she would say, “Oh, we have to be very clear in our condemnation of Tito’s revisionist clique” or something like that.

Sergey Radchenko: It was very important for Mao to actually prove to Stalin that he was a loyal supporter, that he was not a Tito who could betray him. That’s what he tried to do. Very pragmatic on his part.

Mao did declare in 1949 that he would “lean to one side,” the Soviet side, which led later to historians debating this moment, asking, “Was there a lost chance? Was there an opportunity for the Americans to stay in China and avoid 20 years of disengagement?“

Back in the ‘90s when these debates were being held, the conclusion by most historians — people like Odd Arne Westad and Chen Jian and others — was that no, there was no lost chance because Mao was so ideologically connected to Stalin. Also, he had in mind this project of Chinese communist revolution that was so important to him. He had to “clean the house before inviting guests,” i.e., kick out the Americans and then carry out these revolutionary transformations.

I am not 100% sure about this because I don’t think anything is definite until it actually happens. My own view is that perhaps if the Americans had a different approach to Communist China at that time, there would’ve been an opportunity to establish relations.

Stalin and Mao coordinated their policy on this question. The expectation on the part of the Chinese communists would be that the Americans would derecognize Taiwan effectively, or derecognize the Kuomintang government and extend recognition to the communist government. That was the key issue for them. Beyond that, it’s not like they were absolutely determined to break diplomatic relations with the United States.

Jordan Schneider: This older brother, younger brother, father-son dynamic between Stalin and Mao plays out wonderfully the way you talk about the face-to-face meetings they have. Mao gets really disrespected the first time he goes to Moscow. Stalin makes him wait for a while and basically says no to everything he asks. Slowly but surely over the course of Stalin and even more dramatically with Khrushchev, the national power balance as well as the revolutionary legitimacy (i.e. who is leader of the communists) shifts over time.

Let’s maybe use that as a way to start talking about the Korean War. How did their relationship and this Chinese-Soviet dynamic play out in 1950, which brought about that horrific conflict?

Sergey Radchenko: The Korean War has been studied by so many people. We have had a lot of documents that have been studied closely by historians since the 1990s on the Chinese side, the Russian side, and even on the North Korean side. This is really a well-studied conflict.

I wondered if I’d have anything new to say about this. In a massive book like this, I had to say something about the Korean War. I was actually quite lucky in the sense that I did find some new evidence. It wasn’t a smoking gun type evidence, but it came close. I’ll explain the nature of the evidence.

Basically, the story goes like this. Kim Il-sung in North Korea wanted to reunify the country and kept asking Stalin for permission, saying, “Comrade Stalin, the moment we cross over the 38th parallel, there will be revolution in South Korea. Everything will turn out just fine. It’ll be very quick.” Stalin would refuse him permission to do that time and again. The reason for that is pretty obvious — Stalin was worried about American intervention. He was a very cautious individual in this particular instance.

Then he changed his mind. It’s not exactly clear when he changed his mind, but we know that in late January 1950, Kim Il-sung was at a reception. Kim was a little bit drunk, went up to Shtykov, who was the Soviet representative in Pyongyang, and said, “We want to reunify South Korea.” Shtykov reported that to Stalin. Stalin sent a cable back to Shtykov for informing Kim Il-sung that “this matter requires preparation.” That is already a yellow light, not a red light.

Subscribe now

Later, Kim Il-sung went to Moscow, and they effectively agreed that the invasion would start. He changes his mind around late January 1950, or at least he tells Kim Il-sung around that time.

Various theories have been advanced. The most obvious is that at that point, there’s already going to be an alliance with China, so as a worst-case scenario, the Chinese could fight this war if Kim makes a mistake, or if the Americans intervene. But I think if Stalin thought that the Americans would intervene, he would never have authorized Kim Il-sung to do that.

The question is, why does Stalin change his mind from thinking that the Americans might intervene to thinking that they will not intervene? That is where it becomes complicated.

First of all, we have Dean Acheson’s remarks in the press conference, which are straightforward, where he says, “America has a defensive perimeter, which does not include Korea.” That is probably the most misguided statement ever made by an American foreign policymaker. That, in retrospect, was a very bad idea.

Beyond that, there’s been speculation by historians that Stalin had spies who had access to debates at the center of American power in Washington. The problem with that is, of course, we didn’t have any evidence. We can just say, “Stalin has spies everywhere. That’s how he knew about certain things like NSC-48, for example.“

What I found was very interesting. I was reading a discussion between Mao Zedong and Anastas Mikoyan in 1956 during the 8th Party Congress in Beijing. Mikoyan comes to Beijing in September 1956. They’re having discussions. Stalin is dead for more than three years, right? Suddenly the discussion turns to North Korea, and Mao says, “Why did you agree to let Kim Il-sung cross over and start the war?“

When I saw this, I thought, “That’s a moment right there. That’s very interesting.” Even Mao himself did not know what was going on. Stalin did not inform him.

Then I see Anastas Mikoyan’s response, which is, “Our intelligence intercepted cables by the Americans that said that they would not intervene in the conflict.” That is super interesting because you get that one little piece, one acknowledgement, one little piece of information that you can weave into the narrative and say, “The role of intelligence was very, very important.” This is basically from the horse’s mouth.

Mikoyan says that three years after Stalin died, “Actually, the reason we did this was because we misinterpreted the intelligence that we intercepted.” That is where it all goes wrong.

The nature of the intercept, I couldn’t find that in the archives. By the way, this document of Mao Zedong’s conversation with Mikoyan is actually from the Chinese archives, not the Russian archives. When I went back to the Russian archives, I couldn’t really find anything about that. It’s probably still somewhere locked away in the KGB vaults.

Jon Sine: My thinking is this is one of the places where there might be a lesson for today that screams at you most clearly. Do you think Joe Biden was playing 4D chess, learning from Dean Acheson, when he kept making these mistaken statements about Taiwan that his staff would later clarify by saying, “That’s not really what he’s saying,” while he maintained, “We’re going to protect Taiwan”?

Sergey Radchenko: That’s a whole different conversation regarding strategic ambiguity versus strategic certainty and which approach is better. The discussion has typically gone like this — If Biden remains unclear or maintains ambiguity about what the United States would do, that might potentially trigger Chinese intervention because they would underestimate American resolve to defend Taiwan. If, on the other hand, he stated very straightforwardly, “We will actually defend Taiwan,” then this would tempt the Chinese to test American resolve by invading Taiwan.

You can twist this argument in many ways. The bottom line is that you have internal debates and external communication between leaders. You have statements that can be misinterpreted and often are. This is how we slide into conflicts and wars — sometimes by misinterpreting what the other side wants us to do, or will do, or will not do, by underestimating the other side’s resolve.

The best example from our present situation is that Putin underestimated the resolve of Western powers, particularly the United States and Europeans, to help Ukraine. He underestimated the extent of their commitment, although that commitment has now started to fade away in some instances. Certainly in February 2022, he did not anticipate that level of commitment. Why? Because Crimea was a different story and nothing meaningful happened after Crimea, so he learned from that experience. It turns out the response this time was very different.

This parallels the early Cold War. There would be one type of strategy, and then the Soviets would do something completely reckless, like initiating the Korean War, which would reinforce the thinking in the United States that the Soviets must be confronted and pushed back against, even in a place like Korea. From Korea not mattering to anyone and nobody being able to find it on any map, suddenly it becomes the center of American attention, with people going there to die on behalf of the free world. Who would have thought? That’s a change of attitude, and it happens over weeks and months.

Jordan Schneider: This is one of the big lessons I took away from your book — the leaders are playing this weird game. Dean Acheson said it and probably believed it. Then when it happened, the politics changed, and everyone decided that it’s one thing to project it out, but when you’re faced with the reality, the news, the headlines, and the threat to prestige that all these people weigh so heavily, then perspectives change.

It’s also remarkable how you illustrate, by delving deeply into these leaders and their profiles, that they often don’t even know what the plan is. You see them winging it frequently. So even the idea that you can try to guess what your adversary is doing and play 3D chess when they themselves or even their lieutenant is thinking, “I don’t know what we’ll do next” — it’s wild to internalize that there are so many unknowns where one path could lead to the end of humanity.

Sergey Radchenko: Exactly. That’s the weird thing about Cold War history. Things could have taken terrible turns. They could have also turned out better than they ultimately did. We did have terrible crises during the Cold War, and at those particular moments, we should be grateful that leaders, despite all of their confusion and inability to understand what the other side was thinking — or maybe because of it — decided to de-escalate and pull themselves away from the brink. That is something that should be acknowledged.

Jon Sine: One of my lessons, and this might be the bridge from Stalin to Khrushchev, is how much these prestige battles mattered. It makes me think much of the Cold War was really just a sad, pointless endeavor of people jockeying over imaginary status points and where they sit in various social groupings. They’re getting pulled into games by third-party players while imagining it’s part of a broader struggle they have defined for themselves among their own group.

The Korean War is an interesting point because this is where America comes to push back, and what ends up happening is we now have a divided Korea. Fast-forward to today, and it seems like one of the most worthwhile interventions from the Cold War. But then you look to other cases, Vietnam being an example of an atrocious intervention. I’m wondering how you think about that — when America decides to intervene and when it becomes a really costly and pointlessly costly endeavor versus something that was worthwhile?

Sergey Radchenko: That’s a difficult question. It entails a moral judgment, and it’s difficult to make because, for example, the Korean War could have ended up as a nuclear war. Didn’t Douglas MacArthur want to use nuclear weapons, or at least threaten to do so? It could have happened but didn’t, and so if it did and we ended up having a nuclear catastrophe in Northeast Asia, would we be better off “dead rather than red“? It’s difficult to say.

In the end, it kind of worked out despite millions of casualties. It was a terrible, destructive war with so many deaths, but we can say in retrospect, “Look, the Americans stood up to defend freedom” — not quite democracy, frankly, because Syngman Rhee’s regime was not exactly democratic, but years and decades later it all worked out. Was it worth it? Probably.

By the same token, you might even say the same thing about Vietnam. One of the things I discovered while looking at Soviet documents — and the Soviets had remarkable access to the Vietnamese Politburo. Somehow they had spies there because in the Soviet archives you have speeches of Vietnamese Politburo members. So internal documents and all sorts of things from Vietnam that you wouldn’t expect. The Soviets clearly knew a lot more about what was going on.

You read these Vietnamese documents and recognize that actually Eisenhower’s domino theory was not wrong in the sense that that’s exactly what the Vietnamese wanted to do. They wanted to reunify their country and then promote revolutions. They obviously wanted to control revolutionary movements in Laos and Cambodia, but in the late 1960s they were discussing starting a communist revolution in Thailand, and I don’t know why that particular project didn’t work out for them.

When you read something like this, you think, “Wait a second, this is the domino theory. That’s what they’re talking about.” The Americans got involved and that was very tragic. You can turn to almost any historian of the Vietnam War and they’ll tell you, “This was a terrible mistake.” But you could also raise this counterfactual: “What if the Americans did not get involved and how would it have worked out for Southeast Asia?” Which is, by the way, very nice and prosperous now. All of that is to say that these are hard judgments. Very hard judgments.

Jon Sine: When reading your book, I was thinking that there is a fine line between describing these players battling for prestige and simply narrating that one side actually believed in a domino theory while the other side correctly presumed there was one. Where do you draw the line between objectively analyzing that each side is engaged in this game? Is there a place to step out and say, “You can analyze this correctly but you don’t necessarily need to engage in the game“? I guess it goes beyond the realm of a historian, but it was something I was trying to think about for lessons today. We’re not in any sort of domino theory situation now, but...

Sergey Radchenko: Exactly right. You might ask, “The domino theory — Vietnam was interested in taking over Southeast Asia. So what if they did? Would American global interests still be deeply undermined or fatally eroded somehow?” That is where you have to make the strategic judgment about what’s important and what is not important for America, and what’s worth paying the price for.

We all have to agree that having communists take over Southeast Asia was perhaps not the best idea for freedom and democracy, but is it worth sacrificing all those lives and basically getting America stuck in this quagmire far away from its shores? What would have happened if the Vietnamese were victorious earlier? Would that not have worked out differently?

Perhaps the Vietnamese had so many problems that their hopes to dominate Southeast Asia would have run aground anyway for internal reasons, also because the Chinese would be upset with that as well — as what ultimately happened in the 1970s with the Sino-Vietnamese conflict. Maybe the Americans should not have gotten involved precisely because this was far away from their core interest.

But what about Korea, then? You might come back to Korea and say, “That is also far away from America. Should the Americans have gotten involved, or should they have just followed Dean Acheson’s advice and stuck to their defensive perimeter?” It’s all about strategic judgments.

This matters today as well. Let’s say the Russians overrun Ukraine and establish a puppet regime or annex Ukraine. Would America’s global interests or core national interests be fatally undermined? You can make the argument both ways. You could say yes, because American credibility would be at stake — others will say, “America is a paper tiger, they cannot be trusted, they cannot even defend Ukraine, so how will they defend Taiwan?” and other things start falling apart.

Or you might say, “Actually, it doesn’t really matter.” Ukraine is far away from America’s core interests. America’s core adversary is China, that’s where it should be focused. Some people have been arguing exactly that, and so Ukraine is a distraction.

As a historian, what can I say about this, apart from noting this was always a problem and always part of the discussion? The best we can do really is see how historically things have worked out. Sometimes they have worked out well, and sometimes they have not. But there was always a price to pay. That is a very unsatisfying answer from a historian.

Jon Sine: It’s not unsatisfactory, but I would also draw in the domestic aspect as well. You go to pains to show how much it undermined the Soviets to spend so much of their effort and attention in these Third World competitions that ultimately were of little significance. We still feel the effects in the United States from our voyage into Vietnam in terms of undermining faith in the government. I think that’s another aspect that needs to be considered.

Sergey Radchenko: That’s right. The Soviets were ultimately the losers of the war in Vietnam. You’d think the Americans were the ones who lost out, but as I argue in the book, it’s actually the Soviets who lost out. They thought they won, but what they got was an ally they constantly had to pay for, and that was a gigantic drain on Soviet resources. They just kept paying for the Vietnamese, and the Vietnamese said, “Sorry, you have to forgive all those loans because we cannot do anything.” That actually contributed to Soviet over-extension. Who’s the real winner then? That’s a big question.

Jordan Schneider: Let’s reel it back to the Stalin-Khrushchev transition. I love how you painted how the fight to kick out Beria ended up creating a fork in the road where Germany could have ended up in a different place, but everyone kicking out Beria ended up making that path toxic. What should people reflect on when understanding that power transition?

Sergey Radchenko: That is really an interesting period. We still don’t know all that much about it. We know enough from the memoirs and recollections of people like Nikita Khrushchev about what actually happened. We have the film “The Death of Stalin.” That’s where most students learn about Beria’s demise from.

The problem for that period — late Stalin up to maybe ’53, ’54 — is that we don’t have sufficient documentation. It’s not that it’s unavailable; I think it just doesn’t exist. Things like Politburo records of what they were discussing in the Politburo — the Presidium, as it was called — we don’t have that. There are some gaps in the record.

There is an interesting story about Germany in all of that. Around this time, the Soviets get the idea that East Germany should perhaps not try to build communism. It’s not working out well. They try to communicate that to the East German leaders, who are very hard-line, committed orthodox communists. These leaders keep pushing policies which result in a full-fledged uprising in Berlin in June 1953, which has to be suppressed with tanks, tragically.

The Soviets get this idea of maybe having a confederation or something similar. What they actually say is they could have a bourgeois Germany — a bourgeois Germany that is not aligned to the West necessarily, but is also not a communist country.

Now, the question is who’s pushing this? That becomes difficult to answer because you have people like Beria who seem to be in favor of this. Stalin’s hideous henchman comes across almost as a liberal, letting people out of the Gulag, saying to the Germans, “Look, you don’t need to build communism.” Beria is one. Malenkov seems to be supporting this as well, so it’s very difficult to say that it’s just Beria, because Malenkov also entertained similar ideas.

What I argue in the book is that Beria is arrested, and in the process where they try to implicate him or pin all kinds of terrible things on him, they say, “Well, he’s a Western spy, he’s a British spy. He was trying to undermine communism in Eastern Germany.” This is where internal politics and external foreign policy overlap. At this point, Malenkov could have said, “I also was in favor of the same policy,” but notably, he doesn’t. That’s how it works.

Beria is executed, and East Germany remains basically communist. Later, West Germany becomes admitted to NATO, and at that point, this idea of the neutral Germany that is bourgeois but not aligned to the West falls by the wayside. The Soviets then commit to maintaining the regime in the GDR, which is very costly. It’s a very costly proposition and difficult to do because people keep running away, which is one of the reasons that ultimately they have to build a wall.

What I’m trying to do with this book is show that you can actually have a variety of policies. At different times, like in ’53, they thought, “What about this approach?” It didn’t work out for different reasons, and then a different policy emerged.

Jon Sine: Your research basically finds the same thing as Joseph Torigian, that especially in those moments of transition of leaders, it’s really not about policy discussions at all that determine what’s going on. It’s internal struggling, what he calls a knife fight with weird rules.

Sergey Radchenko: That’s right. Joseph’s book is fantastic on this. In fact, we work from some of the same materials.

That’s what you get. Once you go into the real depth with these records, what you see is really just people above all. You see people positioning for power, for influence, backstabbing each other, and so on. Big questions of policy get drawn into this in weird ways, but sometimes in ways that you would not expect necessarily.

This is something interesting because I don’t know how you would explain that from a theoretical perspective. It doesn’t really work out. Acquaintance with archival materials like Joseph Terian and myself have been able to do is very useful.

Jordan Schneider: It’s like the theory is just schoolyard politics, which is kind of incredible. You have these moments in this book of big statesman-like times where Khrushchev feels really good when he meets with Eisenhower, and Brezhnev feels like he can do business with Nixon. But those are the times where your grasp on power is most secure.

When it’s not secure domestically or internationally, then it becomes about the unseemly parts of human nature that are evolutionarily the things you would see 10,000 years ago, not what you would necessarily expect humanity to be able to pull off and aspire to in the 20th century.

Sergey Radchenko: Jordan, I would actually argue against this. People don’t evolve so fast. In some ways, we do behave like monkeys in the forest because that’s evolutionary behavior, the struggle for “I’m the king of the hill” and “Everybody defer to me.”

If we see that with monkeys, why wouldn’t we expect people to have the same tendencies? Fundamentally, those psychological motivators of human behavior are really important. This is where over-reliance on some grand narratives about ideology or realpolitik starts falling apart. Yes, of course, all of those things matter. But in the end, we are monkeys in the forest.

You see these behaviors historically repeating themselves again and again for centuries. There must be something fundamentally human to this kind of behavior — this struggle for power, struggle for recognition, the desire for legitimacy, the desire for prestige. All of those things seem to be deeply rooted in human psyche.

Jon Sine: Once upon a time, I wrote an essay that I was going to use to get into my PhD program on reorienting how we think about international relations around these Evolutionary Psychology ideas of status and prestige. Needless to say, I’m a fan of your narrative.

But you can also root it in something else — philosophy, or at least great works of art, which you do. You quote from Dostoevsky in describing, I think it really applies most of all to Khrushchev above anybody else, “Am I a trembling creature, or do I have the right?” That just jumps out at you as something correct when you look at Khrushchev.

Maybe let’s fast-forward to the Stalin speech because in a way, he’s not so much a trembling creature when he’s up there thoroughly denouncing Stalin. This will be the thing that he’s known for. You look at some of the Marxist historians who say after reading “China Today,” this is where they say it all started to go wrong. Tell us about that.

Sergey Radchenko: Khrushchev decides to denounce Stalin — this is known as the Secret Speech — denounced Stalin for his various crimes, for repressions, for the way that he conducted the Second World War. The party is absolutely shocked. When this verdict is delivered, the party faithful are there, and they cannot believe it. Although for a time, things were kind of going in that direction anyway, so you had silent de-Stalinization already shortly after Stalin’s death. But in February 1956, it becomes official. It’s not publicly announced because it’s a Secret Speech, but it’s circulated inside the party.

It’s a very interesting question as to why Khrushchev does that. When you read about this, you almost feel sympathy for Khrushchev. You think, “Well, here’s a good guy.” But then you also have to remember that he did all kinds of other things, like ordering the Soviet invasion of Hungary, which serves as a counterbalance to this very positive view of Khrushchev.

Why did he do that? Was it for political reasons in order to consolidate his power, or did he really think that Stalin was a terrible person? I think probably both. It helped him consolidate his power, but there’s no reason to think that he was lying when he was saying that Stalin was a criminal.

Now, some people said, “Well, where were you? You were also participating in all this stuff. You benefited from this because your career overlapped with absolutely horrendous purges. You built your career on the bones and blood of your predecessors.” There’s something to that as well.

But when all is said and done, you look at this Secret Speech of February 1956, you cannot help but feel that this was the right thing to do. You cannot help but feel like this is one of the good moments for Khrushchev. There are some others, like the way he decided to back out from the Cuban Missile Crisis, where you feel, “He did the right thing.” That’s how I felt when I was looking at it.

But the Chinese were not very happy. Mao Zedong was saying, “Wait a second. You did not consult us.” This is part of the problem. Mao Zedong himself did not like Stalin, but he certainly did not like being surprised like that by Khrushchev. He said that, “Stalin was a great sword, and Khrushchev had discarded the sword, and now our enemies will seize it and try to kill us,” or as he would also say, “Stalin was a great rock, and Khrushchev lifted it only to drop it on his own feet.“

Mao predicted problems for the communist movement, and of course he was right because problems broke out in places like Poland that summer of ’56. Then ultimately, in the fall of 1956, we had the situation in Hungary that just went off the rails very quickly. All of that is connected to de-Stalinization.

In my book, I talk at length about it. I don’t focus so much on the domestic impact of de-Stalinization. There are other people who have written fantastic books about it, but what I look at is how it impacted countries like Poland, Hungary, North Korea, and then I also talk about China’s role in all of that.

This is where Mao Zedong, for once, feels like he’s the great strategist who’s determining the fate of global communism because he sends a delegation to Moscow led by Liu Shaoqi. Liu Shaoqi says in conversations with Khrushchev, “What you’re doing to Poland, trying to pressure Poland — this is called great power chauvinism. You should stop doing that. On the other hand, in Hungary, that is a kind of revolutionary rebellion, so you should basically suppress that.“

e know there is no Soviet intervention in Poland, and there is Soviet intervention in Hungary — by the way, for reasons that had nothing to do with Liu Shaoqi’s advice. It was for Khrushchev’s own reasons that simply overlapped with what Liu Shaoqi was saying. But the Chinese conclude that this is because of their advice. So they have finally got to a position where they’re determining the general line of the communist world movement, and that, I think, is consequential for the Sino-Soviet split, for all the things that happened afterwards.

Jordan Schneider: Two quotes, some of my favorite from the book, where you found in the Politburo that Mao talked about Khrushchev’s Stalin speech as if Khrushchev had broken the incantation of the golden hoop. Of course, the brocade that the Monkey King wore and now everything’s going to go crazy and Mao’s going to be able to be free. The superstitions of what can and can’t be done to challenge the Soviet Union are now blown wide open.

I also love the detour that you had of these incredible sources of the pro-Stalinist folks in Denmark and Norway. The story’s been told a lot in the US, but at some point, I think it was the Danish guy who was so pissed off — when he was getting the readout, he was saying, “Khrushchev, what were you doing? You should’ve arrested this man, Stalin. What a monster.“

The ability for some folks to see through the cognitive dissonance of it and say, “Okay, we should just give up. This is a completely bankrupt thing.” And then other folks in different situations are more or less bought into the system, or their incentives are different. It’s funny that the players with the least actual political power, the intellectuals in the West, were often the ones who clicked most quickly to the fact that a communist system able to bring about someone like Stalin is just a bankrupt endeavor on its face.

Sergey Radchenko: It took a while actually, and obviously ‘56 had a dramatic impact on the world communist movement. In many ways I would say 1956 was the beginning of the end for communism.

It was very clear. If you needed to use tanks to suppress a country or to promote your vision of economic development, you’ve got a problem. That is where you have a collapse of party memberships in communist parties across Western Europe. So many things can be linked back to 1956.

It was such an eye-opening moment for Western intellectuals, many of whom just could not believe that Stalin could be such a hideous figure. But that’s a consistent problem. I was struck reading Martin Sherwin and Kai Bird’s biography of Oppenheimer, American Prometheus, where they discuss how nuclear scientists working out in California were just absolutely blind to the brutalities of the Soviet system. You think, “How can smart people who are obviously geniuses in physics be so uncritical as to think that there’s this worker’s paradise and everything is going wonderfully well?“

For Oppenheimer himself, 1939 seems to have been the turning point — the Soviet-Nazi pact. After that, he finally started thinking that perhaps this Stalinist paradise was not so wonderful after all. But Soviet intellectuals continued to believe until 1956.

In 1956 you have the double shock. First, Khrushchev himself says, “Look, Stalin is not the father of humankind, not this great genius or demigod that we held him for, but actually a brutal criminal who repressed innocent people and was just a terrible person.” Then you’ve got the Soviet intervention in Hungary which nails the final nail in the coffin of the Soviet project for many Western intellectuals. That’s why you have this chaos in Western European Communist Parties from CPUSA to the Norwegians, the Danes, Great Britain, and so on.

Jon Sine: You just said in 1956 you could see the beginning of the end of communism. When Deng Xiaoping was writing at the Sixth Party Plenum in 1981, rewriting the history of what happened in China, he’s very careful not to reject Mao in total, seeming to draw a lesson from 1956.

This is a very prominent lesson today among certain Marxist historians who look back at the Soviet Union and how it fell. 1956 is a crucial year, but you didn’t just root it in the Secret Speech. You pointed to tanks going into Hungary as a crucial thing. Do you agree with their analysis? Was there a way to save Stalin’s reputation? Did the Chinese keep a 70/30 Stalin type of interpretation? What are your thoughts on that?

Sergey Radchenko: That’s obviously how Mao would have liked to see things, which is why he proposed to discuss Stalin in those terms. That Stalin committed some mistakes — mainly not believing in the revolution in China and mistreating Mao. But generally speaking, he was a great Marxist-Leninist. That would have been a better approach.

Stalin was a horrible person, and saying this frankly and openly was a better thing to do. In fact, the problem with the Soviet Union and with Khrushchev’s de-Stalinization was that they did not go deep enough. They deposed Stalin but the system remained, and I would argue the system still remains. Is that a good thing? I don’t think so.

There needs to be a different approach to people like Stalin in places like Russia and to people like Mao Zedong in places like China. But of course the Chinese will say, “We are careful. We don’t want to undermine the legitimacy of the communist state. That’s why we still print Mao Zedong on our money and that’s why his portrait hangs over Tiananmen Square, because he is the founder of the Chinese state. If you say that he was a horrible criminal responsible for the deaths of 35 million people, well... that might delegitimize the Communist Party.” Are they wrong? I think they’re probably right — of course it will delegitimize the Communist Party.

Jon Sine: Nothing delegitimized the Soviet Communist Party more though than their failing economy. You mentioned not de-Stalinizing. There was the Kosygin reform which could have radically restructured, gotten rid of the ministerial planning apparatus. I feel like if they were going to go down that route, they needed to get away from the Stalinist planning system at some point. It could take you to maybe the early modernization, but they never seemed to get out from under it.

Sergey Radchenko: Khrushchev’s reforms brought about a lot of chaos in the USSR. Constant restructuring, compressing ministries, taking out whole parts of the apparatus. I say in the book that Khrushchev was like this guy who felt the communist project was a really great project and you just have to tinker with it here and there. Take out a pin here, put something there and then it would work just fine. The problem was that it fundamentally wasn’t working, so that realization took some more time to creep in. That’s already after Khrushchev.

We can see the key point, as you rightly pointed out, is that the Soviet project was not delivering for the Soviet people. It was not delivering enough because the Soviets set themselves up not just with the claim that they would improve the standard of living — because actually the standard of living was improving — but that they would outstrip the United States.

They claimed they would be better. We have the kitchen debate, that famous encounter between Nikita Khrushchev and Richard Nixon, of which we have a video recording. I was so happy to find the Soviet version of it in the Russian archives, which was hilarious to see what was not captured on tape. They go around and Nixon shows this and that and says, “Well, look, this is an American house.” And Khrushchev responds, “No, this is made of wood. This is rubbish. We now make stuff of plastic. This is so much better.“

You have this competition with the United States. The idea was that it would deliver for the Soviet people and it simply wasn’t doing that. In some areas, there were breakthroughs, but by and large, the Soviets had meat riots. Think of the 1962 riot in Novocherkassk where the KGB had to be deployed — the military police — to break up a protest with casualties. In 1963, the Soviets were spending their gold reserves to import grain for the first time. That is already where you’ve got some serious warning signals coming.

That’s why they launched the Kosygin reforms. But then that kind of goes off the rails. At the same time as all these problems in the economy multiply, they also discover oil and gas in much larger quantities in western Siberia and they think, “Okay, that will save us.” To a certain extent it actually helped them for another decade or so but ultimately, of course, the whole thing fell apart.

Jordan Schneider: I want to back up because I don’t want to move too quickly past the second half of the 1950s. You actually have this incredible moment of optimism with Khrushchev, where you have Sputnik, of course, and then this techno-utopian vision that Khrushchev has. He believed that in the future, the Soviets would have a bright future, a fairytale-like socialist abundance with people having so much to eat that they would, quote, “be careful not to overwork their stomachs” as robots did all the work. People would just hang out, only working for an hour or two per day.

Share

You make this fascinating connection between what Sputnik does and how it changes Khrushchev’s psychology, leading to some of the most dangerous moments in human history. Could you explain that connection? Khrushchev has this big technological breakthrough, decides he’s going to catch up and surpass the US in 15 years, and what does that lead to in terms of the risk of war in various hotspots around the world?

Sergey Radchenko: As you say, Jordan, these are interrelated processes. The Soviet defense program, the missile program, and the Sputnik program were obviously intricately related. In fact, Sputnik was launched on an ICBM.

What was happening in terms of defense priorities? The Soviets got their bomb in 1949, still under Stalin. There was a modest buildup of nuclear bombs in 1950, but then they really put a lot of effort into the development of missile technology. By around 1955, there were some serious breakthroughs.

First, you had a proper Soviet thermonuclear test — not the one in ’53, which wasn’t the later design — but the proper thermonuclear test in 1955. That made Soviet nuclear weapons much more powerful, giving Khrushchev this sense of great power: “Look, we’ve got these massive weapons that can destroy whole cities.” That was an empowering feeling.

Then you had breakthroughs in missile technology. By 1956-57, missiles increased their range. In 1957, the Soviets conducted the first test of an ICBM — intercontinental ballistic missile. Khrushchev realized this basically meant the Soviet Union was invincible as a power. It could reach the United States. It could destroy the United States.

For the first time, he developed this feeling — later shared by all Soviet leaders and even today by Russian leaders — that Russia was a mighty power precisely because it could destroy the world. That’s where its claim to greatness came from. It could destroy the world, and nobody could mess with it. For the first time, Russia could be safe because nobody would dare to invade a nuclear power.

Khrushchev understood that, which is why he began massive cuts in the armed forces and shut down projects like the battleship program that was Stalin’s prestige project. Why? Because he believed in nuclear weapons. He felt they gave the Soviets might and a new voice in international politics.

Although he didn’t plan to use nuclear weapons, he used them for blackmail and discovered the usefulness of that already in 1956. During the Suez Crisis, Khrushchev threatened to destroy Great Britain and France. Those powers ultimately backed out from Suez for reasons that had little to do with Khrushchev’s threats — mainly American pressure on the British. But that’s not what Khrushchev thought. He believed, “I was so successful in doing that. Now, let’s play this card again and again.“

That, Jordan, got him into dangerous situations, ultimately leading to the Cuban Missile Crisis, because he felt that the Americans or anybody else would have to retreat when faced with Soviet might. It gave him this trump card to play.

On the Sputnik side, you had the launch on October 4th, 1957. That’s why we talk about the “Sputnik moment” — a term we now use to describe breakthroughs by America’s adversaries. The original Sputnik moment was a great breakthrough which symbolically showed that the Soviet Union could outcompete the United States in a key area of science and technology.

Unfortunately for the Soviets, they weren’t able to capitalize on this. Though they could capitalize on Sputnik itself, they couldn’t provide sufficient investments in scientific and technological terms into their own economy where their economy would compete with the United States. There were structural problems with the economy as well, and it simply wasn’t competitive.

The Sputnik moment faded away fairly quickly, but what remained was the Soviet nuclear threat, which was real and which became ever more serious as time proceeded.

Jon Sine: The Cuban missile crisis remains one of the most hotly debated historical occurrences. From the US perspective, it’s incredibly important. The Sputnik moment you mentioned factors into that. Another element is the menage a trios between China, the US, and Soviet Union — a constellation of powers competing for greatness.

The Soviets seemed to have a trickier game to play because they were balancing their great power aspirations between themselves and the US while also competing for leadership over the communist revolutionary world with China. They were attempting this double dance, which came to a very dangerous head in the Cuban missile crisis, with that process playing into Khrushchev’s decision to try to win over Fidel Castro.

How does bringing in the Chinese and the competition for prestige dynamic change how we think about the Cuban Missile Crisis?

Sergey Radchenko: That’s a great question. The Cuban Missile Crisis is a very overstudied episode of the Cold War. What I contribute in my analysis is precisely this Chinese angle, which is fascinating. We haven’t really been talking enough about that.

If you think about the Cuban Missile Crisis, what questions are we still trying to answer? The first question is why Khrushchev sent missiles to Cuba in the first place. There are different competing theories about it.

The prevailing American theory for years was that it was strategically necessary for the Soviet Union. Khrushchev realized that his ICBMs — for all the Sputniks of the world — weren’t particularly good. They weren’t accurate; they were problematic. So he had to put intermediate-range ballistic missiles in Cuba to target America more accurately and reliably.

That has been the traditional explanation for Khrushchev’s decision. Looking at the Soviet and Russian records, there isn’t a single moment where Khrushchev actually raises this as an explanation — not once. I didn’t find a single piece of evidence to support that. It’s more about political scientists trying to figure out why Khrushchev would do this than Khrushchev actually explaining his reasoning.

In the 1990s, Sergey Mikoyan, who was the son of Anastas Mikoyan whom I previously mentioned, proposed a different theory. He suggested it wasn’t about addressing strategic problems at all. It was about saving Cuba because of the Bay of Pigs and the fear that the Americans would take over Cuba — a realistic fear, let’s be honest. Khrushchev wanted to save Cuba.

Is there evidence for this? If you look at the Russian records, including the Presidium discussions, you actually find evidence for it. Khrushchev says on at least two occasions, “We just wanted to save Cuba from American invasion.” This is very interesting because it supports this particular theory.

In my analysis, I ask why he was so obsessed with Cuba. What was that about? I argue that it was because he was in competition with China, where the Chinese were saying, “The Soviets have betrayed revolution. They’re weak. They’re not standing up to American imperialists.” Under those circumstances, losing Cuba — which was basically a socialist country run by a real revolutionary, a wannabe communist — would be unacceptable to Khrushchev.

He was really worried about the Chinese angle. There’s substantial evidence for this because throughout and after the crisis, he was very sensitive to Chinese criticism. He kept telling the Cubans, “Don’t believe the Chinese. We don’t want to sell you out. We are actually the only ones helping you,” and “Look what the Chinese haven’t done for you anyway.“

The Chinese had made their own inroads into Cuba because they were talking to Che Guevara, who had a very close relationship with the Chinese ambassador. This presented a big challenge for Khrushchev.

Bringing us back to the angle about prestige and the Soviet desire for recognition and greatness, there’s another interesting piece of evidence that connects to my theory. Khrushchev wanted to be treated as an equal to the United States. When he was making the decision to send missiles, it was in the context of a discussion about American missiles in Turkey.

From his position, if the Americans had nuclear Jupiter missiles located in Turkey, why couldn’t the Soviets have missiles in Cuba? That wasn’t fair. Khrushchev commented, “We will effectively give the Americans a little of their own medicine.“

If you psychoanalyze this phrase, what does it mean? Does it mean he was really concerned about strategic problems or reliably hitting Washington? He never wanted to use nuclear weapons to begin with. No, it was more about equality — why were the Americans allowed to have missiles near Soviet borders, but the Soviets weren’t allowed to do the same? This brings in the question of equality, status, and greatness: “We are on par with the United States. We can destroy them, therefore they should not expect special treatment.“

That’s how I describe the opening phase of the crisis. The relevant chapter also discusses how the crisis itself unfolded. I was fortunate to have access to remarkable materials that add to our understanding of how Khrushchev ultimately decided to back out from the situation.

The key piece of evidence is that Khrushchev really thought Castro was going off the rails at one point, especially after Castro proposed to nuke the United States in a first strike. Castro later denied this, saying he never meant anything like that, but that’s how Khrushchev understood Castro at the time. Khrushchev was shocked, thinking, “What is he talking about? This guy’s crazy.” At this point, Khrushchev became frightened.

One of the things you find in the Russian archives is how early Khrushchev decided to back out. Kennedy gave the quarantine speech on October 22nd, 1962, and then Khrushchev dictated a letter to Kennedy on October 25th, already effectively backing out of the crisis. It only really lasted for three days.

Analyzing Khrushchev’s language and concerns is fascinating. All of that material is now available in Moscow for researchers — though I don’t recommend going there.

Jordan Schneider: The dynamic you see in the Cuban Missile Crisis is one that plays out over many crises, where we have lots of influential actors say, “Let’s just send some nukes. It’s the path of least resistance. It’ll solve our problem.” We mentioned this earlier in the Korean War context, and it happened here as well.

But you do see this other very human reaction where the person who will actually make the decision often seems to be the only one truly weighing what the second-order consequences mean, beyond just the narrow military, theater-level advantage you might get by being the first to pull the trigger.

Sergey Radchenko: That’s precisely the point. When you have the responsibility, you must make that final decision. The Soviets had extensive plans for using nuclear weapons in Europe to wage nuclear war. They practiced for it and carried out military exercises with nuclear weapons.

The military planned for it, but they didn’t decide on this matter. The leadership had to make that decision — Khrushchev ultimately, because the responsibility was in his hands. When he considered what he could potentially be authorizing, he was deeply concerned.

This is where personal psychology becomes so important. I discuss a fascinating episode in the book about how Khrushchev contextualized his decision on the Berlin Crisis by recalling his experiences from the Second World War and the story of Nikolai Voshchugov.

Back in 1941, after the Germans invaded the USSR, Voshchugov, one of the commanders on the Soviet side, came to Khrushchev and said, “I’ve lost my tank army.” Khrushchev asked, “What do you propose to do?” Voshchugov pulled out his handgun and shot himself in front of Khrushchev.

What’s interesting is that Khrushchev recounted this story when discussing his decision — making during the Berlin Crisis. Why is that important? Because Khrushchev realized that human rationality had limits. Why would this man take his own life like that? Was it rational? No. Similarly, starting a nuclear war isn’t rational, but people might still do it. This realization added to his reservations.

Despite all the nuclear plans, exercises, training, and available weapons, Khrushchev did not plan to fight a nuclear war. He reasonably concluded that he didn’t want to engage in nuclear warfare.

The same applies to the American side. Consider President Eisenhower’s approach to the second Taiwan Strait Crisis in 1958, when the Joint Chiefs of Staff presented the nuclear option, saying, “Use these weapons.” Eisenhower refused to consider it because he was the decider, as George Bush would later say. They were the deciders who had to make these choices, and the weight of responsibility was immense.

Jon Sine: What’s interesting is that Mao at times seemed unconcerned about nuclear weapons. I believe he once said — perhaps to Khrushchev — “If they used a nuclear weapon on China, they could take out half the population but we’d still have 300 million Chinese,” or something to that effect.

This brings me to the Sino-Soviet split, which is perhaps the most important aspect of the Cold War. There are many contenders for that title, but it really comes through in your book and is one of the most interesting aspects. I’d like to ask about its origins. You mentioned potential ideological debates but don’t find them convincing. The split started under Khrushchev but reached its most critical point — a literal war — under Brezhnev. Could you talk us through that?

Sergey Radchenko: This is another topic historians have debated for years, myself included. I wrote another book on the subject called Two Suns in the Heavens where I presented a particular view on why the Sino-Soviet split happened. We also have scholars like Lorenz Lüthi from McGill University, a good friend of mine, who wrote a different book arguing for the importance of ideology.

This debate goes back to the 1960s. In line with my general skepticism about ideology, I argue that ideology wasn’t really what mattered most, which is counterintuitive. There was extensive propaganda and ideological rhetoric on both sides — the Chinese accusing the Soviets of revisionism (revising Marxist-Leninism), while the Soviets called the Chinese dogmatic.

So much material was produced in various proclamations, statements, and letters that you might think ideology was obviously important, but I don’t believe that’s what truly drove the relationship or caused the split. Fundamentally, the Chinese didn’t want to be underdogs in this alliance, particularly Mao Zedong. He believed he deserved a better position and wanted to lead the alliance forward in ways he felt were correct. He didn’t want to defer to people like Khrushchev, whom he didn’t consider particularly bright or insightful.

It was partly a conflict of personalities and partly a conflict over leadership. Khrushchev gave a remarkable description of the reasons for the split in a conversation with Castro, who asked him when visiting Moscow in spring 1963, “What’s going on between you and the Chinese?” Khrushchev replied, “I also don’t know what it’s about. They say they’re against world war, we’re against world war. They say they’re for revolution, we’re for revolution, so none of that makes sense.” Then he advanced his theory: “Actually, the Chinese want to play the first fiddle.“

Then he completely goes off the rails in one of the most fascinating snippets of Khrushchev I ever found. He launches into a theory about how even in a circle of friends, some are naturally smarter, with various degrees of intelligence, different colors of hair, and so on — and it becomes quite explicitly racist.

Basically, Khrushchev felt like, “Look, who are the Chinese? We are the ones who had the communist revolution. We are the ones who won the war against Nazi Germany and launched Sputnik into space. So why are the Chinese trying to claim leadership from us? It’s unreasonable. We are the natural leaders.”

From Khrushchev into the Brezhnev era, the Soviets were determined to maintain their primacy in this relationship. They also hoped to bring the Chinese back to their “proper place.” They felt the Chinese had erred and would eventually recognize their mistakes, repent, and return under Soviet leadership.

That’s one reason this conflict continued for so long. Only under Gorbachev did things change, when he finally said, “We don’t want you to be younger brothers. We’re not interested in that.” Then they rebuilt the relationship on a more equal basis.

Jon Sine: One of the key points of criticism was when Khrushchev realigned toward saying, “We’re going to have peaceful competition or peaceful coexistence,” while the Chinese were saying, “No, we should heighten tensions, especially when it comes to the US.” This becomes even more ironic 10 years later with the Mao-Nixon rapprochement.

Sergey Radchenko: That’s precisely the point. That’s why I’m somewhat skeptical of ideological explanations. The Chinese felt the Soviets weren’t revolutionary enough, and then what did they do? They invited Nixon to Beijing, and Mao said, “I like rightists.” That’s an actual quote from his conversation with Nixon. What does that make of the various ideological disagreements they had? That’s a good question.

Stay tuned for part 2!

Charles Yang is the executive director for the Center for Industrial Strategy, a bipartisan think tank focused on industrial policy. Previously, he served as an AI and Supply Chain Policy Advisor at the Department of Energy and was an ML Engineer at an AI hardware startup in San Francisco. Today, he’s here to present some excerpts from his research into how Admiral Hyman Rickover built the nuclear navy.

Strategic competition demands more than technological innovation — it requires building industrial power. The U.S. is realizing the damage done by decades of underinvestment in the nation’s industrial base, which now jeopardizes its ability to compete on the global stage. Today, the production capacity of Chinese shipyards is over 200 times that of US shipyards, and China has used its chokehold on critical mineral processing as leverage to retaliate against US sanctions.

A new bipartisan consensus is emerging around the need for industrial policy — from the passage of the CHIPS and Science Act, to the recent bipartisan introduction of the SHIPS for America Act and the Critical Minerals for the Future Act.

As Congress steps into this more active role, policymakers should learn from the successes of our past. Nearly 75 years ago, Admiral Hyman G. Rickover, “Father of the Nuclear Navy”, pioneered a bold program to develop and operationalize nuclear power in the Navy. Under his leadership, the U.S. government harnessed the power of the atom, building the world’s first nuclear-powered submarine and the world’s largest fleet of nuclear reactors for civilian power.

Lessons From the Past

Rickover spent his entire career in the Navy and is still the longest-serving naval officer in US history. He spent the first 20 years of his career as an electrical engineer, where he honed a strong technical foundation and unique management style. In 1946, he was assigned a 1-year tour of duty at the Oak Ridge site of the Manhattan Project. Rickover immediately recognized the transformative potential of nuclear technology — he spent the rest of his career building the “Nuclear Navy,” which ensured US strategic dominance of the high seas for the rest of the 20th century.

Within the span of 10 years, Rickover created an entire office dedicated to nuclear propulsion, and successfully launched the world’s first nuclear-powered submarine without cost overruns. He conclusively demonstrated the strategic importance of nuclear propulsion in a timeframe no one thought possible and helped the US beat the Soviets to nuclear propulsion for submarines by 3 years. His institutional legacy is the US Navy’s safe construction and operation of nuclear reactors.

As the US gears up for another strategic competition, Rickover’s story can offer helpful lessons for aspiring technocrats. Oftentimes, industrial policy is framed in terms of legislation, but Rickover demonstrates that industrial policy is as much about policy as it is about strong leadership.

Talent, Training, and Management

Rickover spent an inordinate amount of time focused on interviewing personnel — he made the final hiring decision for every naval officer who applied to serve on a nuclear submarine until he retired. And he was an unorthodox interviewer, screening for high agency individuals who could think on their feet — literally! To test their composure, Rickover famously made candidates sit in a chair with the front two legs shortened as he loomed over them during questioning.

One interview account:

For one interviewee who said he liked hiking, Rickover asked him if he had ever hiked the nearby “Goat Mountain”. When he said he had not, Rickover told him to bring him proof he had climbed it by tomorrow morning and he would be hired. It turns out that Goat Mountain was the peak of a structure for mountain goats in a zoo. He went to the zoo, asked a tourist to take his picture, jumped into the enclosure, and climbed to the top. He’s hired the next day!

But it didn’t end at the interview process — Rickover believed in continued technical training for his staff and in building out a talented workforce base for this new technology:

While Rickover worked to staff up quickly in the short term, he also set out to build a deep bench and a long-term pipeline of talent. He required each officer and engineer he hired to submit a self-study plan demonstrating mastery of advanced texts in metallurgy, physics, and chemistry, along with field trips to AEC facilities, totaling 854 hours of study or 16 hours per week. He also worked with MIT to develop a survey course on nuclear physics and a master's degree in nuclear engineering, with a curriculum drawn up and agreed to by Rickover, starting in June of 1949. Rickover also worked with Oak Ridge National Lab to develop a 1-year curriculum in nuclear science and technology, a program christened “Oak Ridge School of Reactor Technology (ORSORT) with the first cohort starting in March 1950. Westinghouse, GE, utilities, naval and private shipyards, and Naval Reactors all sent students to ORSORT and the program started turning out ~100 graduates a year, providing another training center to develop a nuclear industry.1 Finally, Rickover had his engineers provide training lectures to a variety of audiences, ranging from senior officials in BuShips to junior technicians, as well as to explain shipboard problems and applications to scientists at Argonne, Oak Ridge, and Westinghouse/GE.

He was also known for his unique style of management. Not only did he interview every naval officer in his office, he also maintained direct lines of communication with every nuclear sub commander and project officer on-site with contractors, giving him early awareness of every issue. The demanding oversight he extended over his technical staff under his command pushed them to have greater awareness of their own direct reports:

Rickover was also an intensely demanding and scrutinizing manager. As most writing then was done on carbon copy paper, every night Rickover would collect the “pinks” of every piece of writing from his various teams i.e. the carbon copied half, and read over them at home, including drafts. When his officers protested as to how they should be expected to keep track of everything in their purview, including drafts reports from staff below them, Rickovers responded “It’s up to you to see that I don’t know more about what’s going on in your shop than you do”. By enforcing tight lines of supervision over his officers, Rickover ensured that he maintained full visibility into each team, including the project facilities at Knolls, Bettis, and the shipyards, allowing him to catch problems early on. It also enforced a culture of direct accountability and oversight across the organization.

Rickover’s focus on hiring, training, and close project management represented his philosophical approach to how to build complex systems managed by humans.

Near the end of his career, Rickover testified to Congress after the Three Mile Island Reactor accident. He spent the vast majority of his testimony talking not about regulatory reform, but about the lack of training and inadequate culture of responsibility among the operators.

Industrial State Capacity

Rickover’s scrutinizing style of management extended to the private companies he worked with. He pioneered the practice of project officers, who lived on-site at the projects and who would report directly to him any delays or unforeseen issues, so that Rickover could escalate immediately and ensure the project remained on track.

Government contracting was, and still is today, a largely passive and administrative activity. While Rickover acknowledged that the government was the “customer” and the contractor was the one responsible for delivering, Rickover’s unique approach to program management was exercising tight oversight over the contractors. Rickover hired technical experts into his office and then sent them out as project officers to oversee the various contractor sites. There, the project officer was expected to be the active representative of the Naval Reactors Office, reporting directly to Rickover any issues with contractors and ensuring the contractor was on track to deliver the product as expected. In every sense, Rickover’s project officer was to be his eyes and ears on the ground. Rickover took great pains to ensure there was no customer capture, telling one of his project officers, “Don’t go to dinner with them. Your wives must not get friendly with their wives. You’re not even to let your dogs get friendly with their dogs…when you do that, you become one of them…you don’t represent me anymore”.

Rickover’s success in scaling industrial technology was demonstrated early on with Zirconium production. In 1949, the world had only produced a shoebox worth of purified Zirconium, but the material showed promise as a fuel cladding material due to its durability under high temperatures without blocking the emitted neutrons needed to enable fission reactions. AEC opened up a simple contract for private companies to bid to produce Zirconium, but none of the companies were able to scale up production. Rickover took over production a year later, applied his practice of close project management with the (now defunct) Bureau of Mines, and only then passed it off to industry:

But by 1949, when Rickover was looking to scale up promising fuel cladding material production, the AEC had already decided to run contracts through another AEC division. Unable to exert the centralized control over the contractors, the AEC manufacturers were slow to scale up a high-quality production process. In 1950, after a year of delay, Rickover finally received permission to have the Westinghouse Bettis site directly manufacture Zirconium metal and worked with the Bureau of Mines (BuMines) to purify the Zirconium. Under Rickover’s scrutiny, Bettis scaled a novel purification process to thousands of tons of production capacity. Rickover opened up contract bids for Zirconium only after having derisked this novel technology. When the Secretary of the Navy later asked Westinghouse how they managed to scale up this process, the response he got was “Rickover made us do it”.

Bureaucratic Innovation

Building big things requires lots of people. Rickover was not only an exceptional manager of people and deeply technical, but his 20-year naval career before Oak Ridge taught him how to wrangle government bureaucracy — and discern which rules mattered and which didn’t. For example, Rickover was interviewing an officer who thought the monthly reports on the gasoline usage of his base’s motorboats were pointless and wasteful. Rickover told him to simply remove the tickler file that tracked the reports from the boss’s secretary file and to send over a note the next day alerting Rickover that the task had been completed. The interviewee did and was hired.

Subscribe now

Rickover’s bureaucratic skill is exemplified by his success in rallying the Navy behind the nuclear-powered submarine. He believed this was a feasible, near-term project, despite widely-held convictions to the contrary — including those of the Atomic Energy Commission (AEC). Even Robert Oppenheimer (who served as one of the first AEC commissioners) doubted nuclear propulsion early on.

In light of initial resistance from the civilian AEC, Rickover formulated a unique bureaucratic innovation to position himself within two chains of command — one within the Navy and the other within the AEC.

Rickover was also able to realize his bureaucratic innovation to occupy a spot on the org chart both at AEC and in the Navy BuShips, something he first formulated while at Oak Ridge. This way, if the AEC refused something, he could respond that “this is a priority for the Navy” and vice versa. Similar to how the Manhattan Project reduced risk by pursuing parallel technological approaches, Rickover would reduce his bureaucratic risk by pursuing parallel chains of command. This unique structure lives on to this day, with Naval Reactors shared between the semi-autonomous National Nuclear Safety Administration (NNSA) in the Department of Energy (DOE) and the Navy.

Rickover firmly believed that the right team and the right culture could build incredible industrial technologies at scale, even within the government. While discourse in Washington DC often focuses on regulations or money, Rickover’s life brings a uniquely human-centered view of industrial policy: one that recognizes the importance of state capacity, technical personnel, and most importantly, public leaders with the vision and drive to build technology.

You can read the full story of Rickover and how he built the world’s first nuclear-powered submarine on Charles’s substack.

Founded in 2023, Moonshot AI is one of China’s four new “AI tigers” that’s attracted massive valuations and big-name investors including Alibaba and Tencent. The firm is known for its chatbot, Kimi, whose most recent release highlights improved math, coding, and multimodal reasoning capabilities.

The following piece is a translation of an interview with one of Moonshot’s founders, Yang Zhiling. With a bachelor’s degree from Tsinghua University and a PhD from Carnegie Mellon University, Yang boasts an impressive resume that includes time working at Google Brain and Meta AI. He was also a technical contributor to some of China’s earliest large models, including Pangu 盘古 and Wudao 悟道. His research publications are numerous, and he is the first author of two highly-cited papers in the natural language processing (NLP) field: Transformer-XL, which proposed a method that extends the context length of transformer models, and XLNet, which introduced a way for models to better understand complex data relationships.

This interview was published on the official account of Overseas Unicorn on February 21, 2024. In it, Yang outlines his vision of Moonshot AI as a combination of “OpenAI’s technology idealism” and “ByteDance’s business philosophy.” He covers a couple of key points:

• Moonshot’s goals and how it plans to compete with OpenAI,

• The pursuit of AGI,

• Data challenges and the potential of multimodality and synthetic data,

• Personalized AI models,

• Yang’s approach to leadership and his vision for a global tech future.

Yang Zhilin of Moonshot AI: How Can a Newly Founded AGI Company Surpass OpenAI?

Original article, Archive link.

Interviewers | Tianyi, Penny, and Guangmi. Editor | Tianyi. Typesetter | Scout.

01. AGI: AI is essentially a bunch of scaling laws

Overseas Unicorn: We compare training LLMs to landing on the moon, and the name “Moonshot AI” [literally “dark side of the moon”] is also related to moon landing. How do you view LLM training by startup companies? Under conditions of limited GPU and computing resources, is it still possible to achieve a “moon landing”?

Yang Zhilin: “Moon landing” has several different production factors. Computing power is certainly a core one, but there are others as well.

You need an architecture that simultaneously satisfies both scalability and generality — but today, many architectures actually no longer meet these two conditions. Transformers satisfy these two conditions in the known token space, but when expanded to a more general scenario, they don’t quite work. Data is also a production factor, including the digitization of the entire world and data from users.

So among many core production factors, by changing other production factors, you can make computing power utilization more efficient.

At the same time, regarding “moon landing,” computing power will definitely need to continue growing. Today, the best models we can see are at a scale of 10^25 to 10^26 FLOPs. This order of magnitude will certainly continue to grow, so I believe computing power is a necessary condition. This is because machine learning and AI have been researched for 70 to 80 years, and the only thing that actually works is the scaling law, which is the expansion of these various production factors.

We are actually quite confident that, within a one-year time window, we will be able to achieve a model at the scale of 10^26 FLOPs, and that resources, ultimately, will be reasonably allocated.

Overseas Unicorn: For OpenAI to train their next-generation model, we estimate they have at least 100,000 H100 GPUs, with single clusters reaching 30,000 GPUs. OpenAI is clearly pursuing the “moon landing,” with the possible shortcoming being that they don’t focus as much on user and customer experience. Where will Moonshot AI’s path differ from OpenAI’s? What can Moonshot AI do that OpenAI won’t do?

Yang Zhilin: A key point in the short term is that everyone’s tech vision is not exactly the same. Many areas are not OpenAI’s core competitive strengths (for example, image generation); DALL·E 3 is at least one generation behind Midjourney. GPT’s long-context capabilities are also not state-of-the-art. The lossless long-context technology we recently developed performs better than OpenAI in many specific scenarios because it uses lossless compression technology. You can use it to read a very long article, and it can effectively reproduce specific details and make inferences about the content. Users will discover many scenarios themselves, such as chucking 50 resumes at it and having it analyze and screen them according to their requirements.

To achieve differentiation, I believe we need to look at how large the tech space is: the larger the tech space, the greater the differentiation that can be achieved at the technical, product, and business levels. If the technology has already converged, then all anyone can do is follow the same path, resulting in homogeneous involution.

And I’m actually quite optimistic, because there is still a huge tech space. AGI technology can be divided into three levels:

The first layer is the scaling law combined with next-token prediction (this foundation is the same for everyone, and the catching-up process is gradually converging). On this path [of scaling law with next-token prediction], OpenAI is currently doing better because they have invested the right resources over the past four to five years.

The second level has two core problems. The first is how to represent the world in a general-purpose way. True “general-purpose” representation is like a computer using 0 and 1 to represent the entire world. Transformer-based language models can represent a book, an article, or even a video — but representing a larger 3D world, or all the files on your hard drive, is still difficult. They haven’t achieved token-in-token-out, and are actually still far from the so-called unified representation. Architecture actually solves this problem.

Overcoming the bottleneck of data scarcity through AI self-evolution is another issue at the second level. Today’s AI is actually like a black box, and this black box has two inputs: a power cable and a data cable. After inputting these two things, the box can produce intelligence. Subsequently, everyone realized that the input from the data cable is limited — ie. the so-called data-bottleneck problem. The next generation of AI needs to unplug the data cable, so that as long as power is continuously input, intelligence can be continuously output.

These two core problems lead to enormous space at the third level, including long context, cross-modal generation, the model’s multi-step planning capabilities, instruction-following capabilities, various agent functionalities, and so on.

These higher-level elements will all have enormous differentiation, because there are two important technical variables in between. I believe this is our opportunity.

In addition to the technical level, our values differ somewhat from OpenAI: we hope that, in the next era, we can become a company that combines OpenAI’s technology idealism with the business philosophy of ByteDance. I believe the Asian mindset towards commercialization has certain merits. If you don’t care about commercial value at all, it’s actually very difficult to create a truly great product, or to make an inherently great technology even greater.

Overseas Unicorn: What kind of story should AI model companies tell? Should they frame their narrative around the pursuit of AGI, like OpenAI, or focus on becoming a super app? Are these two narratives in conflict, and how should they be balanced?

Yang Zhilin: The way a company tells its story depends on investors’ mindsets. For us, the more important question is understanding the relationship between these two goals.

AGI and product development are not a means-to-an-end relationship for us; they are both goals in themselves. In the pursuit of AGI, I believe the so-called "data flywheel” is crucial, even though it’s a somewhat old concept.

Products like ChatGPT haven’t yet fully established a continuous evolution loop based on user data. I think this is largely because base models are still evolving — when a new generation is developed, previous user data becomes less useful. This is tied to the current development stage — today, progress is driven by the scaling laws of base models, but in the future, there could be a shift toward leveraging the scaling laws of user data as a source of progress.

Historically, almost all successful internet products have ultimately relied on scaling user data. Today, we can already see signs of this with MidJourney. By leveraging the scaling laws of user data, it has managed to outperform simple base model scaling. However, when it comes to language models and text generation, the scaling effects of base models still far outweigh those of user data. That said, I believe this will eventually shift towards user data scaling — it’s just a matter of time.

This is particularly important now, as we face data bottlenecks. Human preference data, for example, is extremely limited but also indispensable. I believe this is one of the most critical challenges for every AI-native product today. A company that doesn’t care enough about its users may ultimately fail to achieve AGI.

Overseas Unicorn: What’s your view on MoE (Mixture of Experts)? Some argue that MoE isn’t truly a form of scaling up and that only scaling up a dense model improves a model’s capabilities.

Yang Zhilin: You can think of models with MoE and without MoE as following two different scaling laws. Fundamentally, a scaling law describes the relationship between loss and parameter count. MoE changes this function, allowing you to use more parameters while keeping FLOPs (floating point operations per second) constant. Meanwhile, synthetic data changes a different relationship — it allows for data scale growth while keeping FLOPs unchanged.

Following a scaling law is a predictable path, and people try to modify specific relationships within these laws to achieve greater efficiency. That extra efficiency becomes their competitive advantage.

Right now, many believe that simply implementing MoE is enough to achieve something like GPT-4. I think this view is too simplistic. Ultimately, the more fundamental challenge is how to establish a unified representation space and a scalable data production process.

Overseas Unicorn: If compute were sufficient, would anyone build a trillion-parameter dense model?

Yang Zhilin: That depends on how fast inference costs decrease, but I definitely think someone would. Right now, inference costs are too high, so everyone is making trade-offs. However, if compute weren’t a constraint, training a trillion-parameter dense model would undoubtedly perform better than a model with only hundreds of billions of parameters.

Overseas Unicorns: Anthropic has been emphasizing model interpretability, which has sparked a lot of debate. What’s your perspective on interpretability?

You just mentioned that models are a “black box,” and we still don’t fully understand how the human brain works either.

Yang Zhilin: Interpretability is fundamentally about trust. Building a system that people can trust is important, and the applications related to this might be quite different from something like ChatGPT — such as integrating long-context models with search.

If a model never hallucinates or has an extremely low hallucination rate, interpretability wouldn’t even be necessary, because everything it says would be correct. Also, interpretability itself can be seen as part of alignment — for example, chain-of-thought reasoning can be considered a form of interpretability.

Hallucinations can be addressed through scaling laws, but not necessarily in the pre-training stage. Alignment itself also follows a scaling law, meaning it can be solved as long as the right data can be found. AI, at its core, is just a set of scaling laws.

Overseas Unicorn: What are your expectations for AGI? At its core, isn’t the transformer still a statistical probability model? Can it lead to AGI?

Yang Zhilin: There’s nothing wrong with statistical models. When next-token prediction is good enough, it can balance creativity and factual accuracy.

Factual accuracy is usually a challenge for statistical models, but today’s LLMs can exhibit highly peaked distributions. If you ask a model a question like, “What is the capital of China?” then the model can assign a 99% probability to the character “Bei” (as in Beijing). At the same time, if I ask it to write a novel, the probability distribution of the next word would be much more evenly-distributed. Probability is really a method of general-purpose representation (通用的表示方式). In this world, there is a vast amount of entropy. We need to capture the deterministic elements while also allowing the inherently chaotic aspects to remain chaotic.

To achieve AGI, long-context will be a crucial factor. Every problem is essentially a long-context problem — the evolution of architectures throughout history has fundamentally been about increasing effective context length. Recently, word2vec won the NeurIPS Test of Time award. Ten years ago, it predicted surrounding words using only a single word, meaning its context length was about 5. RNNs extended the effective context length to about 20, LSTMs increased it to several dozen, and transformers pushed it to several thousand. Now, we can reach hundreds of thousands.

If you have a billion-token context length, then the problems we face today would no longer be problems.

Additionally, lossless compression is essentially the process of learning determinism from chaos. An extreme example is an arithmetic sequence — given the first two numbers, every subsequent number is deterministic, meaning there is no chaos, so a perfect model can reconstruct the entire sequence. However, real-world data contains noise. We need to filter out this noise so the model only studies the learnable content. During this process, we must also assign appropriate probabilities to uncertainties.

For example, if you generate an image, its loss will be higher than that of generating text because images contain more chaos and information. However, the key is to capture only the aspects you can control, while treating the remaining uncertainty probabilistically. Take a water cup as an example — whether its color is green or red is a probability-based variation, but the shape of the cup remains unchanged. Therefore, the priority is learning the cup’s shape, while its color should be treated probabilistically.

Overseas Unicorn: What patterns exist in the increase of context length? Is there any technological predictability?

Yang Zhilin: I personally feel that there is a Moore’s Law for context length. However, it’s important to emphasize that accuracy at a given length is also crucial. We need to optimize both length and accuracy (lossless compression) simultaneously.

As long as we ensure the model’s capability and intelligence, I believe the increase in context length is very likely to follow exponential growth.

02. Multimodal: most architectures aren’t worth scaling up

Overseas Unicorn: Everyone anticipates multimodal technology will explode in 2024. Compared to text, where do the technical challenges of multimodal lie?

Yang Zhilin: Currently, state-of-the-art video generation models actually use at least an order of magnitude fewer FLOPs than language models. It’s not that people don’t want to scale them up; it’s that most architectures aren’t worth scaling.

In 2019, the most popular architecture was BERT, and people later asked why nobody scaled up BERT. The truth is that architectures worth scaling need to have both scalability and generality. I don’t think BERT lacked scalability, but you can clearly see it lacked generality — no matter how much you scaled it, it could never write an article. Multimodal has also been stuck on architecture issues for the past few years, lacking a truly general-purpose model that people are willing to scale. Diffusion clearly isn’t it — even if you scaled it to the heavens, it could never be AGI. Today, auto-regressive architectures have brought some new possibilities, sacrificing some efficiency to solve the generality problem.

Auto-regressive architectures themselves are scalable, but tokenizers might not be, or eventually tokenizers won’t be needed at all. This is a core problem for 2024.

Overseas Unicorn: If tokenizers aren’t scalable, do we need a completely new architecture beyond transformers?

Yang Zhilin: Just talking about transformers themselves, I don’t think there’s a major problem. The core issue still is solving the tokenizer problem. The transformer architecture has actually already undergone many changes — today’s implementations for long-context and MoE aren’t standard transformers. But the spirit or ideas behind transformers will definitely exist for a long time. The key is how to solve more problems based on these foundational ideas.

Overseas Unicorn: If context length becomes infinitely long, we wouldn’t need tokenizers anymore?

Yang Zhilin: Correct. Essentially, if a model is strong enough, it can process any token, pixel, or byte. With infinite context length, you could directly input everything on your hard drive to it, and it would become your real new computer, taking actions based on all that context.

Overseas Unicorn: Leading model companies like OpenAI and Anthropic think a major bottleneck in 2024 will be data — so they have high expectations for synthetic data. What’s your view on synthetic data?

Yang Zhilin: A scalable architecture is the foundation — and this architecture must first support continuously adding more data before data truly becomes the bottleneck. The data bottleneck we’re talking about now will be encountered in the text modality in 2024, but introducing multimodal data will delay this problem by one to two years.

If the bottlenecks in video and multimodal can’t be solved, then the text data bottleneck will become critical. We’ve actually made some progress on this — if the problem is constrained, such as mathematics or code writing, data is relatively easy to generate. For general-purpose problems, there isn’t a complete solution yet, but there are some directions worth exploring.

Overseas Unicorn: Will the bottleneck in 2025 be energy? Because by then, individual clusters will be very large, which will bring energy challenges.

Yang Zhilin: These problems are actually connected. Eventually, multimodal might solve the data problem, and synthetic data might solve the energy problem.

By the GPT-6 generation, players who master synthetic-data technology will show clear advantages. This is because there are two types of data: “pre-training” data and “alignment” data, the latter of which is more costly to obtain. If you master data-generation technology, the cost of alignment might decrease by several orders of magnitude, or you could produce several orders of magnitude more data with the same investment, changing the landscape.

I think 2025-2026 might be an important milestone: most of the model’s computation will occur on data generated by the model itself.

By 2026, the amount of computation used by models for inference might far exceed training itself; you might spend 10 times the cost on inference, and then one-tenth of that cost on training. A new paradigm will emerge: inference becomes training, and this inference doesn’t serve any users — it only serves to generate synthetic data for itself.

If this happens, the energy problem is also solved, because inference can be distributed. It doesn’t violate any laws; it’s essentially energy conservation. I’m just changing the computational paradigm to allow energy to be solved in a distributed way.

03. Super App: Model Fine-Tuning May Eventually Not Exist

Overseas Unicorn: The search and recommendation systems behind Google and Douyin have strong flywheel effects: their algorithms can provide real-time feedback based on user behavior, continuously improving user experience. LLMs, however, currently can’t provide real-time feedback on user behavior. What will the flywheel effect of AI-native products be?

Yang Zhilin: I’ve thought deeply about this question. The ultimate, core value of AI-native products is personalized interaction, which is something previous technologies haven’t implemented well. So this question is actually about personalization — how to enable users to gain highly personalized interactive experiences the more they use your product. For many products today, the degree of personalization is almost zero. Previously, we could only do personalized recommendations, but now users can interact with products. This interaction is highly anthropomorphic and personalized. How do we achieve this?

I think this is fundamentally a technical issue. In the traditional AI era, achieving personalization required continuously updating models, using small models to solve specific problems. In the large model era, one way to achieve personalization is through fine-tuning — but I believe fine-tuning may not be the fundamental method and may not exist in the long term. Why? When your model’s instruction-following ability, reasoning ability, and contextual consistency ability become stronger, everything only needs to be placed in memory. For example, your large model’s memory can have a bunch of prefixes to follow, reducing costs dramatically. Ultimately, the process of personalizing a model is actually your entire interaction history — which is a collection of your preferences and feedback. This feedback is more direct than products from previous eras because it’s generated entirely through conversational interfaces.

Based on this judgment, the next question is: how to achieve long-context-based customization at the technical level to completely replace fine-tuning?

I believe we’re moving in this direction now. Future models won’t need fine-tuning but will instead solve problems through powerful contextual consistency and instruction-following capabilities. The long-term trend should be personalization of the underlying technology, which will be a very important change.

For example, GPT-4 brought a new computing paradigm where creating GPTs doesn’t require fine-tuning. Previously, customization was achieved through programming, but today it’s achieved by making the model’s prefix very complex, and extracting what you want from this general-purpose set. Personalization achieved this way is truly AI-native personalization, and a traditional recommendation engine plug-in will definitely be eliminated by this new approach.

Overseas Unicorn: How did you make the decision to first develop lossless long-context?

Yang Zhilin: I think the most important thing is to begin with the end in mind. Large models as new computers definitely need large memory, because the memory of old computers has increased by at least several orders of magnitude over the past few decades, and old computers also started with very little memory. The second point is that the ultimate value of AI is personalization.

Overseas Unicorn: OpenAI also has some long-context capability now.

Yang Zhilin: But they haven’t truly viewed the user interaction process as a personalization scenario. For example, if we prompt ChatGPT with something, regardless of whether it’s today or tomorrow, as long as the model version is the same, the effect is basically the same. This is what I mean by a lack of personalization.

Ultimately, everything is instruction-following. It’s just that your instructions will become increasingly complex. Today, your instruction might start with 10 words, but later it could be 10,000 words or even 1 million words.

Overseas Unicorn: Chatbots have always been the ideal for AI scientists. If each user has hundreds of conversations with a chatbot daily, and the chatbot system can collect and understand more user context, will it ultimately far exceed the matching accuracy of search and recommendation systems? Like interactions between colleagues or family members, where just one sentence or even a glance is enough to understand each other.

Yang Zhilin: The key is crossing the trust threshold.

I think the ultimate measure of an AI product’s long-term value is how much personalized information users are willing to input into it, and then lossless long-context and personalization are responsible for turning these inputs into valuable outputs.

New hardware forms may also be needed — but I think models and software are still bottlenecks. To dig deeper, the prerequisite for users to input a lot of information is trust — you need a sufficiently engaging and human-like AI. You can’t say, “I’m setting up product features specifically to get your information.” The end result should be that users and AI become friends, so users can tell the AI anything.

Inflection Pi’s motivation is actually good — wanting to establish strong trust — but Pi may need to take another step forward. How to build trust with users? Human society probably won’t accept being assigned a lifelong companion; that seems somewhat against human nature.

Overseas Unicorn: Moonshot AI wants to create a super app. What does your ideal super app look like? How big does it need to be to qualify as “super”?

Yang Zhilin: It’s about breaking out of niche adoption. When all your relatives are using it, only then have you truly become a super app. And I believe that improvements in AI capabilities will lead product adoption. For example, if character.ai were a perfect multimodal model today, I think its chances of breaking out of its niche would be at least 10 times greater. Ultimately, an application’s ceiling is reflected in the year-over-year increase in connections between AI and humans.

04. Moonshot AI: People with the ability to unlearn make the best talent

Overseas Unicorn: What does the ideal CEO for an AGI company look like?

Yang Zhilin: On one hand, there needs to be a tech vision. You can't just keep doing things that have already been proven to work by others. A real AGI company must have its own unique technical judgment, and this judgment should influence the overall direction of the company. If the top leader can't make decisive calls, that won't work either. At the beginning of the year, we were already working on auto-regressive multimodal models and lossless long-context, but these only became extremely popular in the last couple of months. Even today, lossless long-context is still not widely accepted as a consensus. If you only start noticing these trends now, there won’t be enough time to iterate, and in the end you'll just become a follower.

Another point is having a profound understanding of AI-native product development and then adapting the organization to this new mode of production. In the past, product development was about understanding user needs and designing features accordingly. But in this new era, design needs to be completed during the manufacturing process. ChatGPT’s design was finalized through its creation — it wasn’t built by pre-defining a bunch of scenarios and then finding corresponding algorithms. Similarly, Kimi users uploading resumes and using it for screening was a completely untested use case before we launched, yet it emerged naturally from real-world usage.

Resource acquisition is also crucial, with compute power being the primary cost driver. In the early stages, funding is key, but later on, product commercialization becomes necessary. However, commercialization cannot simply copy mature models from the previous era; it requires innovation. A good CEO and team should have some experience but also possess strong learning and iteration capabilities.

Overseas Unicorn: But maybe some investors can’t tell whose “tech vision” actually leads the pack.

Yang Zhilin: I’m not too worried about this problem. What we have now is the best distribution mechanism: it’s close to a real free market and we will end up with the most efficient resource distribution. What we need to prove to others is not the value of our vision, because a vision is an abstract thing. We need to prove our worth through delivering real models and products. Anthropic received much more funding immediately after it released models like Claude. The market is fair.

Overseas Unicorn: From the perspective of building product- and company-competitive moats, the industrial era relied on economies of scale, and the internet era emphasized network effects. Will there be a new paradigm in the AGI era?

Yang Zhilin: In the short term, changes in organizational structure drive technological advancements — better technology is achieved through better organization, which then directly translates into a superior product experience.

In the long term, network effects are still likely to dominate. The question is: how will they manifest? Traditional two-sided networks from the internet era may still exist, but not necessarily in the form of users and content creators. For AI-native products, the two-sided network effect may be reflected in personalization, where users and the AI engage in a co-creative relationship.

So right now, I see two key areas worth exploring: the continuous improvement of model capabilities and the development of two-sided network effects. These will shape new paradigms in the AGI era. Midjourney has already seen explosive growth through its two-sided effect, while Stable Diffusion, as an open-source model, faces the challenge of being too fragmented on a single side, instead relying solely on base model improvements.

Overseas Unicorn: From the hiring perspective, how do you define strong talent?

Yang Zhilin: I break it down into experience and learning. The ability to learn is a general-purpose capability, which not only includes learning but also unlearning — especially unlearning previous experiences of success. Let’s say you built YouTube from 0 to 1; you might find it harder to work on AI products now than other people do, because you have to unlearn a lot of things. Learning is more important than experience. Maybe in 5 years, the AI industry will cultivate a large number of so-called mature roles. Currently, I don’t actually think that dividing people by roles is all that meaningful, since every person needs to be multi-faceted.

Overseas Unicorn: What kinds of researchers possess “tech vision”?

Yang Zhilin: The core ideas are twofold: focusing on the big picture while letting go of the small details, and maintaining an endgame mindset. I’ve worked with many researchers, and a common issue is over-optimization — getting caught up in refining details while missing the broader perspective. For example, we saw that transformers solved the context length limitations of LSTMs, but if we take a step further back, we realize that each generation of technology is fundamentally about extending context length.

Overseas Unicorn: How many more of these people do you think Moonshot AI still needs?

Yang Zhilin: Objectively speaking, the real limit for us is still supply. Currently, experienced AGI talent is very rare, but there are lots of people with the ability to learn.

But from a demand perspective, the organization cannot become too large — if it turns into just another Big Tech corporation, many of its organizational advantages will be lost. So we will definitely maintain a lean and highly efficient structure. One key judgment is that AGI does not require that many people. In the long run, once we truly “unplug the data,” models at the level of GPT-6 and beyond should be able to evolve on their own, breaking through the limits of human capability.

Overseas Unicorn: How do you assess the difficulty and timeline for catching up with GPT-4?

Yang Zhilin: Hitting benchmark scores on par with GPT-4 is very easy, but achieving its actual performance is definitely challenging. It’s not just a matter of resources — Google has already demonstrated this. In fact, the training cost of GPT-4 isn’t that high; several tens of millions of dollars is not an intimidating figure. This is positive news for us, and we’ve even already made substantial progress.

The most critical factor is having a strong tech vision to anticipate what GPT-5 and GPT-6 will be, and then executing and building the necessary foundations ahead of time. Otherwise, it’ll never be possible to surpass OpenAI. Much of OpenAI’s advantage comes from its foresight — by 2018, it had already committed to what it believed was the right path and spent years building deep capabilities.

Overseas Unicorn: If you were to develop an image-generation AI, how would you approach it? How would you balance language comprehension and image quality?

Yang Zhilin: Midjourney has already done exceptionally well in the single task of image generation. If I were to develop a similar product, I would want it to handle multiple tasks, while still excelling in certain key areas. This is actually the same approach OpenAI attempted, but they didn’t quite succeed.

An AGI company should focus on becoming the default platform — the primary way users interact with AI. Meanwhile, niche user groups will still have specialized needs and ultra-high standards for performance, which is why there’s room in the market for companies like Midjourney. However, if AGI becomes powerful enough, many users will migrate. For example, if I were to repackage all of Photoshop into a single prompt — essentially turning it into an outsourced all-in-one designer — then fewer people would use Midjourney.

Midjourney’s current dominance comes from its first-mover advantage, which enabled it to kickstart a powerful data flywheel. The tricky part is whether such a time window will exist in the future — if not, general-purpose models may eventually outcompete and overtake it.

Overseas Unicorn: Following the strategy of becoming the default platform, how many key user entry points do you foresee in the future?

Yang Zhilin: At least two — one for utility, the other for entertainment.

The way we access information today may become obsolete because, at its core, searching for information is just a means to an end — we do it to complete a task from start to finish. In the future, AI-driven interfaces will likely replace search engines as the primary way users interact with information. Retrieving information is never the end goal; it has just been artificially framed as one. Sometimes we want to accomplish a task, and other times, we want to learn something new. The ideal AGI interface should directly help users complete tasks, rather than simply helping them find information.

Overseas Unicorn: From today onward, how much investment do you think it will take to realize your vision of AGI?

Yang Zhilin: Achieving a fully realized AGI will require investment on the scale of tens of billions of dollars. However, it won’t be a one-time expense — it’s about setting up a self-sustaining loop where the business can generate the necessary resources to fuel further development. This multi-billion-dollar estimate is based on the need to scale up by at least two to three orders of magnitude. Of course, costs will be optimized along the way.

Overseas Unicorn: What should the business model of an AGI company look like? Will it still be seat-based or usage-based?

AGI delivers varying levels of value depending on the task it completes. It may operate more like an outsourced service, pricing each task individually. Beyond that, advertising will undoubtedly play a crucial role. With deeply personalized interactions and conversational engagement, ad monetization could become significantly more efficient than it is today.

Overseas Unicorn: If training costs for models like GPT-4.5, Claude-3, and Gemini-2.0 are around $300 million today — and future models in 2025 could require tens of billions of dollars — does that mean the pursuit of AGI is a trillion-dollar gamble? Have you considered its ultimate impact on human society?

Yang Zhilin: One impact that’s almost certain is a real and tangible increase in productivity. Today, a single piece of software might function at the intelligence level of 1,000 programmers, but in the future, applications could be powered by the equivalent of a million programmers, continuously improving through iteration.

Thinking about the possibilities, everything we take for granted today could change. Training models on a vast range of languages and cultures will inevitably influence values and perspectives. The way people allocate their time will shift — fewer people may work purely for money, and more of human life may be spent in digital or intellectual spaces. Ultimately, we may see the emergence of a massive virtual cognitive ecosystem. To truly build the Metaverse, we may first need to perfect AI.

Additionally, I firmly believe AGI will be inherently global.

Overseas Unicorn: Right now, leading AI models are both powerful and relatively inexpensive, leading to a strong Matthew effect [self-reinforcing cycle in which early winners keep accumulating advantages]. Doesn’t that mean the final market landscape will be highly consolidated?

Yang Zhilin: Within a five-year window, top players might still dominate. However, in 50 years, I believe AGI will be fully commoditized — it will be no different from electricity today.

Despite leading the world in AI innovation, there’s no guarantee that America will rise to meet the challenge of AI infrastructure. Specifically, the key technological barrier for data center construction within the next 5 years is new power capacity.

To discuss policy solutions, ChinaTalk interviewed Ben Della Rocca, who helped write the AI infrastructure executive order and formerly served as director for technology and national security on Biden’s NSC, as well as Arnab Datta, director at IFP and managing director at Employ America, and Tim Fist, a director at IFP. Arnab and Tim just published a fantastic three-part series exploring the policy changes needed to ensure that AGI is invented in the USA and deployed through American data centers.

In today’s interview, we discuss…

• The need for new power generation driven by ballooning demand for compute,

• The impact of the January 2025 executive order on AI infrastructure,

• Which energy technologies can (and can’t) power gigawatt-scale AI training facilities and why Jordan is all-in on GEOTHERMAL,

• Challenges for financing moonshot green power ideas and the role of government action,

• The failure of the market to prioritize AI lab security, and what can be done to fend off threats from adversaries and non-state actors.

Watch below or listen now on iTunes, Spotify, YouTube, or your favorite podcast app.

Constructing Compute

Jordan Schneider: Ben, why were you, as an NSC director, spending time in SCIFs dealing with energy permitting policy on federal land?

Ben Della Rocca: That’s a great question. Spending so much time reading about environmental permitting law and watching law school lectures on how the Clean Air Act works was not how I envisioned spending my time as an NSC director, but there I was doing just that in the SCIF for my AI work.

The United States has a lead in AI thanks to our thriving innovation ecosystem and the immense engineering and other talent. However, that lead isn’t guaranteed. Which country will lead in artificial intelligence will come down more and more to where AI can be built most quickly and effectively.

By “built,” I’m not just talking about the technical engineering challenge of how to do large-scale AI training runs from a computer science standpoint, but also the physical building challenge. This involves developing the large-scale computing infrastructure and energy infrastructure to execute these ever-growing AI training runs.

What became clear to us in the last administration was AI’s significant impact on national security, the economy, and scientific advancement. We also saw an exponential increase in the demand for computing power and energy resources to develop frontier models.

The amount of compute required by frontier AI models is increasing by a factor of 4-5x annually based on publicly available statistics. That’s an exponential pace of growth. Even when you factor in increasing energy efficiency of computational resources and other countervailing factors, you’re still looking at gigawatts of electricity needed to execute training runs at the frontier within the next few years.

There are multiple power-related challenges when it comes to developing and deploying AI. Assuming current trends continue, you need gigawatt-scale training facilities to develop models at the frontier — that’s one set of issues. Then there’s a separate but related set of issues around developing a robust network of potentially smaller-scale data centers around the country to actually use these tools effectively in different locations. This is a significant energy challenge as well.

At the National Security Council, I led the White House’s work around AI infrastructure and developed the AI infrastructure executive order that came out in January 2025. What the executive order tries to do is directly address that first challenge — how to build these large gigawatt-scale training clusters that, assuming the current paradigm of AI training continues, will need to be built for the United States to maintain its lead.

Jordan Schneider: That’s a tall order you set for yourself, Ben. In your mind, what parts of the executive order will matter most when building out AI data centers in the US?

Ben Della Rocca: The executive order includes a wide range of things that address not only how we bring gigawatt-scale data centers online in this country, but also a broader, distributed network of data centers around the country. There’s a lot in there, but I’ll highlight three key mechanisms.

1. Construction on federal lands — the centerpiece of the executive order establishes a mechanism by which AI data centers can be built in a streamlined and more efficient way on federal sites owned by the Department of Defense and Department of Energy. There’s a huge value proposition to building on DOD and DOE sites because by building on federal lands, you bypass many of the state and local land use permitting requirements that usually make data center construction and similar projects take such a long time. Additional burdens on the federal permitting side are taken on as a result, but the federal government can and will be doing much under the executive order to make those processes proceed as expeditiously as possible.

2. Regarding bringing new power generation online — operating gigawatt-scale data centers will require a lot of new power added to the electric grid. This is challenging for numerous reasons, including permitting requirements that complicate construction broadly, delays with interconnection to the electric grid, and other required approvals. The executive order directs the Department of Energy to establish requirements to collect and share information with data center developers regarding unbuilt power projects that have already received interconnection approvals to accelerate the power procurement process. It also directs the Department of Energy to engage utilities to reform their interconnection processes for faster progress.

3. Regarding transmission — the Department of Energy is directed to use some of its powerful authorities to partner with private sector transmission developers in building transmission lines much more efficiently and quickly than business as usual, and to take part in the planning process as well. You also have other actions to bolster the supply chain for transmission and grid equipment, which would be really useful for the long-term vitality of this industry.

Between these sets of things and other actions to expedite the permitting process within the federal government’s authority, the executive order lays out a pathway for building gigawatt-scale facilities on the timelines that we expect leading developers will ultimately need for AI training.

Jordan Schneider: Tim, what's missing from the executive order?

Tim Fist: We highlight two big gaps.

1. The EO comes along with a clean energy requirement. All the energy that you're producing to power these data centers needs to come from clean energy sources. It allows the use of natural gas with carbon capture, but that technology can’t be implemented on the timescale required.

2. While building on federal lands owned by DOE and DOD allows you to bypass a lot of state and local permitting issues, it does open up the issue of NEPA, which automatically applies if you’re building on federal land.

Our recommendation around that is using the Defense Production Act to speed up permitting on federal land, as well as resolve supply chain issues.

DPA gives the President broad authority to intervene in the economy where this is seen as necessary to ensure the supply of technology that's deemed essential to national defense. Our claim is that AI definitely fits within this scope.

Because the most powerful AI systems are now being developed by private firms, a lot of the DoD's future capabilities are likely going to come from models that are trained in data centers operated by private firms.

OpenAI recently announced a partnership with Anduril to bring its models to the battlefield. ScaleAI has built a version of Meta’s Llama, which they call Defense Llama, to help with military planning and decision making. Palantir is building platforms for DoD as well.

Arnab Datta: We mentioned two authorities in the DPA. Title I, which is prioritization, would allow the federal government to tell contractors to prioritize transformers, turbines, and other hardware for AI data center use. The other authority in the DPA is Title III, which is a financial assistance authority. There are also authorities within the DPA and Title III where you can streamline some of these permitting issues that Tim described.

Jordan Schneider: Can you explain how DPA authorities would help?  

Arnab Datta: I’ll use a practical example — right now, natural gas turbines are sold out. GE Vernova said they could be sold out past 2030.

DPA would allow the president to say, AI is a national security priority, and those data center turbine contracts need to be fulfilled first.

I don’t want people to think that our idea and what we’re proposing is about getting out of environmental rules to build this infrastructure. The important thing is that we’re talking about using these to streamline the procedural laws associated with environmental review.

This doesn’t necessarily mean that an environmental review wouldn’t be conducted prior to leasing federal land, for example. Finding ways to limit the likelihood of litigation could be very important here, and that’s where some of the national security exemptions are most useful.

The Geothermal Goldmine

Jordan Schneider: Let’s take a step back here, because we got really deep, really fast. The renewable regulations will obviously disappear in the next few weeks, and NEPA was just nerfed by the Trump administration.

The worry for a hyperscaler might be driving their data center through the process and then getting sued four years from now in a different administration. But even then, the data centers will already be built. We’ll have AGI by then. No administration is going to shut that off because you took advantage of a permissive regulatory environment.

I’d like to discuss the hard constraints when it comes to actually building and deploying the electricity needed for these data centers. A fascinating part of Tim and Arnab’s work was going through all the potential technologies that could provide the marginal electricity needed for these data centers, and ranking them by potential. Let’s do overrated and underrated. What are three overrated sources of electricity that get too much attention in the broader discourse about the future of AI?

Arnab Datta: I don’t love the word “overrated,” but there is sometimes an implication that natural gas makes this super easy. The reality is that gigawatt-scale energy projects, particularly off-grid projects that many compute companies are moving toward, represent massive investments.

By off-grid, I mean not connected to the transmission system — you’re building close to your data center and it’s exclusively powering your facility. This is an extremely expensive and risky investment regardless of how much cash you have on hand. Building energy infrastructure at a gigawatt scale is very costly.

When discussing natural gas, yes, it’s a proven technology, but there are stranded asset risks. If an AI data center outlives its useful life in that location, or if something becomes more cost-competitive and a company wants to switch, you face challenges. Natural gas is important and should definitely be part of the mix, but I wouldn’t say it’s an easy decision. There are supply chain challenges, and the notion that “we can solve this with natural gas” oversimplifies the issue.

Jordan Schneider: Tim, what do you think?

Tim Fist: The obvious one is the much more long-term technology — fusion. We see some hyperscalers signing power purchase agreements for fusion energy. A power purchase agreement is a commitment to buy a fixed number of kilowatt-hours at a future point, and the fact that they’re buying this for fusion is rather incredible when there isn’t a viable commercial fusion reactor that’s ever been demonstrated.

This technology is clearly so far off that it won’t matter over the timeframe we’re concerned about, which is how we ensure we can build this infrastructure over the next five years.

Ben Della Rocca: Each approach has real downsides. There are real risks, problems, and inefficiencies that aren’t fully recognized.

To pick one where there’s sometimes optimism that should be qualified — we need to be realistic about nuclear options in particular. Nuclear energy is something I would be very excited about in a longer-term timeframe, such as the 2030s. However, it’s very difficult to see that being part of the solution for the late-2020s challenges we may face regarding AI’s energy use. We should be realistic about the timing involved.

Tim Fist: We’re at this weird point in history. Twenty years ago, the answer would have been obvious because renewables were completely non-viable, and there weren’t exciting next-generation technologies coming along.

Right now, we’re at a point where multiple technologies are approaching the same level of cost competitiveness simultaneously. Large-scale battery plus storage is now better than natural gas in many areas. Advanced geothermal is becoming super interesting but hasn’t been properly scaled or demonstrated yet. Small modular reactors are just coming online and are probably the next big thing once we can scale them up.

In the near term, natural gas seems like the obvious solution. However, it will likely become an obsolete technology within about 10 years. That’s the core problem we’re trying to grapple with.

Jordan Schneider: I was disappointed about dams. I thought they could be a viable option, but you disabused me of that notion. They’re too slow or not big enough. There’s no new innovative dam technology that’s been developed over the past 70 years, which I was disappointed to discover.

Arnab, what are you excited about?

Arnab Datta: I’m incredibly excited about next-generation geothermal energy. This is energy produced from heat in the Earth’s crust — the heat beneath our feet, as it’s often called. We’re pioneering this energy technology because of our experience with fracking and how we developed drilling techniques that led to the shale revolution.

There hasn’t been enough demonstration yet, though some companies are innovating quite rapidly. The scale potential is remarkable, and it’s an area where the US can really lead because we have an oil and gas workforce where 61% of workers have skills directly transferable to geothermal. We have a supply chain for fracking and shale production that’s ready to go and transferable to next-gen geothermal. The potential is incredibly high, and I wish we were doing more to support it.

Jordan Schneider: Can we stay on the technology for a moment? How do I drill a hole and get electricity out of it?

Arnab Datta: Basically, you’re drilling into the Earth’s crust where there’s substantial heat, and you’re pumping fluids down into that heat. The fluid gets heated up, and you’re circulating it back to a steam turbine that generates electricity. That’s the simple explanation.

Jordan Schneider: So it’s just like a steam boiler with the Earth’s core as the power source? That’s incredible.

Arnab Datta: Yes. There are multiple types of systems. Traditional geothermal requires three elements coming together naturally: heat, fluid, and a reservoir. These are natural geothermal reservoirs.

What next-generation geothermal does is create artificial reservoirs. You’re digging and fracking to create cleavages in the crust, and then cycling fluid through it. This is safe, to be clear. It has been tested and demonstrated. This isn’t something that’s going to damage the Earth. That’s the basic explanation for how it works.

Tim Fist: The amount of heat energy stored in the Earth’s crust that you can access via enhanced geothermal vastly exceeds the amount of energy in all known fossil fuels by several orders of magnitude. This is an abundant source of low-carbon energy without any of the intermittency problems of solar and wind that you can also access using many of the same tools we’ve developed for large-scale fracking.

This technology has already been deployed. Google is powering some fraction of its data centers with this. At this point, it’s primarily a scaling problem.

The areas where you can extract the most heat from the Earth’s crust using these methods also overlap substantially with the areas where you have federal land that can be readily leased. It’s a perfect recipe for solving this problem.

Jordan Schneider: The Earth is warmer under Nevada?

Arnab Datta: The heat is closer to the surface. To add one thing to what Tim said as an example: you have to drill three wells to produce about 10 megawatts of energy in something called a triplet. To reach five gigawatts, which is our goal by 2030, you would need to drill 500 of those triplets.

Jordan Schneider: This is child’s play.

Arnab Datta: That means 1,500 wells. We have drilled 1,500 new wells multiple times in the shale region of this country in a given year. This is not that many new wells to drill, if we can perfect the technology. We’re very well-positioned to take advantage of this, if we can get there.

Ben Della Rocca: I would underscore that geothermal is the single energy source I’m most excited about, in terms of technologies that are underrated by the broader public.

AI itself provides an opportunity to create much of the backstop demand that can funnel capital to the industry and incentivize development and technical advances needed to make the United States a global leader in this technology and advance our energy leadership more broadly.

As Tim mentioned, traditional geothermal resources are primarily available in the western United States. This overlaps heavily with places where large amounts of land are owned by the Bureau of Land Management. One of the sources of delay with building geothermal projects on federal lands has been federal environmental permitting reviews, which take time.

The executive order has directed the Department of Interior to find ways of conducting these reviews much more quickly — eliminating redundant reviews at multiple stages in geothermal projects and creating what are called “priority geothermal zones.” These are areas where the Department of Interior will focus its permitting efforts to move the process along as expeditiously as possible.

Ideally, these zones would overlap with places where AI data centers are being built, to ensure that all efforts are moving in the same direction. There’s a lot more to be done, but we’ve seen a valuable starting point to accelerate development in the geothermal space.

Jordan Schneider: What is the environmental consideration? We don’t even have oil gushing out. Are there endangered species in rock 20,000 feet below the Earth’s surface? It’s just ten guys and a drill.

Ben Della Rocca: That’s a great question. Certainly, the environmental repercussions are fewer than with traditional fracking for the oil and gas sector. With any construction project like this, you have to build a power plant, which involves some change to the natural environment. If there’s an endangered species right where you want to build the power plant, that will be a factor in the environmental analysis. Drilling deep down can potentially cause some impacts to the broader region as well.

The environmental burdens are significantly less, which is why there’s potential for the permitting to go more quickly. It’s a question of marshaling the right policy resources to ensure we’re all moving as quickly as possible given the lesser concerns with this technology.

Jordan Schneider: What are some lessons from the shale revolution that can potentially apply to the US government helping incentivize the development and production of geothermal?

Arnab Datta: In the 1970s, coming out of the Arab oil crisis, we made a conscious effort to support the non-conventional production of energy. By the mid-2010s, we were the leading producer of oil and natural gas.

How did that happen? There were four key policy interventions that occurred over those decades that I would emphasize. My colleague at Employ America, Skanda Amarnath, and I wrote about this last year.

First, there were numerous research and development and cost-share programs to innovate in drilling and develop new techniques. The Department of Energy worked directly with Mitchell Energy, sharing some of the drilling costs to test non-conventional means of production.

Second, there were supply-side production tax incentives and demand-side price support. On the supply side, there was a Section 29 tax credit — essentially a production tax credit for non-conventional sources. An analogous current example is the Inflation Reduction Act, which supports production for new types of energy. It’s important that those credits remain in place.

On the price support side, there was targeted deregulation in the Natural Gas Policy Act in 1978 that exempted energy produced from non-conventional sources from the existing system of price controls. This essentially created a price support incentive. People describe it as good deregulation, but its ultimate purpose was to create a more competitive price environment for this type of production.

Third, there were permitting changes that altered the regulatory environment. The Energy Policy Act of 2005 established a legislative categorical exclusion, meaning certain types of production with specific geographic footprints could undergo the lowest level of NEPA analysis and get approved more quickly.

Fourth, and often underrated, was a highly accommodative macroeconomic environment. The shale boom and major productivity increases happened at scale in the late 2000s and early 2010s when interest rates were low. Companies could take on cheap debt and iterate, with abundant capital available for them to enhance productivity to the point where production levels increased even as the workforce declined because drilling techniques became so efficient.

These four factors help explain what happened in the shale revolution. We need to figure out how to compress that timeline to just a couple of years for next-generation geothermal energy.

Jordan Schneider: How far away are we today from these awesome steam boilers?

Arnab Datta: Fervo is currently building what I believe is a 400-megawatt facility. I don’t know the exact state of where that stands in project development since it’s a private company and I’m relying on public information. They’ve demonstrated that their technology works at a small scale.

Tim mentioned the Google facility. One of their smaller installations is powering a data center at around 40 megawatts. There’s also a company based out of my home city, Calgary, Canada, called Eavor that’s working with horizontal drilling. They have a small demonstration project as well.

The real question is whether we can achieve scale. The major challenge for these companies is securing enough capital to demonstrate that the technology works and can produce utility-scale electricity.

Jordan Schneider: You mentioned capital. Where hasn’t this been coming from, and where should it come from to realize this vision for geothermal over the next five years?

Arnab Datta: That’s a great question. In our report, we discussed the challenge of financing these next-generation technologies — whether geothermal, small modular reactors, or others. The fundamental challenge is that despite the promise, there’s tremendous uncertainty associated with developing these technologies.

The key is finding investors willing to accept this level of uncertainty in project development. They need to be comfortable covering costs when they increase because permits take longer than expected, supply chain issues arise, or interest rates climb, making capital more expensive. All these uncertainties accumulate, making equity investors reticent to invest at a sufficient scale.

There are only a handful of venture capital firms that engage in this type of investing, and they reach their limits quickly. Banks won’t do it because the risk of failure amid such uncertainty is too high. The government is playing a role — if you look at the Office of Clean Energy Demonstrations, they’ve funded demonstration projects for SMRs.

In the shale context, the federal government shared costs with Mitchell Energy for drilling operations. We need some version of that approach now. We also need the federal government to reduce uncertainty.

When you think about technology demonstration, you typically start with a concept paper, hoping to attract investment for the next step — a small-scale project. Once you secure that investment, you aim for a larger-scale project, attracting a bit more funding. It’s a slow, incremental process of building investor confidence. We need to compress this timeline at each stage and reduce uncertainty so companies can invest.

Financing energy projects typically requires three elements — debt, equity, and offtake agreements (meaning someone to purchase the energy once you’ve produced it). Currently, we see headlines about AI companies “investing” in energy projects, but they’re mostly doing this through these power purchasing agreements.

Subscribe now

With next-generation energy, there’s substantial uncertainty from factors that aren’t easily quantifiable — permitting and regulatory timelines, physical feasibility, and potential material bottlenecks. One of those three participants needs to own that uncertainty, and it’s generally not going to be through offtake agreements.

You would need very high-premium offtake agreements to cover a level of uncertainty that would give debt or equity investors confidence that their investment won’t fail. In our report, we emphasize that if the federal government can either own that uncertainty by providing capital or reduce it through streamlined regulatory procedures, that could unlock the tremendous amount of capital these companies possess, enabling them to invest directly in projects upfront.

Currently, there simply isn’t enough upfront capital, and that’s the barrier we’re trying to overcome. Many tech companies are sitting on substantial cash reserves, but they’re not directing it toward upfront energy investment. We believe that through federal government initiatives that reduce or assume this uncertainty, we might encourage companies to invest capital directly.

I should note that Amazon has invested directly in an SMR project in the Pacific Northwest, and that’s a model we’d like to see replicated more broadly and at a larger scale. That’s what we’re trying to accomplish.

Jordan Schneider: You gave me some great new acronyms, Arnab. We have FOAK, SOAK, THOAK, and NOAK — first of a kind, second of a kind, third of a kind, and Nth of a kind. We’re still in the “first of its kind” universe. Let me push back on this hype train a little bit. Three years we’re going to go from some cute demonstration projects? I take your point that if you can figure out the pumping mechanism, you have a lot of people happy to live in Nevada for a while and drill big holes in the earth. But AGI’s coming soon. Is this really going to get us there?

Tim Fist: This underscores the validity of an all-of-the-above energy approach, where we’ll want to take multiple shots on goal. We don’t want to put all our eggs in the geothermal basket. If you think about it, we had this staging approach for different technologies that could work really well under this all-of-the-above strategy.

You deploy natural gas plants because you know you can bring those online. They’re going to provide secure, reliable energy, and we know how to build them within a couple of years. Solar and battery storage is a really promising option to build out as much as we can where we can get it online.

Building out geothermal should come alongside that. Small modular reactors would come a bit after that as well. If you want to start thinking about fusion, maybe you’ll bring that online in 20 years. Basically, you want to invest in as many of these technologies as you can at once to address the inherent technological risk with this next-generation stuff.

Ben Della Rocca: One additional thing I’d say to your question, Jordan, is that even within the geothermal space, we can talk about this layered or staged approach where we lean more on some geothermal technologies at one point and then move on to others.

Traditional geothermal technologies are more tried and tested. We’re not necessarily doing first-of-kind projects for some of those geothermal hydrothermal resource projects. Fewer potential gigawatts can be brought onto the grid for that technology, but that might still be enough if we push people to do the right exploration and resource confirmation as they’re building their data centers. That could still be a meaningful part of the solution by 2028.

Beyond 2028, that’s when it might be most realistic for some of the enhanced geothermal projects, which are still in the first-of-a-kind stage, to come online. There’s a bit of phasing that we can do that way.

Tim makes an excellent point that there are different strengths and weaknesses to each of these approaches. It’s unrealistic to think that any one energy source is going to be the sole answer to AI’s energy needs.

With solar and batteries in particular, combining those two resources can be a way to access firm power, and there are downsides to the ability of that to scale in some cases. But it’s also faster to build solar plants, and there’s already been work done to review the environmental impacts of solar developments on some of the Western land under government management. This could make construction of those projects proceed more quickly in certain cases and be an important part of the shorter-term solution.

People really should be looking at a wide range of different options and leveraging different site-specific opportunities.

Transmission and Permitting Reform

Jordan Schneider: Can we talk a little bit about transmission lines and transformers? Arnab, you mentioned that a lot of this stuff may just end up being off-grid where Google’s responsible for building the power right next to its new data center. To what extent does hanging lots of transmission lines over people’s farms or whatever actually matter for all this stuff?

Ben Della Rocca: Transmission is a big part of the equation. You could certainly imagine a world, as you said, Jordan, where all the power resources are co-located and we don’t need to transport any electricity from one site to another. That’s theoretically a solution, but it’s unlikely that when we’re talking about gigawatt-scale data centers, at least in the short term, that we’re going to find sites where you can truly get three to five gigawatts of co-located power. Not saying it’s theoretically impossible, but it would be unrealistic to assume there’s a world where we just don’t need transmission lines to solve this problem.

At a minimum, having transmission lines provides a number of other benefits. First, it puts a much larger range of resources in place. If you have a data center being built somewhere around a variety of different BLM lands that are amenable to different energy sources, you can tap into more of them if you can deliver power from offsite to nearby locations.

Even if you are building power generation on site, there are still many advantages to interconnecting that power to the electric grid. It provides stability benefits, removes some of the need to build a microgrid or other sorts of redundant electrical facilities on site, and mitigates some of the financial risk of your project. If you end up using less power than expected, you could resell it onto the grid. Transmission lines are going to be important no matter what.

Subscribe now

I’d like to highlight a couple of things that the executive order tried to set in motion that could help us be forward-leaning on building the transmission infrastructure needed for AI data centers going forward. The most important is that the Department of Energy has some very important statutory authorities to address these problems.

One relatively well-known authority worth mentioning is the ability to establish National Interest Electric Transmission Corridors (NIETC), which are areas the Department of Energy can designate to play a backstop role in accelerating certain permits and approvals if they’re taking a very long time in a way that impedes efficient transmission development. This could be useful in the longer term, though it does take a meaningful period to actually establish a particular region as a NIETC and activate those authorities.

Another set of less well-known authorities that should be fully explored are those allowing the Department of Energy to partner with transmission line developers in powerful ways. Several statutes — such as the Energy Policy Act of 2005, provisions in the American Recovery and Reinvestment Act, and the Infrastructure Investment and Jobs Act — essentially let the Department of Energy create public-private partnerships or work with companies to participate in upgrading, constructing, planning, and financing transmission lines.

Based on analysis that the Department of Energy published under the Obama administration, in some circumstances, these authorities might be used to essentially bypass the need for lengthy state approval processes and allow the Department to play a more efficient role in cost allocation and other processes that state public utility commissions typically handle, which can often take many years to complete.

By using these authorities more aggressively, there’s actually a much faster pathway to building transmission lines, particularly relatively shorter ones. We’re talking about dozens of miles to connect a data center to the grid rather than hundreds of miles of interstate transmission lines. We’re not talking about giant transmission projects, but more targeted transmission builds that DOE can develop with the private sector. This could be a really important pathway to getting more gigawatts on the grid by 2028.

Arnab Datta: I would second basically everything Ben said. Transmission is really important. Some of the short-term solutions Ben identified are exciting, and it would be great to see them implemented.

The reason firms are moving off the grid, making that such a significant factor, is because transmission is just so difficult right now. This underscores the need for longer-term reform and broader permitting reform, with Congress actually taking action. There’s only so much you can do through the executive branch, and it tends to be more imperfect. We need to fix that.

Tim Fist: It takes, on average, 10 years at present to build a new transmission line in the United States. That’s mostly due to these holdups in permitting.

Ten years ago, we built 4,000 new miles of transmission lines every year. Now, it’s more like 500 miles. This is decreasing by a factor of eight.

Jordan Schneider: Ben, what’s a transformer and why does it matter?

Ben Della Rocca: Conceptually, a transformer changes the voltage of an electric current. This transformation process is essential to bring electricity from transmission lines or higher voltage environments to voltages that an actual end-user facility can accept. We need transformers within electric infrastructure to make the power usable for its intended purposes.

The challenge with the transformer industry is that there’s limited capacity with current resources allocated to transformer development to supply an adequate number of transformers needed to build all the power infrastructure that AI is demanding.

Much can be done to support the transformer industry through loan guarantees or other financing options provided or encouraged by government. These measures would allow the transformer industry to invest in the capital expenditures needed to expand their facilities or train new workforce and add workers to existing facilities. All these steps are important for reducing transformer lead times, which I believe are currently in the two-and-a-half to three-year range. Transformers are certainly an important part of the supply chain aspect of this problem.

Jordan Schneider: Since we were talking about the Defense Production Act earlier, now that corruption is in and FARA is dead, how far can a president who really just wants to potentially let their main consigliere who happens to be building giant AI... Just get all the gas turbines before everyone else? Is there any recourse here for something crazy like that?

Arnab Datta: Broadly with DPA use, when utilizing any aggressive or assertive legal authority, it’s important to try to get political buy-in, even if you believe the legal authority is bulletproof and you can do what you want. You still want political buy-in.

The fact that Ben’s here discussing how the Biden administration prioritized this because they saw it as a threat, and the fact that the Trump administration thinks this is a threat and that there should be a national security aspect to AI data center build-out shows there is some consensus.

The DPA is up for reauthorization, which typically happens in a bipartisan fashion. This presents an opportunity to take advantage of that consensus and say, “We are appropriating X amount of dollars and authorities for the DPA to be utilized to help our energy infrastructure build-out for AI data centers,” and put some safeguards on it. If Democrats are concerned about the hypothetical scenario you just described, that can be a negotiating chip for what is typically a bipartisan reauthorization.

Generally, if your concern is around corruption with this issue, there is an opportunity here because the DPA is up for reauthorization. There were actions during the Biden administration regarding using the DPA for heat pumps that Republicans didn’t like, and there were hearings last year on this. Building some kind of legislative consensus could be useful here.

Ben Della Rocca: Arnab has made many great points, and I’d add that in this issue space, the role of litigation shouldn’t be underappreciated in terms of the type of check it can play. This has long been the dynamic with infrastructure projects in many different industries.

The way litigation works around the National Environmental Policy Act and other permitting-related statutes is that lawsuits can allege that permitting requirements haven’t been fully fulfilled. This can result in injunctions that ultimately delay projects while court proceedings are ongoing.

In effect, this means there’s significant value in making sure all the T’s are crossed and I’s are dotted when pursuing an option that uses national security authorities. If you do something that goes outside the bounds of the law or isn’t an airtight legal case, the odds of litigation increase, which can ultimately result in projects being delayed.

If you can’t get past the pre-construction stage because you’re dealing with extensive litigation, that can have huge consequences for the timeline of building artificial intelligence infrastructure, and time has a real premium in this space. The need to ensure that laws are followed very closely here shouldn’t be understated, for a wide variety of reasons.

Jordan Schneider: A lot of the dynamics you both just pointed out also apply to all the NEPA stuff we were discussing. Ben was doing some clever things here and there to try to make it easier for firms, and then Trump just cancelled NEPA, which is legally questionable. We’ll find out. On one hand, it’s probably exciting for Google and Amazon. On the other, you’re opening yourself up to a whole new legal attack surface that wouldn’t exist if we were living in a Harris administration that followed the direction the executive order laid out more directly. Anything else to add on that dimension?

Ben Della Rocca: Your point about the uncertainty here is exactly right. Trump’s rollback of NEPA as it has existed for years certainly has the potential to speed things along, but it doesn’t ultimately eliminate the fundamental statutory requirement, which is for agencies to essentially do their best to review the environmental consequences of their actions.

Without clear regulation, there’s going to be significant ambiguity and uncertainty about what that means, and there will still be years of past practice that courts may look to when determining the exact content of the statutory requirement. The actual magnitude of the impact from efforts to rewrite NEPA regulations remains to be seen. It may take years to play out.

Arnab Datta: One quick thing to add is that it’s important to note that while the NEPA regulation passed by the Council on Environmental Quality was rescinded, every agency still has its rule in place for conducting a NEPA analysis, and those remain in effect. Right now, that’s the rule of the road. There’s a long-term uncertainty that Ben is right to discuss here, but in this immediate moment, the regulatory framework still exists for agencies, and that’s important for people to know and continue to comply with.

Attaching Strings

Jordan Schneider: Tim, you wanted to add a requirement to make these hyperscalers take AI security more seriously before they get access to government financial help. What is the market failure here, and what are the sorts of things you think the government should add to their requirements in order to get all of these special dispensations?

Tim Fist: US AI companies are currently building models that they believe could, within just a few years, reshape the global balance of economic and military power. Consider AI systems that can autonomously carry out massive cyber attacks, automate scientific R&D processes, or serve as substitute remote workers for many kinds of jobs. If this is true, we really need to protect these systems against theft by bad actors.

It turns out that many of the security problems you need to solve are at the data center level. Protecting against sophisticated threat actors like nation-state hacking groups is both extremely difficult and expensive. If a company invests adequately in security, they risk falling behind competitors who aren’t making similar investments.

A core part of the executive order ties assistance around loans and permitting to strong security requirements. This creates a set of requirements that hyperscalers and AI companies can follow to raise the level of security protecting their critical intellectual property against these threats. By connecting it to this assistance, you transform something that would put you at a disadvantage relative to competitors into a strong commercial decision.

We outline several ideas for what this could look like. Specifically, it means finding the best existing standards and applying them across the board, developing new standards and guidance specific to the threat model for attacks on AI model weights, and providing government assistance for supply chain security, physical security for AI accelerators, background screening for personnel to protect against insider threats, and counterintelligence playbooks.

The basic idea is to create a strategic partnership between the government and the AI industry to improve security, with incentives on the other end to make it worthwhile.

Jordan Schneider: The piece that has really struck me when reading Dario’s article about his world in which America gets ahead and accelerates toward AGI faster than China, then the incentives for the Chinese government to leave these data centers alone falls to basically zero.

If you can steal the model weights, then maybe you want OpenAI and Anthropic to continue existing to make cool stuff that you can take and deploy. But if this is the technology which is to rule all technologies, you start to get into a US-Iran 2000s/2010s dynamic where stuff like Stuxnet or drone attacks become a concern. There’s water for cooling everywhere — what if there’s just a giant leak that fries all your servers?

The physical security of these tens or hundreds of billions of dollars being thrown into data centers hasn’t really been discussed much. You can see the potential future where that ends up being a critical part of what the US government and the firms themselves need to focus on for safety.

Tim Fist: I’m more optimistic about protecting models than you are, with the caveat that we need to think about the scope of things that it’s useful to protect.

To be more specific, I expect that over the next few years, the most powerful models developed by US frontier labs are going to be deployed internally first. There are three main reasons for this:

First, as capabilities grow, there will be numerous misuse concerns that labs will want to address before wide deployment.

Second, deploying internally before wider release makes a lot of technical and economic sense as you can use the model to accelerate your own R&D before releasing it more broadly.

Third, it makes sense to first train the big expensive model and then distill it down to a version that’s more economical to serve to users. This is reportedly now common practice across basically all the frontier labs.

If this is true, then protecting cutting-edge models can be done in a more favorable security environment where your attack surface is relatively smaller because you’re initially only deploying for internal use cases.

Eventually these models will likely get stolen, but protecting the bleeding edge from immediate theft is still worthwhile as it allows you to use those models to maintain your overall lead by investing your inference compute into AI research and development, and using those models to develop things like AI-powered cyber defense.

There’s some hand-waving in this theory of victory, and there are many unknowns, but seriously trying to predict this is worth it. The alternative is freely handing it over to China — putting all this money into power and chips and then giving the products freely to China to accelerate their own AI research programs.

Preventing denial of service or sabotage operations is also a worthwhile goal. I’ve seen interesting research recently about the susceptibility of current AI data centers to cheap drone strikes as well as attacks on surrounding network and energy infrastructure. I don’t have a view about how expensive this will be to defend against, but it certainly needs to be a huge part of the investments in defense.

Jordan Schneider: One thing I’ll say is that if America is going to win, it’s going to need PRC nationals working in these labs. If we’re doing FBI counterintelligence checks on every AI PhD Berkeley graduate — I’m sorry, we’re just not going to have an AI ecosystem. There’s some middle ground there, but that’s the one piece I was most skeptical of.

I had one random question. There was this very funny chart that Tim and Arnab had where Google, Amazon, and Microsoft all committed to being net zero by 2030, and they’re on this trend line. Then it just starts to go the wrong way once they realize they have to build tens of billions of dollars of data centers.

Do those commitments just go away? In our anti-DEI world, is there anything statutory about it? Is Blackstone going to get mad at them? What’s the forcing function here that would keep them on those trend lines, absent some really amazing geothermal breakthrough?

Arnab Datta: I wrote about this recently. The way to get more adoption of these newer technologies that are firm and emissions-free is for them to become cost competitive and quick to deploy. I don’t know how firm the commitments are from Amazon and Google or how sticky their internal social costs of carbon are.

I’m trying to think about policymakers and what we can do to get to that place — reduce those costs. These commitments are real, but they’re probably not going to stop a company from even putting a coal plant online if they know they can get AGI first. If you care about climate change and decarbonizing, our job is to figure out how to make that happen as fast as possible.

Ben Della Rocca: I agree that making these energy sources affordable is the best way to ensure they’re adopted. The related piece is making sure that the timeline to actually permit them and bring them online is efficient and fast as well.

In some cases, if clean energy technologies or emerging clean energy technologies can be brought online more quickly than other, less clean sources — if the permitting timelines are actually shorter for those technologies — that can provide strong incentives for industries such as AI to choose the faster route. There’s a large financial premium they could earn from bringing their AI models online and operational six to twelve months earlier.

Jordan Schneider: Any final thoughts, Ben?

Ben Della Rocca: There was a lot of work in the last administration to set forth actions that will address AI’s energy needs, and we’ve discussed many potential ways forward in this conversation.

A linchpin to making all of this successful is effective implementation and really prioritizing this work within federal agencies. Ensuring that people are focused on completing this work effectively, fully, and quickly, and making sure that the work starts on time and proceeds according to schedule is going to be extremely important.

Tim and Arnab, in your paper, one of your recommendations was for the White House to have an AI infrastructure czar of sorts to oversee and spearhead this work. This work is ultimately very complex and interdisciplinary. It’s not just a national security challenge — it also includes energy policy and law, and environmental permitting law. It will require strong leadership from the White House and the federal government to make sure that things happen as envisioned.

To underscore a simple but important point — the implementation side of this really matters.

Mood Music

We’re doing a lot more stuff than we used to! To make sure you don’t miss out on our best content, we’ve decided to start publishing monthly roundups.

Chips and AI

**China’s Weird Chip Surplus, Explained

This article examines China’s paradoxical state of AI compute, where reports of both shortages and overcapacity exist simultaneously. China acquired and produced over a million AI chips in 2024, but faces inefficiencies in their deployment. At the same time, a slowdown in foundation-model training has reduced immediate demand, creating a temporary surplus that is expected to reverse as inference demand rises and China pursues the next generation of AI models. The article synthesizes insights from Chinese media, outlining how state-owned telecoms, tech giants, and local governments are investing in compute, the challenges of building true high-performance clusters, and the Chinese government’s recent moves to curb waste and consolidate AI infrastructure. It also assesses the role of Huawei’s domestically produced chips.

DeepSeek: what it means and what happens next

This interview with Kevin Xu, a former GitHub and Obama Press Office staffer, explores DeepSeek’s unique organizational structure, approach to talent, its parallels with OpenAI, and the broader tensions between open-source collaboration and nationalistic technological competition.

Anthropic's Dario Amodei on AI Competition

We interviewed the CEO of Anthropic! In this podcast, Dario discusses his essay “Machines of Loving Grace,” and advocates for stricter export controls to widen the US-China capability gap and ensure AI systems are developed responsibly. The conversation covers DeepSeek’s unexpected advancements, concerns over AI espionage and model distillation, and the technical and policy challenges of preventing China from scaling up to US levels. Amodei also reflects on AI’s potential impact on democracy, suggesting that AI can be used to strengthen governance and public deliberation rather than simply entrenching autocracies.

DeepSeek’s New Burdens: Does DeepSeek need Beijing, or does it need Beijing to stay out of the way?

With the help of an anonymous contributor, this article speculates on the challenges DeepSeek faces as it attracts attention internationally and from the Chinese government. With growing compute needs, we consider whether DeepSeek could partner with a hyperscaler like ByteDance, and how that would shift its culture and business model. Since then, new reports have emerged verifying that DeepSeek is considering such a partnership as well as weighing the benefits of outside fundraising. This essay also considers the extreme case where DeepSeek receives “national champion” status, gaining state funding, privileged data access, and procurement deals — but at the risk of government interference, loss of autonomy, and geopolitical backlash.

SemiAnalysis + Asianometry on the AI Mandate of Heaven

In this podcast, Jordan has way too much fun making an AI tier list, speculating on Tim Cook’s successor, and discussing Google’s $75 billion AI investment with Dylan Patel and Doug O’Laughlin from SemiAnalysis and Jon from Asianometry.

Watch below on YouTube or listen now on iTunes, Spotify, or your favorite podcast app.

China

Friday Bites: China on Dario, DeepSeeking Truth, Ali + DeepSeek, and a Procurement IP Manifesto

This grab bag features translated Chinese responses to our interview with Dario, as well as Alex Colville’s excellent report about how DeepSeek subtly propagandizes even when the model is theoretically uncensored. Other features include how to use DeepSeek to research China, rumors of hyperscaler partnerships, and an anonymous procurement reform call to arms.

Xi’s Hard Tech Avengers

This article guides you through the less-well known guests who attended Xi’s business symposium, with profiles of the propaganda czars, Politburo tech strategists, and neoauthoritarian academics as well as tycoons of semiconductors, chemicals, agriculture, and 3D printing.

US Policy

**Strategic Ambiguity vs. Clarity

We took a break from DeepSeek mania to publish Republic of ChinaTalk editor Nicholas Welch’s comprehensive guide to the debate on Taiwan policy. It condenses arguments from 50 op-eds and academic papers into 12 key points, with five supporting strategic clarity (arguing it deters China, reassures allies, and aligns policy with reality) and seven advocating for continued ambiguity (which proponents say maintains flexibility, prevents provocation, and keeps Taiwan focused on self-defense). The article highlights areas where more data is needed to evaluate these arguments, and highlights historical parallels to past U.S. foreign policy missteps.

American Power in the Age of Economic Warfare

In this interview, Eddie Fishman explores the evolution of American economic warfare and the effectiveness of sanctions as a geopolitical tool. Fishman, a former State Department official, traces the transformation of U.S. sanctions policy, from the failures in Iraq and Cuba to the unprecedented success of Iran’s financial isolation and the mixed results against Russia. He highlights how globalization turned the U.S. dollar into a powerful chokepoint, leveraged through secondary sanctions and financial penalties to pressure adversaries. The discussion also covers the institutional dysfunctions that hinder swift action, the challenges of enforcing export controls on China, and the need for a centralized agency to handle economic statecraft.

The NSF, Seriously? + AI Safety's Death

In this piece, Jordan critiques the Trump administration’s cuts to basic science funding, particularly its impact on the NSF’s Technology, Innovation, and Partnerships (TIP) directorate, which was designed to drive translational research and bolster U.S. competitiveness. Jordan argues that gutting forward-thinking programs like TIP is a strategic mistake that undermines national security, and suggests that meaningful budget cuts should target bloated defense programs instead.

Trade: How Free is Too Free?

We released a transcript of our 2019 podcast with trade historian Doug Irwin in light of Trump’s tariffs on China, Canada, and Mexico. This interview explores the long history of U.S. trade policy, tracing its evolution from the revenue-focused tariffs of the early republic to the restrictive protectionism of the late 19th and early 20th centuries, followed by the reciprocal trade agreements that defined postwar globalization. The conversation also examines the rise of executive power in trade policy, the role of institutions like the WTO, and the shifting landscape of U.S.-China economic relations.

Gridlocked: Transformer Shortage Choking US Supply Chains

Lily Ottinger and Caleb Harding reported about the transformer shortage in the U.S. and its implications for industrial policy, national security, and economic growth. Transformers, essential for electricity distribution and infrastructure projects, face soaring demand due to grid aging, electrification trends, and AI-driven energy needs, yet domestic supply meets only 20% of demand. Supply chain disruptions, labor shortages, lack of standardization, and reliance on foreign manufacturers — particularly China — exacerbate delays, raising concerns over security vulnerabilities from potential cyber or physical attacks. The report explores solutions such as expanding domestic steel production, incentivizing transformer manufacturing, standardizing designs, and aggregating a strategic transformer reserve.

Innovation Emergency with Trump 1.0's Patent Director

This interview with former USPTO Director Andrei Iancu explores the role of patents in fostering innovation, the increasing dominance of China’s IP system, and the future of U.S. patent policy under the Trump administration. Iancu highlights how congressional inaction causes outdated IP laws to struggle in the face of emerging technologies, while China’s streamlined patent process and aggressive IP strategy have given it a competitive edge. He warns that weak patent protections in the U.S. risk stifling investment in high-risk, high-reward industries, potentially ceding technological leadership to China.

In the trades…

In the wake of DeepSeek mania, ChinaTalk earned shoutouts in the The FT, NY Magazine, Al Jazeera, and Foreign Policy (which called our coverage ‘savvy’). Jordan also appeared on NPR’s Here & Now, The Circuit and Hard Fork to talk about China’s AI development.

Mood Music

Today we’re doing an anonymous Q&A with KL Divergence, a robotics PhD currently in industry working on humanoid robots. For an introduction to humanoid robotics in China, see our article here, and for a deeper look into who’s leading China’s humanoid market, see our latest translated interview with the CEO of Unitree Robotics.

This Q&A covers:

• When and how AI-driven robotics will reach a tipping point in viability,

• Challenges and solutions for collecting data to build a robotics AI model,

• Successful strategies for companies to compete in humanoid robotics.

When will AI + robotics reach a tipping point in viability?

This is extremely difficult to predict.

Here's my non-answer: whenever the world achieves a data flywheel for robotics, i.e. accumulate a dataset large-enough (and algorithms to use it) that allows some robots to achieve a diverse set of somewhat useful tasks, with enough reliability that people allow those robots to operate in their factories, logistics centers, homes, offices, etc.

Once a robot has a “reason for being” in a space, and works well enough, the data flywheel will spin, and the robots will get better and better. This is the same process we are seeing play out in self-driving cars, and why Waymo’s early advantage in deployment is such a big deal, making Tesla dance and move forward plans for Robotaxi. I don't think we will see this happen all at once, in all application domains.

Today, this has already arguably happened for a specific application: robotic picking/packing in e-commerce logistics. Amazon, Dexterity, Covariant, Berkshire Grey, and Ocado all have massive robotics datasets for this basic task, and already use them to create their own “flywheel.” This is short of what we want though, because that data is only for one task, and is specific to those companies' unique robots.

What’s the pathway to viability? How will AI + robotics diffuse through different industries?

So the next stage will likely be doing lots of different tasks (10s-100s) in a structured environment. I would guess logistics centers and manufacturing. I think this could reasonably be achieved in 2-3 years research-wise, and 5-7 years to become commercially commonplace. Along the tail end of that period, you might see these robots start to appear in retail, hospitality, and food service back-of-house. Think: robots doing laundry or restocking shelves. Next, offices. And last, homes.

Will we be seeing AI-driven robots in homes?

We’re 10+ years away even as a question of research, if we ever get there.

Homes (and to some extent offices) are much more difficult than commercial/industrial spaces because of 4 factors: lack of structure and wide variation, safety, and cost.

• Structure and variation: Homes are the ultimate “unstructured” environment. They come in infinite variations, and change from moment to moment as people and stuff move around. One day you might decide to put the cucumbers in the top vegetable drawer, then next you might move them into the bottom drawer. Multiply that by everything a home robot might have to ever interact with and the amount of variation becomes mind-boggling. It is impossible to create a system which quantifies and anticipates it all explicitly. The realization has been the impetus for the move towards learned — rather than programmed — robotics AI systems over the last 10 years.

• Safety: It’s an engineering achievement to make a robot that can complete tasks and weigh only as much as a smallish human. If that thing falls in a house (dead battery, malfunction, etc.), the stakes are high: it might fall on a pet, break a glass table, or knock over a candle. Contrast with a controlled commercial environment, where people working near the robots can be specifically trained, the environment is arranged so that failures don’t lead to catastrophic danger, and the robots might even be cordoned off behind a cage to minimize the impact of accidents.

• Cost: Most proposals for robots in the home have them providing typical domestic labor: cooking, cleaning, tidying up, etc. People already pay for these services in their homes, and it is invariably some of the lowest-paid work in any economy. A humanoid robot has a similar part count and manufacturing complexity to an electric car, so it’s intuitive that the most optimistic cost estimates land the price of these machines at similar numbers: $15-50k, depending on the source. How much would a home robot have to do for a family to justify a $25k price tag with a 5 year life span, assuming no recurring service or subscription costs?

So why do we see some players in the robotics/AI space, humanoid or otherwise, proudly touting their goal of putting robots in homes? My best guess is that it’s a more compelling narrative to attract investment and inspire talent — and it’s not too hard to pivot back to industrial robots anyway.

What are the challenges of getting good training/testing data for AI-driven robots?

You already answered this well in your piece. It's because we need to get robots into the real world to collect lots of good data, but they currently don't work well, can be unsafe, take up space, etc. They have no economic reason to take up space and human attention where you would want them to collect data.

How are proposed solutions addressing these data challenges?

Ways of addressing:

1. Simulation (but this has flaws, as you mentioned)

2. Spend a lot of up-front capital to collect robot data directly, in hopes of collecting enough up-front to bootstrap a useful robotics foundation model (i.e. vision-language-action model or VLA)

3. Find a way to re-use data from the internet, e.g. watching human cooking or furniture assembly videos from youtube (this is very active research, but so far the results have been disappointing),

4. Master one task at a time (using a combination of 1 and 2, and old fashioned engineering), and hope you collect a dataset diverse-enough before you run out of money (if the dataset is not diverse in tasks, you will have a robot which can do a handful of tasks, but is expensive to train to do new tasks).

What's it like on the ground for a factory collecting this data?

This can vary greatly, but typically you have a task in a factory which has already been defined for a human worker, and it's fairly repetitive. E.g. Tesla's first application is to make Optimus, to grab battery cells rejected into a slide coming out of a battery quality control machine, slot many of them into a grid on a purpose-shaped tote, then walk those totes to a different area of the factory when full. It's very simple and repetitive, and today it's done by humans across dozens of machines. You can imagine other scenarios. For example, sorting packages into bins bound for different geographies in a logistics center. Well-defined tasks and lots of existing automation and built environment (e.g. screens, conveyor belts, well-placed bins and racks) to help humans.

What does it look like to collect the data? That depends on the approach. The most straightforward is teleoperation: a human dons a (usually) VR headset, special gloves for capturing finger movements, and other hardware you'd see in a VR space, and uses them to control the robot directly to do the job. This is “robot-in-the-loop.” It’s slow (the human can't move the robot as fast as they would themselves), and costly (it's actually more expensive than having a human do the job, because it's slower).

Another approach is motion capture: via various methods (camera systems in the work area, body-worn suits, even lightweight worn exoskeletons), we can capture the motion of humans who are already doing the job. This is more speculative, as it’s a difficult research problem to turn these motion recordings into instructions for the robot to achieve the task later.

The last major approach is simulation: usually through the help of a skilled human artist or engineer, create a detailed and functional 3D graphics simulation of the real environment in which the robot is supposed to perform. This allows us to use teleoperation, programmed routines, and reinforcement learning, to control the robot in simulation and collect data on its successes and failures. The weakness of this approach is that the model usually cannot be used immediately on the real robot, because it’s extremely difficult to capture all of the important behavior of a real work task, even a very small one, in a simulation. Roboticists refer to this problem as the simulation-to-reality (sim2real) gap.

On the research horizon, there are a variety of approaches that may allow us to generate or make use of data without actual or simulated robots. A “holy grail” of robot learning for the past decade or more, has been to create a robot learning system which can “learn from watching YouTube videos.” What all of these approaches have in common is that they seek to lower the cost of data for robotics models, by finding ways to make use of lower-quality data (i.e. weaker supervision). The key missing technical piece in most of these approaches is to find a way to map from non-robot behavior in one environment to the actions a robot would take to do the same task in a new environment.

What would indicate a successful humanoid robotics strategy?

How many robots does the competitor have in the real world doing tasks and collecting data, and (importantly) how diverse is that set of tasks? Humanoids are a very expensive way to automate just one thing, so the investment needs to be amortized across many different jobs.

What strategies are robotics firms taking to compete in the market? What will determine who succeeds?

Boy, this is a big question. I won't try to answer the whole thing, but I'll give you a framework.

There are a few fundamental assets to look at here: technical talent (people), chips (compute), robots (how much do they cost? what are their capabilities?), data, and distribution (customer relationships, pilots). Any robotics+AI company or partnership effort needs to assemble all of these ingredients to be successful.

Resources which are less scarce:

• Robots: you have a whole article about how China is commodifying robots. However not everyone agrees that *good* robots will be so plentiful (perhaps because of protectionism), and others (e.g. Figure, Boston Dynamics) believe they can create an edge by having the *best* hardware.

• Customer relationships: Tech demo deals like the Figure-BMW, Apptronik-Mercedes, and Agility-Amazon partnerships are very low-risk for the larger company and easy to make. CEOs at humanoid companies tell me they have no problem getting 100s of leads.

So a successful strategist will try to gain an edge in the scarcest resources: talent, chips, robots, and data.

• Chips: notably — every major NA effort has decided that they need to team up with a giant foundation model provider to have the chips and frontier models to compete. Figure-OpenAI, NVIDIA is in-house, Tesla-xAI, and Boston Dynamics has teamed up with TRI's foundation model team.

• Robots: Most people attempting to make their own, however Skild, NVIDIA, and Physical Intelligence have all taken a partnering or purchasing approach for robots instead. Whether a competitor sees robots as a competitive advantage, or an expense, is a major dividing line in strategy in this area.

• Talent: immensely cut-throat. Until very recently robotics+AI was a very niche field. An investor told me he believes there are ~25 people in the world who could lead one of these companies well. Even below leadership, the number of people with any training at all in this subfield is in the low 100s. The best-paying outfits in the world with the best reputations take 6 months to hire someone, and are often just waiting for new PhDs to graduate to fill positions. Talent with <1 year of professional experience but relevant education (usually a PhD) can fetch $500k-1M/yr in this field, and/or significant equity, depending on the size of the company.

And finally, data is the most strategic asset these companies seek to accumulate long-term. At the end of the day, the firm who has the best data (or best strategy for getting it) wins the game.

Hangzhou-based Unitree Robotics is among the top players in China’s robotics industry, developing best-selling civilian quadruped and humanoid robots.

Unitree’s H1 humanoids captivated over a billion people with a traditional folk dance performance during China’s 2025 Spring Festival Gala. Just weeks later, Unitree’s CEO was the youngest front-row participant in Beijing’s highly-anticipated private sector summit.

To better understand the Chinese robotics industry and where it’s headed, we’ve translated and annotated qan interview with Unitree’s founder and CEO, Wang Xingxing 王兴兴. Originally conducted in April 2024 by Titanium Media (TMT), the interview covers:

• Why LLMs aren’t enough for the robotics industry, and why Wang predicts the emergence of a large-scale AI model for general-purpose robotics by the end of 2025,

• Factors driving the global humanoid robot boom, and why China is uniquely poised to succeed in this industry,

• The techno-optimist vision for the economy of the future, powered by humanoid robots as well as machines of alternative forms,

• The timeline for mass adoption of AI-powered general-purpose robots,

• Unitree’s strategy for competing against foreign and domestic robotics firms.

We’ve added some editorial notes for your enjoyment, including commentary by anonymous robotics PhD and current industry player KL Divergence.

Interested in learning more? For past ChinaTalk coverage, see Angela’s work on China’s leap into industrial robotics and China’s humanoid robot industry.

The Translation

Original Article | Archive | Title: “Dialogue with Wang Xingxing: Humanoid Robots Will Reshape All Industries Within My Lifetime” | Author: Rao Xiangyu 饶翔宇 for Titanium Media (TMT) 钛媒体APP | Editor: Zhong Yi 钟毅

On February 17, 2025, a highly anticipated private enterprise symposium was held in Beijing.

At the event, the presence of Wang Xingxing, the founder of Unitree Robotics and a post-90s entrepreneur, attracted market attention. Wang was seated in the front row among the business representatives, alongside industry giants such as Zeng Yuqun, Jack Ma, Ren Zhengfei, Wang Chuanfu, Lei Jun, and Pony Ma. Among them, Ren Zhengfei, Wang Chuanfu, Liu Yonghao, Yu Renrong, Wang Xingxing, and Lei Jun delivered keynote speeches.

As a startup company, Unitree Robotics and Wang Xingxing have experienced what can be described as a “rocket-like leap” in growth.

[KL Divergence: Perhaps. Though consider that Unitree has been an established player, particularly in quadruped robots, for 10 years. They have worked hard and scraped their way up inch by inch by serving the small markets which existed for their product (mostly researchers). Their recent prominence has come from riding the wave of interest in humanoid robots by creating low-cost, easy-to-use, but not particularly advanced or capable, humanoid robots for research.]

Public records show that Wang Xingxing earned his bachelor's degree from Zhejiang Sci-Tech University. Due to poor English proficiency, he failed to gain admission to Zhejiang University for his master's studies and was instead placed at Shanghai University. In an interview, Wang once mentioned, “During my three years in high school, I only passed English exams three times in total.”

From 2013 to 2015, while pursuing his graduate studies, Wang, despite having limited resources and funding, independently designed hardware and control algorithms and combined them with industrial motors to develop the robotic dog XDog. This project won second prize in the Shanghai Robotics Design Competition. After graduating, Wang embarked on an entrepreneurial journey focused on robotic dogs.

[KL Divergence: Actually, much of the hardware and control algorithms were from publicly-available robot and actuator designs published by Western researchers, such as Sangbae Kim (MIT) and Dan Koditschek (UPenn). What Unitree really excelled at was (1) iterating high-performance actuators, robot designs, and research-grade control algorithms, and (2) leveraging the Chinese supply chain to create low-cost, high-performance, highly-reliable combinations of these key technologies. In other words, they productized the research.]

Unitree Robotics, founded in 2016, initially specialized in the development of quadruped robotic dogs and successfully sold them worldwide, becoming one of the leading players in the industry in terms of product shipments. By 2023, the company ventured into humanoid robotics and quickly became one of the most closely watched companies in the field. In 2025, Unitree Robotics’ latest humanoid robot appeared on the stage of the CCTV Spring Festival Gala, garnering widespread public attention.

In April 2024, Wang Xingxing, the founder of Unitree Robotics, had an exclusive interview with Titanium Media APP. The relevant content can be found below.

Wang told TMT APP that the fundamental reason behind the humanoid robotics boom is the emergence of large AI models. Previously, it would take one to two years for a humanoid robot to learn to walk, but now, with AI algorithms, this can be achieved in just one month.

Regarding the future development of humanoid robots, Wang expressed strong optimism. He believes that by the end of 2025, at least one company worldwide will have developed a general-purpose robotic AI model. This foundational model, he explained, is like a complete set of building blocks, with large language models being just one piece. Other crucial components include visual perception, tactile sensing, decision-making, and interaction systems.

Looking at an even longer timeline, Wang told TMT, “Within our lifetime, humanoid robots will be able to revolutionize every industry, from manufacturing and agriculture to services and industrial sectors. Taking it a step further, governments could potentially deploy 100,000 humanoid robots to construct an entire city. On a smaller scale, robots could even shrink down to the size of cells, transforming all aspects of our natural environment.”

Below is the full interview between TMT and Unitree Robotics founder Wang Xingxing, with slight editorial adjustments.

“The Turning Point for Humanoid Robots Has Not Yet Arrived”

TMT: A few days before our meeting, Boston Dynamics, a star company in the robotics field, announced that its hydraulic-powered humanoid robot would be phased out, and future developments would focus on electric-powered products. What are your thoughts on this?

Wang Xingxing: Boston Dynamics has been making robots for many years, and they’ve also been working on commercialization for a long time.

As for hydraulic drive systems, I had already believed before 2013 that this approach could not be commercialized. The reason is simple: it relies entirely on precision mechanical components, and once you involve such components, costs will never come down. Moreover, all hydraulic systems leak oil. That’s why you hardly see hydraulic systems in consumer vehicles anymore — they’ve all been replaced by electric drive systems.

So, if Boston Dynamics wants to continue developing humanoid robots, switching to electric drive is definitely the right path. The only surprising thing to me is that I assumed around 2018 that they had already started working on an electric version. But later, when they had made no detectable progress, I just stopped paying attention.

TMT: Compared to hydraulic systems, is electric drive better suited for large AI models?

Wang Xingxing: Compared to hydraulic systems, electric drive is all advantages and no disadvantages. As for whether electric systems are better suited for AI, that’s harder to say. However, electric drives have lower production costs, offer greater motion flexibility, are safer, and also lighter in weight.

TMT: Now that Boston Dynamics has switched to electric-powered robots, combined with their existing training data, do you think they could iterate faster than competitors in this new wave of competition?

Wang Xingxing: It’s hard for me to say. However, we remain quite confident, because we’ve been working on quadruped robots for many years, and a lot of the algorithms and components we’ve developed can be directly applied to humanoid robots.

Another important point to note is that most of the top AI talent in the U.S. isn't at Boston Dynamics—they’re at Google, NVIDIA, and OpenAI. Boston Dynamics' strength likely lies in hardware development and traditional humanoid robot control systems.

TMT: So, would you say that the emergence of large AI models is a major turning point for humanoid robots?

Wang Xingxing: I don’t think we’ve reached that turning point yet.

Right now, it’s more like a starting direction. There's a common misconception—many people think that large language models like ChatGPT can be directly applied to robots, but in reality, that’s not the case.

TMT: Why not?

Wang Xingxing: Because LLMs aren’t designed for robotics in the first place. ChatGPT operates purely on text logic, and its entire training dataset is based on text data. It doesn’t perform well in robotic environmental perception — this is a global challenge, not just a problem for one company.

While the humanoid robotics industry does use AI, the technology is actually very different from large model technology.

TMT: But some companies have claimed that large AI models can already recognize different types of plates, allowing robots to identify and pick them up.

Wang Xingxing: That’s not something we can verify. It was just a video, and no one has confirmed its authenticity.

Besides, there’s no data proving that if you swap the plate for an apple, a pear, or something else, the robot would still be able to recognize and handle it correctly. Personally, I don’t see any evidence of real technical breakthroughs coming from Silicon Valley — it still seems quite conventional (中规中矩).

TMT: So, large AI models are not the key turning point for humanoid robot development? Are they less important than people think?

Wang Xingxing: The models themselves are not important for robots, but the underlying technological direction they represent is very important.

Right now, large models are mainly focused on language models — but no one has yet developed a true large-scale model specifically for robotics.

TMT: That brings us to the big question — what triggered the humanoid robotics startup boom in 2023?

Wang Xingxing: The reason is really quite simple: Tesla started working on humanoid robots.

Elon Musk has already disrupted industries like automobiles and rockets, growing them into massive sectors. Now that he's entering humanoid robotics, governments and various institutions want to get started early, rather than waiting for Musk to succeed first and then trying to catch up.

[KL Divergence: I think this is a little bit just-so and playing to the audience a bit too much. The fundamentals are more important. Elon and Optimus is definitely the spark which ignited the wildfire. But the kindling was years and years of slow and steady progress on batteries and electric motor power density made it finally possible to create practical (as in, strong and light enough) electric humanoid robots, around 2020-21. Elon's team caught on to this a little early, because these are also technologies that Tesla happens to to be deeply invested in. But others were doing it already, just quietly.]

At the same time, ChatGPT and other LLMs have expanded the public’s imagination of AI’s potential. You could say these models ignited excitement and enthusiasm across the industry. Right now, what we’re seeing is just the beginning — the momentum will only grow stronger.

As hardware and AI technology advance each year, the impact of humanoid robotics on the world will be massive and transformative.

“It’s simple, really not as complicated as most people think”

TMT: Current large AI models are just the beginning. What's the future direction of the industry or where should everyone's efforts be focused?

Wang Xingxing: There are many directions. The first step is adapting AI for robots - developing robotic vision, perception, understanding, execution planning, and various operations.

I'm excited just like everyone else. I personally feel this industry will develop rapidly, including robots, large models, and AI. I believe by the end of 2025, at least one company globally will develop a relatively general-purpose robot large model.

Our company hopes to develop it ourselves, but realistically speaking, the probability is higher that an American company will achieve it first.

[Angela: Wang is optimistic. Depending on what goalposts you set, training a robot “foundation model” requires large, multimodal datasets that take time and capital to collect or synthesize – including vision, sound, touch, motion, social and environmental interaction, and so on. In some ways, Wang’s prediction aligns well with the Chinese government’s stated goal of mass production of humanoids by 2025 and world leadership by 2027. At the same time, he realistically recognizes the strength of US innovation. The main takeaway here is that to Wang — and likely to many of his industry counterparts — the humanoid robot race is accelerating towards some decisive moments.]

TMT: So that brings up the question of open source versus closed source.

Wang Xingxing: If we develop it, it definitely won't be open source.

TMT: Is there a unified model between robot large models and robot dogs?

Wang Xingxing: Most robot dogs are implemented through reinforcement learning, which is a relatively mature technology.

Robotic large models or robot world models can be applied to all robots, not necessarily humanoid or dog-shaped ones - they're universal tools. I've always believed that robots don't necessarily have to be humanoid; the humanoid shape is just one of many possible forms. I've never insisted they must be humanoid.

TMT: But the mainstream view is that humanoid forms are better because our whole society is built for human-shaped frames.

Wang Xingxing: They might like to say that, but I've never believed it.

You can build entirely new physical worlds. Why would you need a humanoid form for mining? Why would you necessarily need a human shape for building houses? Of course, humanoid forms are important, or relatively important, but they're not everything.

For example, at home, people might prefer humanoid robots for performing scenarios or accompanying you on trips. But for building houses or transporting things - physical labor - there's no need for them to be humanoid. Plus, humanoid forms might give people a sense of owning a slave if you make them do unpleasant work, making their owners feel uncomfortable.

TMT: Would you feel sorry for them?

Wang Xingxing: Current AI hasn't reached that level yet; it can’t perceive such things.

But if its AI could perceive pain or negative emotions, then yes, that might be problematic. But there’s no need to feel sorry now, because it’s still just an inanimate object [死物, literally a ‘dead thing’] with limited intelligence.

TMT: I'm curious about something — even though their intelligence is limited, when you push them, why do they display human-like staggering movements?

Wang Xingxing: Because that’s what the AI was trained to do through reinforcement learning.

TMT: So it’s imitating human behavior?

Wang Xingxing: Some behaviors aren't imitation; they’re determined by natural laws. You could say physical laws constrain these robots to move in certain ways. If an alien had a human shape, its movement would probably be the same as well.

TMT: Currently people break down robots into cerebrum, cerebellum, and the physical body. What's your view on this?

Wang Xingxing: I’ve never liked separating the cerebrum and cerebellum so distinctly. One model is enough - why divide it into two? I don't think it's necessary.

Of course, there might be various modules within the model, but overall I prefer treating it as a single model. From walking to fine-grained operations, we implement everything using AI in a completely end-to-end manner. From visual perception to leg execution, one model handles it all - no intermediate mathematical formulas whatsoever.

TMT: Can the hardware capabilities keep up?

Wang Xingxing: For robots, it's just a few joints — it’s really not that hard. Just sensors feeding into the model, and then the model outputs to the joints. That's all.

TMT: Your understanding of humanoid robots seems simpler than others'.

Wang Xingxing: It is simple, not that complicated.

TMT: For example, others might think dexterous hands are difficult for fine operations because they require more accurate recognition and finer motion control.

Wang Xingxing: It's very difficult if you use traditional technology, so you can't rely on traditional approaches. Without technological innovation, there's no point in working in this field. Of course, you can't express it so directly — better not to go too far beyond public understanding, otherwise I'd probably get cursed out (骂死).

TMT: What specifically do you mean by non-traditional?

Wang Xingxing: It's new AI, end-to-end. It means not having to manually write lots of software programming rules in between, nor perform traditional image recognition.

TMT: How do you implement that?

Wang Xingxing: Modify the model. The underlying AI is the same, but your entire model structure and algorithms are different. I can't explain this too specifically - it would be hard to understand. For example, you don't need traditional image annotation or image understanding at all. You can input images and videos into a model, and the output is directly the robot's joint trajectories, then you just train it. You can still do image annotation, like labeling images of apples. But annotation has only one function: interacting with humans, helping it better understand people. For the robots themselves, there's no difference between an apple and a pear.

TMT: Compared with the mainstream opinions, your logic and industry judgments are unique.

Wang Xingxing: The mainstream viewpoint still has many issues. As a startup, if your thinking is just mainstream, it just won’t work out well for you. You must see the development direction for the next few years, and once you see it, plan ahead accordingly - then you're certain to win, or at least not lose. If you only see what everyone else is talking about, others can certainly do better than you - how could you stand out?

TMT: In your view, what will the next few years look like?

Wang Xingxing: I can't get too specific, but what's certain is that the industry will progress extremely rapidly.

TMT: How fast are we talking here?

Wang Xingxing: It's basically beyond imagination. The current pace of AI deployment in factories — globally, technological progress is extremely fast and has almost proven viable.

TMT: Currently, no company can fully utilize robots for work.

Wang Xingxing: But the entire logic has almost been proven. This doesn't mean robots can do everything, but work-capable, end-to-end robots are nearing maturity. A more general-purpose robot model will likely be developed by a company globally before the end of 2025.

TMT: That fast?

Wang Xingxing: It could be even faster. Some people have already seen where this is going - though it sounds a bit boastful, I feel I've seen it too. Following this direction, with some additional time, manpower, and money, it can basically be developed.

“All technological breakthroughs have a large element of luck"

TMT: What specifically does a robot model refer to?

Wang Xingxing: You can think of it this way: first, it has strong mobility capabilities applicable to most terrains, possibly with some mobility skills exceeding humans. For instance, its obstacle-crossing ability, speed, jumping ability might be better than humans. Another aspect is working in factories - it can do many tasks without requiring manual programming. Through large model capabilities, with just a little teaching, it can learn by itself and then perform well.

TMT: Is simulation training in virtual environments still necessary?

Wang Xingxing: Probably not all that necessary. Once you've trained it well and validated it, you don't really need simulation anymore. Of course, completing the hardware won’t happen right away, but I think that's just a matter of time. As for AI, there's still some uncertainty. Although I just said I'm personally optimistic it will emerge before the end of 2025, it might not happen - it could take 3-5 years before it's developed. It depends on humanity's collective luck - sometimes it just comes down to luck.

TMT: How do you understand this kind of luck?

Wang Xingxing: Many technological breakthroughs depend on luck. For example, if Einstein hadn't existed, someone else would probably have discovered his theories. But it might have been delayed by several years, or even decades. You can consider that all technological breakthroughs have a large element of luck involved.

TMT: Another point: besides algorithms and models, large models need data. Is data collection currently very difficult?

Wang Xingxing: There are indeed many things that need to be done, but there are methods for addressing them. It's not as complicated as people think - many problems aren't as complex as people imagine. You know, in all current technology fields, if you really look, there's nothing truly complex; everything is relatively direct and simple. Even

TMT: So is your industry also divided into two camps - optimistic and pessimistic? For example, you're more optimistic, thinking the whole thing isn't that difficult.

Wang Xingxing: It definitely requires time and intellectual investment, but these are things that can be solved and advanced. They’re not like room-temperature superconductors or controlled nuclear fusion. The biggest problem with room-temperature superconductors and controlled nuclear fusion is that there’s a question mark over whether they’re physically possible. The universe might simply not allow such things to exist, and humans might never achieve them no matter how much time and effort we invest. Artificial intelligence robots are common things, not something extraordinary — just the intelligence of a bunch of humans and animals. Intelligence is a widespread phenomenon. Some animals are very smart and can understand much of what humans say, they just can’t speak. And crows — some crows can even use tools directly. So, intelligence doesn’t have many limitations or physical constraints; it can be replicated.

TMT: What’s the biggest motivator for your work?

Wang Xingxing: To be honest, what moves me personally is AI.

A few years ago, an investor asked me whether our company would ever develop humanoid robots, and I told him, “We would never do it, even if it kills us.”

[KL Divergence: Great honesty here. It's true. Virtually the entire field considered humanoid robots a hopeless tarpit, which would consume all of your time and money and render not progress. Even in robotics research, humanoids were a quirky backwater reserved for the cranks and over-optimistic.]

Back then, humanoid robots were far too complex. Traditional algorithms simply couldn’t handle such intricate machines. The conventional approach to training humanoid robots relied on highly skilled engineers manually writing mathematical equations to model movement. These equations would then be solved to determine the robot’s motion trajectory. But this method had severe limitations—if the environment changed, the equations often became invalid, requiring entirely new models to be designed from scratch.

This approach also led to an overwhelming amount of code, and as the system grew more complex, it became nearly impossible to maintain manually. However, AI changes everything. As long as the model is well-structured and continuously fed with data and compute, AI can iteratively optimize itself through trial and error. By leveraging reinforcement learning and reward mechanisms, AI can automatically retain successful training outcomes and discard ineffective ones, dramatically improving training efficiency.

Recent progress in AI technology—both in capability and speed—has far exceeded my personal expectations. That’s why, despite having worked on humanoid robots for just over a year, our performance is already exceptionally good. The reason we’ve been able to move so quickly is simple: thanks to advancements in AI.

The benefit of AI is that once you’ve built a strong model, the rest is just a matter of compute—you don’t have to manually fine-tune everything. If you need to test a scenario, OK, all you need to do is feed the system more data. This is also why Tesla’s autonomous driving team is significantly smaller than Chinese autonomous driving teams. I know for a fact that Tesla’s team has only a few hundred people, whereas some companies in China have teams numbering in the thousands.

TMT: Is this also why newer players have been able to surpass Boston Dynamics?

Wang Xingxing: Exactly. If we were competing with Boston Dynamics purely using traditional algorithms, we wouldn’t stand a chance. The reason is simple: Boston Dynamics has an entire team of PhDs from MIT, and there’s no way China could outmatch them in that domain.

TMT: Looking ahead, what do you think will be the key differentiator among humanoid robots?

Wang Xingxing: Robotics is an integrated product. Unlike fuel-powered vs. electric vehicles, where the underlying technologies are fundamentally different, the differentiators in humanoid robots will be more incremental—primarily in specific engineering optimizations, such as motor scale, motor placement, workspace dimensions, structural design, and leg configurations.

The same principle applies to AI. Take large language models — they’re fundamentally pretty similar. The biggest points of differentiation are in the details rather than in fundamental design; OpenAI’s GPT architecture is still relatively clean.

“In our lifetime, humanoid robots can reinvent all industries and the natural environment.”

TMT: Commercialization is also important. How can startups survive in an increasingly competitive landscape?

Wang Xingxing: The business logic is very simple. As long as your product is better than your competitors’ in all dimensions, then you will profit. What remains is the question, how big is the entire market? Right now, our company has a strong market position, so we have captured most of the easily accessible revenue opportunities.

TMT: What do you mean by ‘easily earned revenue’?

Wang Xingxing: From having high shipment volumes. We sold quite a few quadruped and humanoid robots last year.

TMT: How many did you sell?

Wang Xingxing: It's hard to say exactly, but it's under a few hundred. However, we definitely sold the most in the domestic market.

TMT: Who bought them?

Wang Xingxing: A variety of buyers, including research institutions, AI companies, and businesses pursuing real-world applications.

TMT: How can you move so fast and manage to sell your products?

Wang Xingxing: Because we have a strong foundation. There’s significant overlap between robotic dogs and humanoid robots. Our company holds advantages in technical R&D, AI algorithms, manufacturing, and sales channels. We already have an established customer base and ready-to-market products. Other companies have to build everything from scratch, which takes time.

TMT: Is your revenue sufficient to support R&D?

Wang Xingxing: Our company maintains healthy gross profit margins, complemented by ongoing funding.

TMT: For humanoid robot startups, is the ability to secure funding a core advantage?

Wang Xingxing: It’s hard to judge the industry right now because it’s too hot. Many companies with basic foundations have raised some funds, which are at least enough to keep them afloat.

There’s certainly no shortage of funds in this industry. When we started, we were poor. Compared to back then, things are completely different. Now, some companies have been around for only a year and already have a valuation of 1 billion yuan. It's astonishing. The industry isn’t short on capital, and neither are they.

But I think that before the industry truly takes off, having too much money is pointless. It’s difficult to allocate effectively, and if spent indiscriminately, it could easily be wasted. At this stage, neither the technology nor the business models have been fully validated, so throwing money around wouldn’t be wise.

Take bike-sharing, for example. It worked because the business model made sense. Once that’s proven, the only thing left is scaling up, and there’s nothing left to do but pour in funding.

TMT: What do you mean when you say the technology and business model haven’t been fully validated?

Wang Xingxing: It means that neither the technical framework nor the commercialization strategy is fully developed. Even if you have the capital, you don’t necessarily know how to deploy it effectively.

TMT: What are the main technical challenges?

Wang Xingxing: For humanoid robots, the biggest question is how to integrate with AI models—we don’t have a definitive answer yet.

TMT: Another observation—most humanoid robotics entrepreneurs today are quite young. (Wang Xingxing is from the ‘90s generation.) Why is that?

Wang Xingxing: It’s simple. Older generations just aren’t as interested in this space. AI technology is evolving at an unprecedented pace and older knowledge is becoming outdated – knowledge of the technology we had five years ago is practically irrelevant. The younger generation is fastest at learning and applying the new advancements. Traditional internet startups had a low barrier to entry — basically anyone could become a product manager. But humanoid robotics isn’t a conventional industry.

[Angela: We’ve written before about how this generation of emerging technology creates space for young, enthusiastic talent to make an impact — DeepSeek is a good example of this. It would be interesting to know if, like DeepSeek, Unitree draws its success from China’s pool of homegrown talent. Wang Xingxing himself never studied or worked abroad.]

TMT: Earlier, you mentioned the potential for a breakthrough innovation. Were you referring to how humanoid robots and AI models can be integrated?

Wang Xingxing: Yes, more or less.

TMT: But aren’t AI models just modular components that can be put together like building blocks?

Wang Xingxing: The differences in AI models go far deeper than that. Take Transformer architectures, for example—there are still endless ways to optimize and refine them. Researchers are even exploring alternatives to Transformer-based models altogether. The AI field is full of opportunities for technical breakthroughs, and there’s still vast room for innovation.

I anticipate that by 2025, we’ll see a significantly improved AI model for general-purpose humanoid robots. When that happens, industry momentum will accelerate even further, to the point where companies from around the world try to enter.

TMT: At that point, do you think hardware or software will be the first to breakthrough?

Wang Xingxing: Software will be the key driver. No matter how advanced the hardware is, without the right software, it’s just an expensive pile of metal.

TMT: So given the current pace of development, as soon as the right software emerges, the hardware will be able to keep up?

Wang Xingxing: Absolutely. Hardware won’t be a bottleneck. If it’s really needed now, we can scale production quickly by aggressively deploying capital. If necessary, we could push manufacturing capacity to its limits — pay engineers 10 times their normal salary, work around the clock, and purchase all the necessary equipment. With sufficient investment, we could have mass production up and running in as little as a few months to a year.

TMT: How does China’s hardware capabilities compare to those of other countries?

Wang Xingxing: China has a significant edge in hardware. The cost-performance ratio is much higher.

TMT: Why is that?

Wang Xingxing: First, in the U.S., hardware development doesn’t receive as much attention—most of the top talent is focused on software. Second, manufacturing and labor costs in the U.S. are much higher than in China.

TMT: It seems like they are prioritizing software, while our strength lies in hardware.

Wang Xingxing: Exactly. Most major U.S. companies focus primarily on software. But at Unitree Robotics, we are developing both software and hardware, because maintaining competitiveness requires full-stack capabilities. As a relatively smaller company, we can’t afford to focus on just one domain. Large corporations can get away with specializing in software and outsourcing hardware, but for us, abandoning hardware development would be an unwise strategic move.

TMT: Why has robotic dog technology matured faster than humanoid robotics?

Wang Xingxing: One reason is that robotic dogs have been in development for a longer period, and their form is more stable. They don’t require complex dexterous manipulation, like grasping and handling objects.

Another key reason is that, in the past, there was a much larger community of developers working on robotic dogs, whereas today that number has declined. In AI, the maturity of a technology is often directly correlated with how many researchers are actively working on it.

For example, large language models have advanced much faster than AI for robotics simply because more people are involved in developing them. Ten years ago, computer vision—especially facial recognition—was in its golden age because so many researchers were working on it. Image-based AI took off because it was relatively easy to experiment with; all you needed was a decent computer.

But robotics is a different story. It requires hardware simulation and real-world testing, which makes it much harder for individuals to participate. That’s why the field has been slower to progress. However, as I mentioned earlier, the industry is now accelerating because a growing number of people are entering the space. More minds working on a problem naturally lead to faster breakthroughs.

TMT: Does Unitree Robotics have a clear product roadmap and timeline?

Wang Xingxing: We will launch new products every year.

TMT: What do you envision for Unitree’s next-generation robots?

Wang Xingxing: The next generation will undoubtedly surpass current models in every aspect—appearance, performance, AI capabilities, and more.

TMT: Can you give a specific example?

Wang Xingxing: Our goal is for humanoid robots to perform real industrial tasks—working in factories, assisting in production assembly, and handling logistics.

TMT: Do you have a release timeline for the next generation of robots?

Wang Xingxing: It’s not convenient to disclose at the moment.

TMT: Unitree Robotics has already completed eight rounds of funding. Do you expect fundraising to accelerate moving forward?

Wang Xingxing: I think we’ll be fine. As the industry gains more attention, we’re seeing increased interest from investors.

TMT: What do you think will be the ultimate future for humanoid robotics?

Wang Xingxing: In the future, humanoid robots could redefine entire industries—from manufacturing and services to agriculture, mining, and construction.

I imagine a distant future in which governments could deploy tens of thousands of humanoid robots to build entire cities from the ground up. At that point, infrastructure is fully automated, housing is provided at no cost, and people no longer need to work because robots sustain the entire economy. That’s entirely within the realm of possibility.

Also, right now when we talk about humanoid robots, we picture machines that are roughly human-sized. But in reality, humanoid robots could build smaller robots, and those smaller robots could build even smaller ones. This process could continue indefinitely, leading to robots at microscopic scales.

Eventually, we might see robots as small as biological cells. Who knows what’ll happen then? What we perceive as bacteria could actually be tiny robotic entities. The entire natural environment could be restructured from the ground up. When that happens, governments will need regulations to prevent unchecked proliferation, or these robots could consume resources uncontrollably.

[Angela: A very science-fiction vision indeed. But Wang’s fantasy of a robot economy resonates with Beijing’s investment in industrial robotics as a path for economic advancement. ChinaTalk will continue tracking such developments in robotics.]

TMT: Do you think we will see this level of technological advancement in our lifetime?

Wang Xingxing: Absolutely. The only missing piece is AI. Once AI breakthroughs happen, everything else will follow naturally.

This will fundamentally reshape the world. I’ve always believed that when we look back at today’s society after the emergence of general-purpose AI and humanoid robots, it will feel as distant and primitive as looking back at the Stone Age.

Announcing: the ChinaTalk book club! We have upcoming shows with the authors of To the Success of Our Hopeless Cause: The Many Lives of the Soviet Dissident Movement, To Run the World: The Kremlin’s Cold War Bid for Global Power, and Learning by Doing: The Real Connection between Innovation, Wages, and Wealth. We’d like to encourage you to read along with us in preparation for the shows!

Manus, a Wuhan-developed AI agent went viral this weekend. Guests Rohit Krishnan of Strange Loop Canon, Shawn Wang of Latent Space, and Dean Ball of Mercatus and Hyperdimensional join to discuss.

We get into:

• What Manus is and isn’t,

• How China’s product-focused approach to AI compares with innovation strategies in the West,

• How regional regulatory environments shape innovation globally,

• Why big AI labs struggle to build compelling consumer products,

• Challenges for mass adoption of AI agents, including political economy, liability concerns, and consumer trust issues.

Listen now on iTunes, Spotify, YouTube, or your favorite podcast app.

What is Manus?

Jordan Schneider: This past Friday, Monica — a startup founded in 2022 — launched a product called Manus. The launch was done through a video in English. Manus is ostensibly an AI agent that you tell to do something, and it can interact with the internet as if it were a person clicking around to book a restaurant, change a reservation, or potentially one day take over the world through Chrome. The rollout was remarkable, with hype building dramatically over the weekend. The product seems to be more competitive than similar offerings we’ve seen from OpenAI and Anthropic. With that context, Shawn, what were your first impressions of what Manus was able to build?

Shawn Wang: My first impression was that it’s a very well-executed OpenAI Operator competitor. It can effectively browse pages and execute commands for you. In side-by-side tests that people in the Latent Space community were running with Operator versus Manus, Manus consistently came out on top. This is backed by the benchmarks they hit on the Facebook Gaia benchmark, which evaluates agents in the real world [which is a public benchmark]. The product is very promising. I’m somewhat suspicious about how well the launch was executed with influencer-only invite codes and people writing breathless threads. We’ve seen this many times in the agent world, but this one people actually seem able to use, which is nice.

Rohit Krishnan: What interested me most was that all our previous conversations about China focused on models — how good their models are, how much money they have, how many GPUs they possess, etc. Now we’re talking about a product. The closest thing to a product from DeepSeek was their API, which is really good with an exceptional model, but the interface was just the same old chat interface. We’ve been discussing agents extensively for a long period. In the West, we still live under the umbrella of fear regarding AI agents, which is why most models aren’t given proper internet search capabilities. It’s amazing to see that the first really strong competitor has come out of China — arguably better, perhaps slightly worse, but definitely comparable to Operator. They made it work with a combination of Western and Chinese models, using Claude and fine-tunes of Qwen underneath. That changes the product landscape as far as I can see.

Dean Ball: I didn’t get an invite code myself but was able to use someone else’s account briefly. I ran it through my favorite computer use benchmark that I’ve organically discovered — trying to book a train ticket on amtrak.com. Operator consistently fails at this task, but Manus succeeded on its first attempt. That says something significant! Many other demos I’ve seen of the product seemed quite impressive and like things that would surprise me if I saw Operator doing them, based on my perception of Operator’s reliability and competence.

This isn’t a story about some shocking technological innovation or about DeepSeek’s unfathomable geniuses, as Jack Clark says, discovering some new truth about deep learning. This is good product execution in a style relatable to Y Combinator circa 2015. It’s a well-built product that works effectively, though it has flaws and glitches like all computer use agents do. What’s interesting is that this represents an advancement of capabilities. DeepSeek might be a more impressive technical achievement than R1 and V3 in some fundamental sense, but DeepSeek R1 wasn’t better than OpenAI o1 — at best, it was comparable. Manus appears better than what I’ve seen from Operator, or at minimum comparable but I think unambiguously better. Thinking about why that’s the case is a really interesting question.

Subscribe now

Jordan Schneider: Shawn, you just hosted a wonderful week-long AI agent conference in New York City. What’s your take on why no one in the West got to this first?

Shawn Wang: There are many skeptics of agents, even among the agent builders themselves. People range across the spectrum in terms of the levels of autonomy they’re trying to create. The general consensus is that lower levels of autonomy are more successful. Cursor being an agent is now worth $10 billion, whereas the people who worked on Baby AGI and Auto-GPT are no longer working on those projects.

Working on level four or five autonomy agents hasn’t been a good idea, while level one and two — more “lane assist” type autonomy agents — has been the better play. With the rise of reasoning models and improvements in Claude and other systems, that is changing every month. The first one to get there, like Manus, would reap the appropriate rewards.

Dean Ball: That intuition you’re describing is definitely something I’ve heard too, and it’s probably right. For a variety of reasons, I won’t be using Manus on a day-to-day basis. Part of that involves security concerns, but even without those concerns, I’m not sure this is a product I would use regularly compared to an agent like OpenAI Deep Research or a Cursor-style product. Those have much more genuine day-to-day utility.

As an investor in this company, I would be concerned that Manus will be, to use Sam Altman’s terminology, steamrolled by the next generation of computer use agents from the big labs. That’s very possible. From a practical business and technological perspective, this makes sense to me.

Rohit Krishnan: The key question I keep pondering is why Manus wasn’t built by a YC company six months ago. We’ve internalized the fear Dean talked about — that anything we build will get steamrolled by Sam Altman. In some ways, that’s correct. We all personally know code assist companies that emerged a couple of years ago and went bankrupt when the big labs effectively took over.

However, I have this heretical notion that despite everyone talking about agents, nobody at the large labs cares enough about them. They don’t seem interested in building products beyond making models smarter and letting them figure out products on their own. We’re stuck in this weird situation where I have access to every large model in the world, but half of them can’t do half the things because nobody has prioritized those capabilities.

O1 Pro can’t take in documents. O3 Mini couldn’t take in Python files or CSVs. Claude can’t search the web. These weird restrictions exist partly from AI safety concerns and partly because nobody has bothered to add these features.

One significant benefit of something like Manus is that people are actually trying to build useful agents for real-life tasks, like booking Amtrak tickets — which is a great evaluation benchmark. This pushes the labs or anyone else to say, “We should probably try to do this.” We can’t just throw up our hands and wait a year hoping the labs will build the next big thing.

The Western success story is effectively Perplexity — the one company that did what the labs would have been closest to doing but never did, and found success. Beyond that, when thinking about other agents we normally use, I can only realistically name Code Interpreter from a couple of years ago and Claude Code, which just released. Both are stripped-down versions that do a few things but still can’t handle basics like search.

When I look at Manus, what stands out isn’t just that they made an agent ecosystem work using external or combined models, which everyone expected would happen. More importantly, they actually went for it. There’s a price to pay — you have to try it. You can give it browser access, let it work for four hours, and get something useful back. Unlocking this capability is important from a product perspective.

Jordan Schneider: I’ll give one more perspective as well, which could be a fun US-centric observation. In the US, we’re very interested in B2B and developer tooling, especially in Silicon Valley. We really love developer tools, building for developers because we feel the pain. In China, there’s perhaps more B2C focus, which actually works to their benefit in terms of finding good use cases.

Share

Rohit Krishnan: What is the previous large software success story from China that took over the world? There’s TikTok, and that’s essentially it. WeChat is amazing, but nobody uses it outside China — maybe in parts of Southeast Asia. Banking software emerged, but nobody really adopted it. Alibaba Marketplaces exist, but they haven’t permeated the West in any meaningful sense.

This might be an unconventional statement, but AI is one of those domains where you can build amazing AI agents using existing models from anywhere in the world. I’m glad we’re starting to see that happening.

Jordan Schneider: It was remarkable how this company, with both its browser and its first product Monica — a ChatGPT-like search browser add-on — targeted foreign users first. That’s notable because running Claude is illegal in China, which makes development difficult.

Reading interviews with the CEO over the weekend, he stated essentially: We’re not really trying to take on the big labs, but we think there’s an opportunity and a big market here. It was somewhat sad reading when he alluded to the politics of AI: “I've come to understand that many things are beyond your control. You should focus on doing well with the things you can control. There are truly too many things beyond our control, like geopolitics. You simply can't control it—you can only treat it as an input, but you can't control it.”

Frankly, I don’t think Chinese AI agents will have much longevity in the US market without hitting some severe regulatory headwinds. However, their skill at playing the global influencer marketing game to generate this hype cycle reflects a real fluency in digital marketing. The fact that they could play this game better than any Western agent competitor — except for Devin, which tried but faced its own challenges — is remarkable. There hasn’t been another major attempt at this over the past year and a half.

Dean Ball: I would go even further. When the first Devin demos appeared, people exclaimed, “Look how cool this thing is!” Then the bubble burst when people realized it was GPT-4 with prompt engineering and scaffolding.

The Western AGI obsession makes us want to conceptualize this as one godlike model that can do everything, and we implicitly dismiss product engineering and practical applications. You see that reflected in public policy, which is obsessed with big models, giant data centers, and similar infrastructure. Those are the only things we seem to take seriously and value.

I’m a deep learning optimist — I’m not going to tell you AGI doesn’t exist or is a Gary Marcus fiction. I’m not in that camp at all. But the AGI obsession has developed into something that feels like a perversion, distracting us from opportunities lying right in front of us.

I’m not necessarily saying Manus represents that opportunity, but there are thousands of possibilities where cleverly stringing together different AI products and modalities could yield interesting results. We just don’t see much of that happening. A year ago, I was more inclined to say, “Well, it takes time,” but a year later, I find myself less willing to make that excuse.

Structural Factors Driving the AI Product Overhang: Why Big Labs Don’t Do Product

Rohit Krishnan: Shawn, what is this? Are the VCs dumb? Are the founders dumb? Are there actually not pennies to be picked up off the ground?

Shawn Wang: There are, and the VCs have woken up to it. I started writing about the rise of the AI engineer two years ago, and now there are VPs of AI engineering at Bloomberg, BlackRock, and Morgan Stanley — they just spoke at my conference last week.

People were very dismissive of the GPT wrapper, viewing it as just a thin layer over the LLM. Now the perception has almost flipped, where the model is the commodity and everything else on top of it is the main value and moat of the product. This is why I started talking about AI engineering, and I think it’ll be a growing job title. It’s what we orient my conference and podcast around.

It’s music to my ears. I’ve been saying this for a while now, and the VCs have caught up. It’s just harder to fund because you can’t just say, “Here’s the pedigree of the 10 researchers I have. Give us $300 million.” Now you have to actually look at the apps and see if they’re well-engineered and fit the problem they’re trying to solve, whether B2C or B2B. That’s much more difficult than throwing money and GPUs at talented researchers and letting them go for it. That approach caused Inflection AI, Stability AI, and other mid-tier startups to burn around $100 million each.

Rohit Krishnan: That’s back to the SaaS era in some sense. You suddenly find a new vertical niche where you can build something, spend time and effort, learn about the specific problem you’re solving — not just intelligence but something more targeted, B2C or B2B. Then you have to tackle it and solve it.

Shawn Wang: The interesting thing about this SaaS transition is you’re charging on value and not on cost, and the margins between those approaches are enormous. Many of us in Silicon Valley realize that if you develop your own models, the next one that comes out is probably open source from China and better than yours. So where’s the value in that?

Everything’s being competed down to cost. Anthropic offers Claude at cost. OpenAI has a small margin, but every other GPU provider serving open source models is just providing at cost because they’re trying to capture market share with VC money. Nobody’s making a margin here.

You contrast the $200 versus the potentially $2,000 or $20,000 a month agents you can offer, because you’re competing against human labor and human thinking time, which we are all limited by. The economics start to really work out. You could start charging for your output instead of charging for your cost of goods sold. That is fundamentally a better business.

Rohit Krishnan: Speak for yourself, Shawn.

Shawn Wang: The fact that you could just start charging for your output instead of charging for your cost of goods sold is fundamentally a better business.

Rohit Krishnan: You wrote a very cool thing about Google’s awkward struggles to make products that people use. What is stopping the big model makers from starting to do things they can charge value-based pricing for? Is it just that they don’t need to and have their hands full making AGI?

It does seem that just selling tokens isn’t going to make you money in the long run. It’s funny because if you’re one of the large labs, if you’re Sam or Dario, you don’t particularly care about that since you already have so much money coming your way. Anthropic just raised $60 billion, OpenAI is valued at $300 billion. These are astronomical figures.

We’ve normalized these numbers in conversation, but they’re absurd by any stretch of imagination. $300 billion is bigger than Salesforce. It’s insane to think about for a company. Why are they getting that money? Because they want to build AGI. Why do they think they can build AGI? Partly because they’re true believers, partly because they have the best research talent in the world who wants to build AGI.

What happens if you tell that research talent that they’ll be working on building agents for awhile? Many of them quit. Arguably, many did. In a weird way, it’s only in larger places like Google where you can potentially have a large enough contingent of people try some unusual approaches and build cool stuff.

They did create interesting products — NotebookLM is actually really interesting. It was a cool new product, new modality, new way of interacting with information. I am surprised that we didn’t see more of it. In typical Google fashion, it just kind of disappeared after a while. They have Colab, which is an interesting product that’s languishing in a corner somewhere.

Everything Google does involves creating a very interesting first product and then slowly killing it by cutting off the oxygen supply over the next five years. For somebody to care deeply about building a product here, it has to start right at the top. It has to come from a mission, because the argument against building a product — the engineers saying, “Just wait a year and everything will get solved" — is really seductive.

Safety, Liability, and Regulation

Shawn Wang: You really need somebody who has a Jobsian level of ability to push back and say, “I don’t care what you guys think. We need to actually build something that really works here.” That’s not a muscle that any of these companies have because none of them have built products. Arguably, the thing that kicked it all off, ChatGPT, was built as a research preview. What we are doing is all being okay with playing around with research previews that consistently sneak their toe in and pretend they’re a bit of a product, but they’re not really.

Rohit Krishnan: Let’s fast-forward to the near future when agents can do more economically useful things than book you a train ticket. Should we start with the safety angle? It’s wild if I’m going to let something exist as me on the internet or in my workplace and I’m responsible for it. Maybe Manus is responsible? Maybe OpeningEye’s responsible? Maybe the AI engineer who goes to Shawn’s conferences is responsible?

This is a very weird world where it’s not just Jordan Schneider as an AI-enhanced worker using chatbots, but Jordan Schneider letting go a little bit and having these automated minions exist under my aegis but also not.

Dean Ball: I haven’t checked their website thoroughly, but I would be very surprised if Manus or the company that built it has a safety and security framework, a responsible scaling policy, or has commented on the EU code of practice.

Rohit Krishnan: I actually looked for this. I could not find one thing that the CEO has said in any relation to any safety discussion or question.

Dean Ball: This thing doesn’t have any guardrails. I don’t think it’s a consideration for them. In some sense, that’s probably part of what makes this better than Operator or Claude computer use, because Anthropic and OpenAI have both legitimate business incentives and internal stakeholders who won’t let the company ship things with no guardrails.

There’s reputational risk. If OpenAI had released something like Operator with zero guardrails, you’d be looking at state attorneys general investigating you, and the FTC and others coming after you, just as they did with ChatGPT. The tech industry is pretty risk-averse on things like this because it’s an inherently risky endeavor.

Those are market incentives, because you shouldn’t be incentivized as a consumer or business user to throw agents into the wild who do things for you and potentially cause problems. There should be some liability for that. You should be incentivized not to do such things, and companies should be incentivized not to release such things.

I’ve been thinking about liability issues in the last few months and have concluded that the court system is going to really struggle. If something happened with Manus, there’s the user who prompted it, multiple LLMs behind the scenes, and a Chinese company that is almost certainly not subject to a legally cognizable claim, unless you want to go to court in Beijing. How is the American tort liability system going to figure this out? I’m skeptical it will do a very good job.

But no liability is a moral failing, too. As the cost of cognitive labor declines, one of the only things left with economic value is trust, pricing risk, and similar concepts. I wonder if frontier AI companies will slowly converge to being more like insurance companies or financial services companies. Those industries are based on trust, pricing of risk, and allocation of responsibility for harm that occurs from realized risks. That feels like what’s economically valuable here, certainly not selling marginal tokens.

Shawn Wang: There’s one proof point that maybe agrees on some level: we’ll never get the O3 API because OpenAI is choosing to release products instead of APIs. That makes sense if you believe your APIs are valuable — you stop giving them to everyone else. It also stops the Manuses of the world in their tracks, because they can no longer use those APIs.

In broadening this general safety discussion, this is just an argument for American AI accelerationism. The simple fact is, if you are more safe and stop yourself from doing anything, then China will do it first, and you’re behind. It’s better to be ahead and in control of the narrative, build in the safeguards at the LLM layer with the post-training that you do, and try to lead from the front instead of the back.

Rohit Krishnan: I have a more contrarian view. Even framing this in terms of safety is incorrect. What are we talking about today? The Manus of today, Operator of today — these aren’t safety concerns. They’re engineering concerns, misuse concerns. We’re using the 2023 version of AI safety, which seeps into every part of the “anything a model can do can be unsafe” conversation, and that distorts how we discuss what these products do.

As Dean said, the liability issue is important once these tools start getting used inside companies. If someone at Pfizer uses Manus to figure something out and creates a wrong drug, there are liability issues. If someone at Cloudflare uses Manus to fix a bug and creates an outage, there are clear questions about where responsibility sits.

But we’re still at the point of making these things work properly in the first place. Think about our example — testing if it can book an Amtrak ticket. We’re not yet at the point where AI agents are so incredibly amazing that we have to restrict them before they engage. I’d like to see them work properly before we leash them.

That doesn’t mean we shouldn’t have a parallel track thinking about liability issues. But these will be hard-won battles that push the frontier forward one issue at a time, rather than “We’ve figured it out for everything from searching medical information to booking tickets to writing open source code or malware."

One problematic outcome of these discussions in recent years is that we’ve conflated all these issues into one, and they’re not the same. I look at Manus and think, “Good. I’m glad somebody without a responsible scaling policy is showing us what can be done,” because there’s no inherent problem with giving something a browser. Yes, there can be prompt injection attacks — that’s new and we need to solve it, but we can’t figure it out without anybody actually doing anything. It’s a chicken and egg issue.

Dean Ball: If you’re a dentist trying to use AI to automate business processes within your dental practice, then the fact that OpenAI has a responsible scaling policy about biological weapons risk evaluation isn’t that important for you. But perhaps more problematically, OpenAI’s model specification says, “Follow the law.” Okay, gotcha.

My view is that we have to almost entirely reject the tort liability system for this because it’s too complicated an issue. This is the kind of situation where transacting parties need to come to agreement about what makes sense in these particular contexts and let contracts do their thing. The courts won’t adjudicate this on a case-by-case basis in any effective way.

The risk of accelerationism is that you accelerate without proper safeguards. Noam Shazeer left Google to accelerate and founded Character.AI. What happened? Character.AI said problematic things to children, made sexual advances to children, and a kid killed himself. I don’t know if you could say Character.AI is responsible for that child’s suicide, but he was talking to the chatbot when he killed himself. That’s a tort case — Tristan Harris is funding it in the State of Florida, with a sympathetic jury and judge.

What’s Character.AI now? It’s a husk. Noam Shazeer’s gone, back at Google, and the company is likely to be picked apart in tort litigation, with other cases against them too.

If you accelerate without figuring this out, something very bad could happen. As they say, bad facts make bad law. Maybe it’s not unambiguously the AI model’s fault, but if there’s a really nasty set of facts, you could get adverse judgments in American courts. The common law is path-dependent, so you could end up with a very bad outcome quickly.

I’m enthusiastic about accelerating adoption and diffusion — Manus is very much a diffusion story. But if we don’t, in parallel, work on risk assignment (not catastrophic risk safety, but determining who is responsible when things go wrong), we could end up in a bad situation rapidly.

Legal Frameworks and Innovation Timelines

Shawn Wang: Do you think it’s primarily financial infrastructure that is needed, like your model of AI companies as insurance companies?

Dean Ball: Legal and financial, yes. What you basically need is a contracting mechanism that is perhaps AI-enabled — AI-negotiated contracts, perhaps AI-adjudicated contracts so you don’t have to deal with the expense of the normal court system. Once you have contracts and liabilities on the balance sheet, you’re in derivatives territory.

It’s a New York problem, not a San Francisco problem at that point. This is certainly an AGI-pilled idea. I wouldn’t do this with Claude 3.7, even though I think it’s great, but I think we could get there in the next couple of years when models are capable of doing things like this.

This is just one approach, certainly not the only one, and it’s a nascent idea for me. But this could be where the money actually is — pricing risk and transforming risk is something America is much better at than China. We’re fantastic at that.

It’s weird because many of my Republican friends in DC hate that fact. They view finance as decadent, as does Chairman Xi. But there might be trillions of dollars of wealth to be created here.

Subscribe now

Shawn Wang: Any financial asset is based on the laws it’s grounded in, and I think the laws have to be figured out here. There’s a bit of a chicken-and-egg situation with that.

Dean Ball: Yes, but if you have contracts, contracts would form a substantial part of the law.

Shawn Wang: They still need to be litigated. One measure I’d be interested to plot is AGI timelines versus legislative timelines. We’re accelerating in AI progress and decelerating in law and Supreme Court resolutions of cases. Our legal infrastructure needs to keep up with AI progress, or we’re in serious trouble.

Dean Ball: I completely agree. That’s the problem I try to get my head around all the time.

Jordan Schneider: We saw the EU just try and completely fall on their face, which was not a good first effort for democracies.

Dean Ball: The problem is you don’t want to create a statute prematurely — a statute with a bunch of technical assumptions embedded in it prematurely. It’s a very narrowly targeted thing. For me, this is all clicking into place, and I think if we got this done in the next two years, we’d be fine.

Jordan Schneider: The future of agents in China is going to be really interesting. There was an argument two years ago that LLMs would have a hard time gaining traction in China because the government would worry about aligning them to avoid anti-party statements. But this is basically a solved problem.

I’m curious about your perspectives on the technical challenge — not just at a legal level of assigning blame, but at a product and operational level of building things that governments and large companies will be comfortable with. Is this just a matter of time? Is there anything fundamentally difficult requiring major breakthroughs? Once we have the technology to make Operator and Manus do really good things, will they be controllable as well?

Dean Ball: I’d be curious if you’d correct this assumption if I’m wrong, Jordan, but my impression of China is that it’s actually a somewhat more ice-cold libertarian country when it comes to liability issues, where there’s a greater “developing country” or “these things happen” mentality.

Jordan Schneider: Yes, until bad things happen, and then your company gets shut down.

Dean Ball: It’s more of a binary outcome.

AI and the Future of Work

Jordan Schneider: Let me take this in a different direction. JD Vance at Paris said, “We refuse to view AI as a purely disruptive technology that will inevitably automate away our labor force. We believe and we’ll fight for policies that ensure AI is going to make our workers more productive. We want AI to be supplementing, not replacing work done by Americans.”

Shawn Wang: This is something AI engineers worry about a lot. A surprising number of them are actually worried for their own jobs, which is very interesting.

The main question is whether you have a growth mindset or a fixed mindset view of the world — whether you believe human desires tend to expand over time. Whenever we reach a certain bar, we immediately move that goalpost one football field away. The idea is that, yes, AI will take away jobs that exist today, but we will create the jobs of tomorrow, and ideally those are the jobs we want to do more of anyway.

Rohit Krishnan: I agree. To a large extent, that sentiment is the most normal politician statement in the world — technological growth is great and will continue making everyone’s lives better. It’s the same thing people have said for a very long time.

The difference here is that there’s at least a contingent of people who look at that and say, “No, this time it’s different.” You might say it’s not disruptive, but it could be massively disruptive in a short period of time to a large segment of society. It’s not just agriculture getting mechanized, but potentially all white-collar jobs.

When I’ve examined this issue, I don’t think massive disruption will happen immediately. The technological, regulatory, and sociological barriers are large enough that we won’t all be unemployed in five years. There are enough things to do. As Shawn mentioned, we’ll have to address the inevitable complications of regulatory frameworks before these technologies can be deployed everywhere.

When I did some rough calculations, we’ll still be bottlenecked by chips and energy in 10 years, which will prevent us from replacing all labor with AGI or AI agents. Does that mean there will be no disruption? Absolutely not. I fully expect disruption.

We already have AI that can plausibly replace large chunks of specific white-collar tasks that I do, you do, legislators do, or Supreme Court justices do. Pick your poison — we could probably replace a chunk of these roles with Claude 3.7 and get better results. We’re already in that world, but complete transformation won’t happen immediately.

JD Vance has to toe the party line: anti-AI safety, pro-acceleration, technological optimism all the way. I support that approach, but I’m not parsing his statement with any deeper meaning than that.

Dean Ball: I find myself more worried about slow diffusion due to multiple factors. The bottlenecks are regulatory, but there are many other bottlenecks as well. I’m much more concerned that diffusion and actual creative use will be slowed. I worry about the uses of LLMs that no one has ever thought of, and I’m concerned that no one will ever think of them. That’s probably the bigger issue we should be addressing through public policy.

In the longer term — and in AI time, that’s about three years — I do think there’s a possibility that some elite human capital might get automated in different ways. Political instability tends to emerge when you have an overproduction of elites in a society. We already have that problem. We’ve already significantly overproduced elites in America, and I worry that will get worse. When you combine that with other political problems America faces, you could have a tipping point phenomenon.

I wouldn’t dismiss it entirely as a risk, but my default assumption would be that the risk is actually on the other side — not diffusing fast enough.

Rohit Krishnan: I have a thesis that I sometimes hold that markets that become extremely liquid end up with polarized outcomes. We’ve generally seen that with capital markets — globalization has meant some companies get extremely large while the middle gets decimated, which is why most gains come from the Magnificent Seven.

We could easily see something similar happen in labor markets. We already see flashes of it. Engineering salaries have a somewhat bimodal distribution. Lawyers experience it too. Once AI enters the picture, we might have a vastly more liquid labor market than ever expected. This sounds nice, except it results in a power law distribution. Polarization is difficult to address in domains we don’t know how to handle cleanly or where we can’t easily establish minimum thresholds.

Jordan Schneider: Any closing thoughts?

Shawn Wang: [In true professional podcaster fashion…] I don’t know if we’ve answered the question that you’re likely to put in the title of your episode: “Is this a second DeepSeek moment?” For what it’s worth, my answer is no.

Dean Ball: I agree. In some sense, I think it’s actually more interesting than DeepSeek. And in another sense, it’s certainly not as impressive of a technical achievement as DeepSeek.

Rohit Krishnan: I’ll argue for yes, because I think DeepSeek was a DeepSeek moment for core research talent. Manus is closer to DeepSeek for product. I’m glad we’re pushing a second boundary as opposed to pushing the same boundary.

Mood Music

Just as the buzz around DeepSeek was beginning to fade, Chinese AI has made waves again with the AI agent “Manus,” launched on March 6th, 2025. Today, we’re here to unpack the Manus launch, explore the business model of Manus’ parent company, and offer a glimpse into the mind of Xiao Hong 肖弘, the founder behind China’s latest viral AI product.

What is Manus?

Manus claims to be the world’s first general-purpose AI agent. It ostensibly outperforms OpenAI’s ChatGPT Deep Research on the General AI Assistants (GAIA) benchmark. Currently in beta testing, access is restricted to those with invitation codes, which are reportedly being listed second-hand for 50,000-100,000 RMB (whether anyone actually paid that much is another question). Users report impressive performance in basic tasks, for instance rebooking airline flights, beyond what Anthropic’s Computer Use and OpenAI’s Operator have thus far provided to users. The product is also experiencing slowing response times, hinting that Monica.ai may be struggling to scale up compute to meet skyrocketing demand.

Manus started in 2022 as an AI-powered browser plugin, backed by ZhenFund (真格基金). In 2023, the company secured Series A funding led by Tencent (腾讯) and Sequoia Capital China (红杉资本中国). What began as a simple “ChatGPT for Google” browser plugin has since evolved into a full-fledged AI agent.

Monica, the company that developed Manus, operates from Wuhan, rather than from China's major tech hubs like Beijing or Shanghai. In early 2024, ByteDance attempted to acquire Monica for $30 million, but founder Xiao Hong (肖弘) turned down the offer. ByteDance’s plan was to absorb Monica’s team and technology into its Doubao AI ecosystem, a move that would have diluted Monica’s distinct market position. Instead, Monica closed a new funding round at the end of 2024, reaching an estimated valuation of nearly $100 million.

The exact AI models powering Manus remain unclear. The company claims to use multiple models for different tasks. Notably, when prompted to reveal its own system files, Manus reveals it may be powered by Anthropic’s Claude models — which would make operating in China illegal. This probably explains why Monica’s website appears to be blocked in China.

Edit: confirmed by co-founder.

Anyway, the fact that Manus appears to disclose more than it should hints at broader potential security vulnerabilities.

Who is behind Manus?

Founder & CEO, Xiao Hong (肖弘), is a serial entrepreneur and a graduate of Wuhan’s Huazhong University of Science and Technology (华中科技大学). He first made his mark by building WeChat-related tools as a student, admitting that while his “academic performance was quite poor,” he partnered with more technical classmates to build tools. In 2015, he launched Nightingale Technology (夜莺科技) and created Yiban Assistant (壹伴助手), a WeChat management tool that secured early backing from ZhenFund (真格基金).

By 2019, Xiao saw a bigger opportunity in enterprise WeChat tools and developed Weiban Assistant (微伴助手). His timing was perfect—when rival WeTool (微商工具WeTool) was shut down in 2020, Weiban became the go-to alternative, attracting investment offers from Sequoia Capital China (红杉资本中国) and Youzan (有赞). Eventually, Minglue Technology (明略科技) acquired Weiban, marking Xiao’s first major financial success.

Sensing the potential of large AI models, Xiao left Minglue in 2022 to create Monica.ai, originally designed as a “ChatGPT for Google” browser plugin.

Co-founder & Chief Scientist, Ji Yichao (季逸超) dropped out of high school at 17 to develop Mammoth Browser (猛犸浏览器). His talent caught the eye of Sequoia Capital China’s Zhou Kui (周逵), who introduced him to investor Xu Xiaoping (徐小平). Xu invested 1.5 million yuan, giving Ji complete creative freedom. Recognizing the large potential of LLMs, Ji joined Xiao Hong to start Monica in late 2022.

Interview Quotes

Unlike DeepSeek’s media-shy Liang Wenfeng, Xiao Hong has done a ton of press. Below are selected translations from several in-depth interviews with Monica’s founder and CEO, Xiao Hong, offering insights into his vision, strategy, and the future of AI agents.

The vibe of Xiao Hong’s interviews is distinct from the AGI-driven idealism blended with national pride we’ve seen from the founders of DeepSeek and Unitree. Xiao is pragmatic and focused on profitability rather than research. A newly published three-hour podcast with Xiao opens with offering this piece of advice:

“I remember there was a Northeastern Chinese restaurant near my university. I made enough money to treat my tech club friends to dinner there every day. Here’s a tip for the audience: if you’re in college, take your most talented classmates out for meals as often as you can. If you wait until after graduation to recruit them for your startup, you’ll have to treat them to Michelin-starred restaurants instead.”

In another interview from January 2024, Xiao openly admits that he didn’t initially believe in AI’s potential, and “remained cautious” despite the hype surrounding GPT-3.5 in the fall of 2022. He describes coming to two conclusions about AI investment, which eventually led him to focus on AI products as opposed to chasing AGI with foundational model research:

"First, I wouldn’t consider working on big models without sufficient business scale. Second, I believe that in China, big model services will eventually integrate fully with cloud computing. I’ve discussed this with our CTO and believe that cloud computing companies will provide customized deployment services, so we don’t need to dive into that ourselves."

…

"I focused more on what big models could do, and what kind of applications I could build with them. In the beginning, many people were financing based on concepts, but by the second half of the year, both domestic and international, there was much less of that. Everyone was returning to business rationality, focusing on finding PMF (Product-Market Fit). By February of 2023, I had a conversation with an investor focused on big models, and no matter how I asked, they refused to talk about products. They weren’t discussing technology or plans. By March, the product’s valuation plummeted. People realized that simply building a single application based on big models might not work, and that’s when the consensus started forming: either focus on technological breakthroughs or work on relatively closed-loop application scenarios."

…

"In March and April of 2023, the fastest-growing product outside of ChatGPT globally was Poe. It was essentially a shell around a big model, and I told investors that if you can perfect the shell, that’s still a big deal. So we decided to do it too, and instead of resisting the demand, we decided to embrace it. In the first half of 2023, Monica integrated all the major models because that’s what the users wanted, and we started by doing that, figuring out how to find more use cases step by step."

Monica’s business model focuses on catering to the overseas market, which likely explains why their website is devoid of any reference to being based in China. Besides English, Monica’s website has dedicated versions in traditional and simplified Chinese, as well as Russian, Ukrainian, Bahasa Indonesian, Persian, Arabic, Thai, Vietnamese, Hindi, Japanese, Korean, and a slew of European languages.

In Xiao’s words, “We chose to target the overseas ToC market because I felt it was a larger, more commercially viable market. The domestic market seemed a bit more challenging.” Their focus shows: in contrast to DeepSeek’s very low key model launches, Manus’ launch came with a whole sophisticated press push like one you would see out of a YC startup, complete with a very well-produced English-language launch video and early access for select YouTubers and twitter influencers.

International expansion comes with its own difficulties, but Xiao believes those challenges made Monica stronger as a company. He’s recently argued that China would benefit from having more firms look abroad:

Xiao Hong: I think we are still in a great era with many opportunities…. First, it's the AI era. Second, I think we are also in a great era of globalization. I'm not a geopolitical expert, but it seems like every country has its own problems — internally, everyone has their own issues. So overall, the world is becoming more conservative and more isolationist, right? But at the same time, no one wants others to be isolationist; they only want to be isolationist themselves. So, everyone hopes that their own entrepreneurs will think more globally.

I believe China’s entrepreneurs of today should be more aggressive in globalizing. If we see overseas markets as better opportunities, it’s not just about market-driven decisions — we should step into international markets to gain experience. We need to participate in global competition, rather than just competing in the markets we are familiar with.

By the way, this process requires a lot of things. When I started this company, none of our founders had lived abroad for an extended period. Everyone’s English proficiency peaked in high school and declined in college! [

I once joked that if, at the same time, there was another founder who had lived in the U.S. and was placed next to me, I would have chosen to work with that founder myself. But this shouldn’t be the way we compare things — it should be about doing our own thing. Secondly, I had a simple belief at the time: the global market is much bigger, and the market itself will provide the tuition fees for founders to learn. (Laughter)

Besides the AI era, another crucial topic is that we are now thinking about things with a globalized mindset.

Unsurprisingly, this business model also relies on collecting vast amounts of user data. Monica’s free Chrome extension requests expansive access to browser data, including permission to log keystrokes, and Manus “crawls” devices to make suggestions. Xiao is betting that widespread adoption of these products will unlock a treasure trove of monetizable insights.

“The data we collect through our browser plugin is critical. Even though this might not guarantee success, it’s a step in the right direction. The private data we gather, along with contextual information, will help differentiate us from the competition. This is one of the key assets we need to grow.”

Xiao is explicitly describing an intent to build an incumbent advantage on a foundation of user data, and TikTok demonstrates how effective that strategy can be. Reliance on eventual mass adoption could partially explain the high-publicity invite-only launch strategy for Manus (although limited access to compute is also certainly a factor).

That said, he is aware that the politics exist and could get in the way of a Chinese-owned AI agent gaining widespread adoption abroad. He spoke about it in a recent podcast alluding to NeZha 2.

I've come to understand that many things are beyond your control. You should focus on doing well with the things you can control. There are truly too many things beyond our control, like geopolitics. You simply can't control it—you can only treat it as an input, but you can't control it.

I recently asked DeepSeek to explain three terms 贪 (greed), 嗔 (hatred), and 痴 (ignorance) [the ‘three poisons’ of Buddishm recently spotighted in the truly excellent animated movie NeZha 2]. It explained it very well: greed is attachment to favorable circumstances; anger is dissatisfaction with adverse circumstances; and ignorance is not understanding the truth of the world. The "truth of the world" is very profound, so I won't discuss that. But greed and anger are problems many people encounter, as are attatchment to favorable circumstances and dissatisfaction with unfavorable ones.

This business-minded pragmatism shines through in Xiao’s vision for the future — instead of techno-optimist visions of AI-powered drug discoveries or a moon colony staffed by robots, he imagines a world where humanity can return to a glorious past:

“I think that the white-collar lifestyle may be a detour for mankind. If you look at it in terms of a curve or over a longer period, say thousands of years, or even the ten-thousand-year span of human history — it's actually quite rare for people to sit in one place and engage in intense mental work without much physical activity. This is probably only a phenomenon of the past hundred years.

For a longer time in history, maintaining physical health and developing spiritual civilization have gone hand in hand. In ancient times, people also needed spiritual and cultural development, but that involved physical labor as well, which helped strengthen their bodies.

In the past hundred years, however, issues like diabetes and high blood pressure have become widespread because people work in this sedentary way. If we look at humanity as a whole, sitting and working for eight or more hours a day is an anomaly.

If AI can take over these tasks, then people can work fewer hours and go back to living more like they did in the past — focusing more on spiritual and cultural enrichment while also taking better care of their physical health.”

To close, here’s a quote from Xiao about how it feels to live through history:

Xiao Hong: From the time I was born in the 90s until now… there have been significant shifts, from PCs to mobile, then the semiconductor industry, which has been booming behind the scenes, the rise of the internet, and now artificial intelligence. I feel like these opportunities are emerging very intensively. When I watched The Godfather, I realized that if I had lived in that era — it was also a time of change — but if you lived in certain periods, you might not have witnessed such rapid technological progress. Sometimes, when we read history books or ancient texts, it feels like things barely changed, which I think would be a little frustrating!

From an Anon contributor who would know:

TLDR: U.S. export controls targeting China’s AI capabilities focus primarily on limiting training hardware but overlook the growing importance of inference compute as a key driver of AI innovation. Current restrictions don’t effectively limit China’s access to inference-capable hardware (such as NVIDIA’s H20) and don’t account for China’s strong inference efficiency. While China’s fragmented computing infrastructure has historically been a disadvantage, the shift towards inference-heavy AI paradigms positions their compute ecosystem to be more utilized and valuable. As reasoning models, agentic AI, and automated AI research elevate the role of inference to advancing AI capabilities, the US should urgently strengthen export controls to hinder China’s inference capacity and develop a coherent open-source AI strategy to maintain competitive advantage.

The Export Control Status Quo is Broken

The global AI competition is unfolding along two critical axes: innovation — the development of advanced AI capabilities — and diffusion — deploying and scaling those capabilities. The United States has prioritized outpacing China in AI innovation by focusing on pre-training as the main driver of progress. However, a new paradigm is emerging where inference, not just training, is becoming central to advancing AI capabilities.

This shift has significant implications for U.S. AI policy. Current export controls aim to limit China’s ability to train frontier AI models by restricting access to advanced chips, based on the belief that scaling pre-training is the primary driver of AI progress. By limiting China’s access to compute resources, these controls aimed to slow its AI development.

Yet, these same controls are far less effective at restricting China’s inference capabilities — exposing a critical gap in U.S. strategy. As inference becomes more central to AI innovation, current policies are increasingly misaligned with the realities of AI development. To effectively counter China’s growing inference capabilities, the U.S. must strengthen its export controls.

Inference Compute is a Key Driver of AI Innovation

The AI landscape is evolving beyond the scaling pre-training paradigm that dominated recent years. Emerging solutions are shifting innovation toward a paradigm where inference compute — not just training compute — has become a critical driver of AI progress.

Three interconnected trends are driving the link between inference and AI capabilities:

Reasoning Models

These models require significantly more inference than traditional LLMs, leveraging test-time scaling laws that suggest a link between amount of inference compute and model performance. Inference demand is further driven by a feedback loop that accelerates AI capabilities: reasoning models generate high-quality synthetic data, which enhances base models via supervised fine-tuning (SFT). These stronger models can be adapted into stronger reasoning systems, creating even better synthetic data and fueling continuous capability gains.

Agentic AI

AI agents — systems capable of taking autonomous actions in complex environments to pursue goals — are often powered by reasoning models, which drives up inference demand. Many agents have access to external tools and environments such as code execution environments, databases, and web search, which enhance their capabilities by enabling them to retrieve information, plan, and interact with digital and physical environments.

Some agents continuously learn by interacting with their environment via reinforcement learning. Unlike standard language models that handle one-off queries, agentic AI systems require persistent inference as they continuously interact with external environments, adapt to new information, and make complex, multi-step decisions in real time — significantly increasing overall inference requirements.

Automated AI Research

Automated AI researchers can design new architectures, improve training methods, run experiments, and iterate on findings. Scaling in this paradigm requires both inference compute to power research agents and training compute to execute their proposed experiments. Greater inference capacity allows more of these systems to operate in parallel, expanding both the breadth and depth of AI exploration. This, in turn, enlarges the search space they can navigate and directly increases the rate of AI innovation.

Greater inference also enhances research agents through iterative reasoning, self-play debates, and automated evaluation — capabilities already demonstrated in AI-driven scientific discovery. As these automated systems achieve early breakthroughs, they become better at identifying promising research directions and architectural improvements, potentially setting off a compounding cycle of progress. Thus, even small initial advantages in inference capacity can compound, leading to a significant, potentially decisive, lead in AI capabilities.

In an era of reasoning models, agents, and automated AI research, inference capacity is not just an enabler — it is a primary determinant of the speed and trajectory of AI innovation. This shift has significant implications for the U.S.-China AI competition and underscores the need for stronger U.S. export controls.

China’s Inference Capacity is Key

Current U.S. export controls aim to restrict China’s ability to train frontier AI models but overlook the growing importance of inference and China’s capacity to scale it. As AI development shifts towards inference, China’s position strengthens considerably due to three key factors:

1. Steady access to inference-viable GPUs

2. Leading inference efficiency

3. Compute ecosystem being better suited for inference rather than pre-training

Access to Inference Hardware: The H20 Loophole

Despite U.S. export controls restricting access to cutting-edge AI chips like the H100 and H800, China maintains strong access to inference-capable hardware through several avenues — most notably through Nvidia's H20 GPU.

The H20 represents a significant gap in current export restrictions. Specifically designed to comply with export controls and serve the Chinese market, the H20 is actually superior to the H100 for particular inference workloads. The H20 outperforms the H100 for inference workloads due to its superior memory capacity and bandwidth. It delivers 20% higher peak tokens per second and 25% lower token-to-token latency at low batch sizes—key advantages given that inference performance is driven more by memory bandwidth and batch efficiency than by raw computational power. With 96GB of HBM3 memory and 4.0TB/s memory bandwidth, compared to the H100’s 80GB and 3.4TB/s, the H20 is highly viable for inference, making it a significant gap in current export restrictions.

China has been importing large sums of the H20. SemiAnalysis estimates that in 2024 alone, NVIDIA produced over 1 million H20s, most of which likely went to China. Additionally, orders by Chinese companies, including ByteDance and Tencent, for the H20 have spiked following DeepSeek’s model releases.

Access to Inference Hardware: Trailing-Edge GPUs

Trailing-edge GPUs remain surprisingly effective for inference workloads. China retains strong access to trailing-edge GPUs due to large stockpiles of the A100, A800, and H800 in 2022 and 2023. Additionally, Chinese firms, including Huawei, Alibaba and Biren, have also developed indigenous chips. The viability of trailing-edge GPUs for inference suggests that China’s inference capacity is stronger than their volumes of cutting-edge GPUs may suggest.

The effectiveness of older GPUs for inference stems from fundamental differences between inference and training workloads:

Long-Context Inference is Memory-Bound, Not Compute-Bound

Unlike training, inference only runs forward passes, avoiding computationally intensive processes like backpropagation and gradient updates. As a result, inference is significantly less compute-intensive than training.

The real constraint for inference is memory. Inference, particularly long context inference, is currently memory-bound rather than compute-bound due to several factors:

1. Model Weights & Key-Value (KV) Cache: For transformer-based models, inference requires storing both the model parameters and a key-value (KV) cache. The KV cache stores the past tokens' key-value pairs, allowing the model to retain context and coherence, and grows linearly with the context length. While compute resources are only required to process each newly generated token, memory usage continuously increases as new key-value pairs for each transformer layer are stored in the cache with every additional token generated. Consequently, total memory consumption rises steadily as the context expands, in contrast to compute needs, which remain stable and do not accumulate in the same manner. As a result, inference often becomes memory-constrained before it becomes compute-constrained, particularly for long-context tasks, where the KV cache can exceed the model weights in size.

2. Autoregressive Bottleneck: Input tokens can be processed in parallel, leveraging the full sequence since it’s known upfront. However, output tokens are generated sequentially, with each new token depending on all the previously generated tokens. This creates a bottleneck during output generation:

   1. Full KV Cache Access: Each generated output token requires accessing the entire KV cache.

   2. Memory Bandwidth Limitation: On long sequences, this repeated full KV cache access for every output token creates a memory bandwidth bottleneck (data transfer rate between memory and processor), which becomes the primary limiting factor.

   3. Constrained Batch Sizes: The size of the KV cache directly limits batch size during output generation. Longer sequences consume more GPU memory, reducing space for batching multiple sequences. This forces smaller batch sizes–the amount of independent user queries that can be processed in parallel–which reduces GPU utilization and restricts inference throughput.

This memory constraint becomes evident when examining FLOP utilization rates. During inference operations, GPUs typically achieve only about 10% FLOP utilization when generating tokens, compared to 30-50% during training. This underutilization occurs because GPUs spend much of their time retrieving and managing the KV cache rather than performing actual computations. The inefficiency grows even more pronounced with newer, more compute-dense chips, where increasingly powerful processing cores sit idle waiting for data to arrive from memory.

The Memory Wall

This inference bottleneck reflects a broader structural limitation in computing hardware. While GPU compute performance has grown exponentially (approximately 3.0x every 2 years), memory bandwidth and capacity have improved at a much slower rate (around 1.6x every 2 years). This growing gap creates a “memory wall” where performance is constrained not by processing speed but by how quickly and how much data the GPU can store and access.

This memory-bound nature of inference has significant implications for hardware viability. While newer GPUs offer exponential improvements in raw computational power (measured in FLOPs), they provide more limited gains in memory capacity and bandwidth — the true bottlenecks for inference workloads.

As a result, inference workloads often cannot fully utilize the computational resources available in cutting-edge GPUs. When memory bandwidth is the primary bottleneck rather than raw compute power, older GPUs remain surprisingly effective for inference tasks. The performance gap between newer and older GPU generations becomes much less significant than their computational performance might suggest.

These technical characteristics create a unique hardware dynamic that changes the calculus around AI chips. Trailing-edge GPUs retain viability in an inference-dominated landscape — a generation-old GPU might deliver 60-70% of current-generation inference performance, making it highly viable for most applications. This shifts the cost-effectiveness equation; dollar-for-dollar, older GPUs often provide better inference performance per unit cost than cutting-edge hardware optimized for training workloads. While trailing-edge GPUs quickly become obsolete for training, they remain viable for inference much longer.

Architectural Innovations and Shifting GPU Viability

A single architectural innovation can reshape which GPUs are viable for inference tasks. DeepSeek's Multi-Head Latent Attention (MLA) highlights this dynamic, reducing KV cache requirements by over 90% and fundamentally changing inference bottlenecks.

By shrinking KV cache memory demands, MLA shifts short and medium-context inference tasks from being memory-bound to increasingly compute-bound. Lower memory demands mean GPUs spend less time waiting for data retrieval and more time on actual computation, significantly increasing GPU utilization rates. For China's AI ecosystem, this unlocks substantially more inference throughput from trailing-edge GPUs.

Custom optimizations further amplify these benefits. DeepSeek has demonstrated that Huawei's domestically-produced Ascend 910C can achieve 60% of Nvidia's H100 inference performance through targeted optimizations. This showcases how software and architectural innovations continually reshape the viability and relative strengths of different GPUs for AI workloads.

MLA renders short- and medium-context inference tasks far more efficient by reducing memory bottlenecks, allowing cutting-edge GPUs to fully leverage their computational power. While this widens the performance gap between cutting-edge and trailing-edge GPUs, it also increases China’s overall inference capacity by making older hardware more efficient. Leading-edge GPUs like the H100 will continue to dominate compute-bound workloads, but MLA significantly boosts the total inference power that can be extracted from China’s existing GPU stockpile.

For long-context inference, the hardware calculus shifts again. When context length becomes sufficiently large, tasks remain memory-bound even with MLA, reducing the performance advantage of cutting-edge hardware over trailing-edge hardware for these specific workloads. Long-context inference tasks are particularly important for reasoning, agentic AI, and automated research applications. The capacity of trailing-edge hardware to support these AI capability-enhancing tasks strengthens China’s ability to advance AI progress despite hardware constraints on cutting-edge GPUs..

Implications for Export Controls

The implications for export controls are significant: inference capacity is growing across the board, and restrictions on cutting-edge hardware won’t prevent China's inference capacity from expanding. Cutting-edge GPUs will retain significant performance advantages for short and medium-context workloads, but trailing-edge hardware remains surprisingly effective for long-context inference where memory constraints persist.

The prolonged viability of trailing-edge GPUs for inference extends the lifespan of China's existing hardware stockpile. Even as export controls limit China’s access to cutting-edge AI accelerators, China’s large stock of A100, A800, and H800 GPUs remains useful for inference applications far longer than they would for training. This sustains China's AI infrastructure and boosts its inference capacity despite limits on acquiring new chips.

Moreover, China has developed indigenous AI chips capable of inference. Huawei's Ascend 910C has demonstrated competitive performance for inference workloads. Notably, the Ascend 910C’s yield rate has doubled since last year to 40%, and Huawei plans to produce 100,000 units of the 910C and 300,000 units of the 910B in 2025, signaling a significant expansion of domestic chip production. Biren Technology's BR100, a 7nm, 77-billion transistor GPU, rivals the A100 for both training and inference. China’s growing production of inference-viable chips, substantial stockpile of trailing-edge GPUs, and continued access to the H20 reinforce its ability to sustain AI capabilities in an inference-heavy AI paradigm despite restrictions on acquiring cutting-edge hardware.

The Hardware Multiplier: China’s Inference Efficiency

Beyond hardware access, China’s advances in inference efficiency have significant strategic implications for U.S. export controls. DeepSeek’s recent innovations — particularly its v3 and R1 models — demonstrate China’s ability to push the frontier of inference efficiency. By implementing innovative techniques like a sparse Mixture of Experts architecture, multi-head latent attention, and mixed precision weights, DeepSeek’s R1 model achieves approximately 27x lower inference costs than OpenAI’s o1 while maintaining competitive performance.

This efficiency advantage effectively counterbalances U.S. hardware restrictions. Even if export controls limit China to 15x less hardware capacity, a 30x inference efficiency advantage would enable China to run nearly twice as much inference as the U.S. This acts as a multiplier on China’s hardware base, potentially giving China greater total inference capacity despite hardware restrictions.

The efficiency gains extend the utility of trailing-edge GPUs in China’s AI ecosystem, as improved inference efficiency compensates for computational and memory limitations. While DeepSeek’s achievements are a continuation of the observed decline in inference costs, this case demonstrates that Chinese AI labs have already developed the expertise to push the frontier of inference efficiency and could choose to withhold future breakthroughs if strategic considerations change.

The Sleeping Dragon: China’s Compute Overcapacity

Additionally, China’s massive but fragmented compute ecosystem is structurally better aligned with inference requirements than training needs. The aggressive GPU stockpiling during China’s “Hundred Model War” of 2023 created substantial compute capacity that became underutilized as many firms abandoned their foundation model ambitions. As Alibaba Cloud researcher An Lin observed, many of China’s claimed “10,000-GPU clusters” are actually collections of disconnected GPUs distributed across different locations or models. While this fragmentation makes the infrastructure suboptimal for training frontier models, it remains viable for inference workloads that can run effectively on smaller, distributed clusters.

Open-source models are particularly well-positioned to leverage this distributed infrastructure, enabling deployment across China’s fragmented GPU ecosystem and transforming previously idle compute into a strategic asset for widespread inference. This approach allows companies to preserve limited high-quality compute for model development while unlocking latent compute capacity.

China’s once-idle compute resources are increasingly valuable in an inference-heavy AI landscape, improving China’s position along both the innovation and diffusion axes.

How Should the U.S. Respond?

An inference-heavy AI paradigm favors China’s AI innovation potential. Its access to inference-viable hardware, leading inference efficiency, and compute overcapacity function better in an inference-driven context than in a pre-training one. U.S. export controls, designed to constrain training, have been less effective at limiting inference. China’s inference capacity remains underestimated. Despite restrictions, access to trailing-edge GPUs, stockpiles, domestic chips, and H20s enable continued progress.

As inference becomes central to AI competition, China’s relative position strengthens, narrowing the U.S. advantage. This shift demands a strategic recalibration: the U.S. must reinforce export controls and develop a coherent open-source AI strategy.

Restricting Exports of the NVIDIA H20

Export controls on AI hardware operate with a lag — typically one to two years before their full impact materializes. This lag effect is central to understanding both current policy outcomes and future strategic decisions for export controls.

Some cite DeepSeek’s latest models as proof that U.S. export controls have failed. However, this outcome is a shortcoming in how the controls were initially calibrated rather than a failure of the broader strategy. The Biden administration initially set narrow thresholds—based on FLOPs and interconnect bandwidth — which NVIDIA circumvented with the H800, designed specifically to remain exportable to China. When controls finally expanded to include the H800 in October 2023, Chinese companies had already stockpiled these GPUs in addition to speculated H100s and H20s, allowing them to maintain frontier development and delaying the policy's actual impact.

This lag highlights how AI hardware and model lifecycles can stretch over many months, so chips purchased immediately before or soon after a policy shift can remain in service for a long time. Consequently, the policy’s full impact may not be evident right away. As older hardware loses its edge for training and frontier development scales, the impact of controls becomes realized through constraints on both the speed of a country’s AI advancement and the extent of its diffusion.

The lag effect of export restrictions is more pronounced for inference hardware. Unlike training, inference workloads can remain viable on older GPU generations for much longer periods, as they depend more on memory capacity and bandwidth than raw compute power. If the U.S. delays restricting inference-oriented chips like the H20 until inference becomes even more central to AI power, the extended lag could substantially weaken the effectiveness of export controls as a defensive measure. By restricting the H20 now, the U.S. can meaningfully limit China’s accumulation of inference hardware before inference becomes the dominant compute paradigm in AI. The sooner these revised controls take effect, the sooner they will impose measurable constraints on China’s ability to compete along both axes of AI competition.

A Strategy for Open-Source AI

Open-source AI is a key vector of competition that requires a strategic U.S. approach. While it fuels innovation, not all models or circumstances warrant taking the same open approach. Open-sourcing an advanced model represents a form of technology transfer to China if that model exceeds the AI capabilities that China has access to. This reduces the U.S. lead on the AI innovation axis, shifting competition toward the diffusion axis — an area where China may be better positioned to compete.

As the compute requirements for pre-training grow, open releases help China overcome its pre-training disadvantage while amplifying the role of inference, where China is stronger. If not managed strategically, open-source AI could accelerate China’s ability to close the gap in both innovation and diffusion. The U.S. must assess whether it retains an edge in leveraging open models for research, application, and deployment. If so, open-source strategies can reinforce leadership; if not, they risk eroding it.

To assess the impact of an open release on U.S. tech competitiveness, we should evaluate how much of an immediate advantage the U.S. is foregoing on the AI innovation axis by open-sourcing a model and compare that to the net effect of how well the U.S. and China can convert open access into gains across both axes. If the U.S. retains a structural advantage in furthering AI research, building applications, fine-tuning, and scaling AI deployment, then open-source strategies can reinforce U.S. leadership. However, if China is more effective at leveraging open models for research, real-world adoption and economic or military applications, then unrestricted open release could benefit China more. This dynamic underscores the need for a structured approach and collaboration between private and public sector regarding deployment decisions.

The Bottom Line

As trends in AI elevate the importance of inference, the U.S. must reassess its strategy to lead along both axes of AI competition. While early export controls are designed to constrain China’s ability to train frontier models, they are less effective in limiting its capacity for large-scale inference. To sustain its competitive edge, the U.S. must expand export controls to address the growing role of inference, particularly by restricting chips like the NVIDIA H20 before their strategic importance escalates further. At the same time, the U.S. must refine its approach to open-source AI, ensuring that its diffusion benefits reinforce, rather than undermine, U.S. national AI leadership. Winning the AI competition requires adapting as fast as the technology evolves, and this is a critical moment for the U.S. to recalibrate its strategy.

It’s been a tough week for the international order. It feels like every TV in every restaurant across Taiwan is blasting nonstop coverage of the Trump-Zelenskyy fallout.

How will Taiwan respond to Trump’s pivot to Putin? Would Taiwan be safer with nuclear weapons? What platforms do Taiwanese people use to debate about politics anyway?

In today’s roundup, we’ll analyze perspectives from Taiwanese legacy newspapers, social media firestorms, and viral political influencers.

Driving Solidarity

We’ll start off by highlighting some reactions on the most popular Taiwanese social media platform, PTT.

PTT is a bit like a Taiwanese version of Reddit. The key difference is that comments are always displayed in chronological order instead of being ranked by popularity. Users can “push” 推, “boo” 噓, or reply to comments to express their opinion. The platform shows whether each comment is being “pushed” or “booed” overall, but doesn’t display the total vote tallies. Like on Reddit, there are sub-forums for topic-specific discussion.

Disclaimer: these forums are hosting open debates with intense back-and-forth between commenters. I’ll be highlighting recurring themes, as well as arguments where both sides are earning push-votes, but I want to be clear that there is no broad consensus on what the Trump-Zelenskyy fallout means for Taiwan at this point.

For example, the following debate emerged below a Mandarin translation of the Foreign Affairs article entitled “Ukraine Will Not Surrender to Russia”:

(Pushed) I support and praise this article, justice will prevail.

(Booed) Then how come there are no soldiers? Conduct an opinion poll or something.

(Pushed) These past couple of days, I've seen quite a few people claim that Ukraine should have originally surrendered to Russia in exchange for peace and prosperity. This kind of argument completely ignores the suffering Ukraine endured under Russian rule in the past.

(Pushed) In the past, we thought that people in democratic countries feared death more than other people — but Ukrainians are not afraid.

(Pushed) The Uyghurs will never surrender, but they will not go to the front line

(Pushed) It is 100000000% reasonable to be suspicious that Trump received personal benefits from Russia or made a blood pact with Russia.

The Taiwanese transliteration of “Zelenskyy” is 澤倫斯基 Zélúnsījī, and in casual writing Taiwanese people refer to him by the nickname 司機 Sījī (literally, “The Driver”) which has the same pronunciation as the last two characters of the transliteration.

From a thread in a military forum about whether Zelenskyy overplayed his hand:

(Pushed) The driver really shouldn’t have talked back to Vance. If he wanted to argue, he could have done it in private.

(Pushed) After apologizing, you still have nothing, so why bother apologizing?

(Pushed) If Little Z doesn’t kneel, America will make explosive corruption accusations against him.

(Reply) East Asian countries are better at licking.

(Pushed) If Ukraine wants to thank someone, it should thank the previous Biden administration. Why thank Trump?

(Pushed) It seems someone is trying to smear and destroy Mr. Z's image. Be careful when responding to this thread.

Indeed, there are signs of disinformation in some discussions of this topic. An FT article entitled “Zelenskyy rejects calls for immediate Ukraine-Russia ceasefire” was posted on PTT with the mistranslated title, “The Driver Rejects Ukrainian and Russian calls for a Ceasefire” (司機拒絕烏克蘭與俄羅斯立即停火的要求), a fact which was quickly pointed out and mocked in the comments.

Marco Rubio is well-known in Taiwan thanks to his long congressional record of support for the island. Here are some comments about him:

(Pushed) Rubio will be replaced soon.

(Pushed) Rubio was once a pioneer in anti-communism, but now he bows down to power.

Underneath an article reporting Trump’s plan to freeze aid to Ukraine in response to the meeting:

(Pushed) Stop it right now immediately!!!!!!!!!!!!!!! I’ve never seen such a cowardly U.S. president!! You truly see everything if you live long enough!!!!!!!!

(Pushed) Will the European big brothers shoulder some of the responsibility? Isn’t this an opportunity for them to show off?

In a financial forum:

(Pushed) Being pro-China is selling out Taiwan, being pro-America is also selling out Taiwan.

(Pushed) In the Budapest Agreement, even China said it would protect Ukraine, but that isn’t happening

(Pushed) Ultimately, [Ukraine] should not have given up its nuclear weapons. Security guarantees are bullshit.

(Booed) Ukraine has no nuclear bombs, so of course it has no bargaining chips.

(Pushed) The driver’s bargaining chip is making the king (Trump) lose face.

(Pushed) Buddha’s mercy 佛祖慈悲 [This phrase is used ironically in situations that are cruel or corrupt to the point of hopelessness.]

Ukraine Today, But Taiwan’s OK?

At the start of the invasion, the DPP popularized the slogan, “Ukraine today, Taiwan tomorrow.” Editor Gu Shu-ren 辜樹仁 of CommonWealth Magazine 天下雜誌 (a Taiwanese publication similar to the Atlantic), addressed fears that Trump will abandon Taiwan after Ukraine in a recent editorial:

Looking back at history, Taiwan's strategic value to the United States has been the key factor in America's decision to either abandon or support Taiwan.

In 1950, when the Korean War broke out, the Republic of China (ROC) government, which had retreated to Taiwan and was on the brink of collapse after being abandoned by the U.S., suddenly became the central hub of the U.S. first island chain strategy in East Asia — a so-called unsinkable aircraft carrier — greatly increasing Taiwan's strategic importance.

In the 1970s, as the U.S. aligned with China to counter the Soviet Union, Taiwan lost its strategic value, leading to the severance of U.S.-Taiwan diplomatic ties and the withdrawal of U.S. troops from Taiwan. …

Today, Taiwan's strategic value to the United States is at its highest since the servering of diplomatic ties, as the primary battleground in the U.S.-China rivalry is now the technology war, with semiconductors at its core. More specifically, TSMC is the most crucial asset for the U.S. in securing a supply of advanced chips and revitalizing its semiconductor manufacturing industry. If the U.S. wants to maintain its technological and military lead over China, it must firmly keep Taiwan within its grasp. …

Ensuring that the U.S. remains dependent on Taiwan’s advanced chip manufacturing — making American national security synonymous with protecting Taiwan — is the most critical factor in maintaining Taiwan’s strategic value to the United States.

Of course, there is another equally important factor. Trump dislikes war, especially costly military interventions where the U.S. cannot be assured of victory. He has repeatedly complained that Ukraine failed to prevent war at the outset. Therefore, avoiding war at all costs is also a key strategy for Taiwan to secure Trump’s support.

Only through this can tomorrow’s Taiwan avoid becoming the Ukraine we saw today.

Reporter Jiang Liangcheng 江良誠 similarly warned that Taiwan would need to become more transactional in its relationship Trump:

“Trump's only vocabulary is actually "money, money, money". All international relations can be measured by money. There is no free lunch in the world. It is impossible to ask Americans to help you defend your country like a plate for free and without any reward. …

However, when it comes to Taiwan's policy toward the United States, Lai Ching-te still sticks to Tsai Ing-wen's international politics, such as the first island chain, geopolitics, and Indo-Pacific security. I'm afraid even Trump doesn't understand these terms.”

The Meihua News Network (梅花新聞網), a Pro-China news outlet owned by a controversial Taiwanese religious leader, argued instead that Taiwan needs to reopen dialogue with Beijing given the reality that the U.S. is an unreliable partner.”

In front of cabinet members and the media, Trump was unwilling to guarantee that the Chinese Communist Party would not invade Taiwan by force during his term, and emphasized that he had a good relationship with Chinese Communist Party leader Xi Jinping. …

“Foreign Affairs” recently published a special article titled “The Taiwan Fixation: American Strategy Shouldn’t Hinge on an Unwinnable War”, co-authored by Professor Kavanagh of the Georgetown University Center for Security Studies and senior scholar Wertheim of the Carnegie Endowment for International Peace. The gist of the article is: Taiwan is certainly valuable to the United States, but if American decision-makers overestimate Taiwan's importance, they will sacrifice the security of maintaining the status quo due to the risk of endless and destructive war; and Taiwan's importance is not enough for the United States to sacrifice tens of thousands of American lives to protect it. Former National Security Council Secretary-General Su Chi 蘇起 described this article as the most powerful article to date advocating the United States to let go of Taiwan. …

Apart from fully relying on the American security umbrella and turning Taiwan into a "porcupine," the DPP also has another option: restoring cross-strait communication and reducing tensions in the Taiwan Strait. If that happens, the so-called "Abandon Taiwan Theory" would naturally dissipate. Rational decision-making should not be obstructed by anti-China or China-hating sentiments.”

By contrast, a popular post from the Taiwanese political influencer James Hsieh argued that Taiwan should be doing whatever it takes to improve relations with the U.S., not criticizing Trump’s Ukraine policy:

“I still see many people online going against the tide, bashing Trump, criticizing the U.S., and supporting all kinds of conspiracy theories.
Here are five reminders:

1. Before the war, Ukraine was extremely pro-China, selling major military technology to China. Just a few days ago, Ukraine even asked China for help.

2. Morally, we must oppose aggression, but in terms of international strategy, we must firmly support the United States.

3. Taiwan is not Ukraine. In terms of historical ties with the U.S., the Taiwan Relations Act, geographical location, type of warfare, and economic strength, Taiwan is completely different. Taiwan is absolutely not a distant European country like Ukraine in America's eyes. Comparing Ukraine to Taiwan is a completely flawed analogy. Saying that the U.S. pulling out of the Russia-Ukraine war implies that it will betray Taiwan is just another favorite conspiracy theory of the dumb lefties (左膠) and the Chinese Communist Party’s propaganda machine.

4. Personally, I hope the Russia-Ukraine war ends quickly so that the U.S. can fully prepare for the Indo-Pacific. This is a practical concern, as China is rapidly advancing its strategic plans. How the U.S. swiftly ends its engagements elsewhere and refocuses on the Indo-Pacific is critical. Just yesterday, Vice President Vance stated that the U.S. military-industrial production can no longer sustain the continuous supply of heavy weaponry to Ukraine.

5. History has shown that during major wars, opportunistic nations take advantage of a great power’s exhaustion to invade smaller neighboring countries. If the Russia-Ukraine war escalates into World War III and the U.S. and Europe are preoccupied with fighting Putin’s alliance, it would be the perfect moment for China to seize Taiwan under the guise of maintaining stability.

If Taiwan's democracy, freedom, and independence from oppression are what you value most, then Taiwan should prioritize its relationships with the U.S. and Japan over everything else — not Ukraine.

Only the U.S. and Japan will help us. Survival comes first before ideals.

Taiwan-U.S. friendship!”

It remains unclear what the Lai administration’s approach will be, but you can be sure that ChinaTalk will keep monitoring the debate as it evolves.

Subscribe now

Will Taiwan Get Nukes?

Zelenskyy’s White House press conference also reignited the old debate about whether Taiwan would benefit from having its own nuclear arsenal. Taiwan abandoned its indigenous nuclear program in response to pressure from the U.S., much like how Ukraine relinquished its nuclear weapons to Russia after the fall of the USSR. Taiwan was estimated to be just two years away from completing a WMD when the U.S. intervened in 1988.

These parallels were drawn explicitly by a CNN profile of Colonel Chang Hsien-yi 張憲義, the Taiwanese nuclear engineer who provided intelligence about Taiwan’s proliferation plans to the CIA. The article was repackaged, translated, and published on the front page of the China Times on Monday.

On PTT, the profile drew comments like:

(Pushed) This person is the reason why Taiwanese independence is impossible.

(Pushed) Nuclear weapons are not something that Taiwan's extremely incompetent politics could handle. If nuclear weapons were in the hands of Chiang Kai-Shek and his family, Taiwan would have ended up like North Korea. The Chiang family would still in power, and there would never have even been a chance for democratization. So many people have no clue what’s going on.

Taiwanese political influencer Mr. Shen 公子沈, who runs a YouTube channel with more than 700k subscribers, posted the following meme on Threads (which is way more popular in Taiwan than the U.S.) with the caption, “With nukes vs without nukes: it’s time for Taiwan to develop nuclear weapons.”

Speaking of bargaining chips…

Reactions to the TSMC Deal

TSMC’s newly announced $100 billion investment in US chip manufacturing led to more online discontent. The following comments from Facebook were curated by Angela Oung:

“Today we are all Ukrainians”

“At least Zelensky has guts”

“ASMC” [American Semiconductor Manufacturing Company]

“So they’re taking our stuff, leaving us with no cards. Think they’ll help in the future? Stop dreaming!”

“Taiwan’s remaining value is becoming a meat grinder like Ukraine.”

“He [TSMC Chairman CC Wei] looks like he has a gun behind his head. Hostage situation.”

“The silicon shield we spent decades building is being handed over by our government without a whimper”

“TSMC: built by the KMT, sold by the DPP”

“Is Lai Ching-te such a pussy that he’s not even gonna say anything?”

“Today Ukraine, tomorrow Taiwan. One step closer to refugee status.”

“Bandits…just like the CCP”

To close, I’ll leave you with another popular post on Threads expressing frustration about Taiwan’s-U.S. relations:

“The U.S. asks us to buy military equipment — we buy it.

The U.S. asks us to extend the length of mandatory military service — we extend it.

The U.S. wants TSMC — we hand it over with both hands.

The U.S. wants us to implement resilient defense — we manage to do it, even if we have to hide and shuffle the budget.

For every single thing the U.S. asks of us, from the issue of eating ractopamine pork in our daily meals to national defense policies involving regional security cooperation, Taiwan follows the U.S.’s demands without question.

But will there come a day, just like today’s Ukraine, where we sign agreements on resource concessions, trading away our country's future rebuilding assets, yet still lack the most basic “security guarantees”?

Ukraine has the support of the entire European continent—but what about Taiwan?

Will today’s Ukraine be a reflection of Taiwan’s future?

Will Taiwan, when that day comes, be even more isolated and helpless?”

To be fair, this commenter is right that Taiwanese pork is way more delicious than the ractopamine pork imported from the U.S. I sincerely hope that every ChinaTalk subscriber has an opportunity to come to Taiwan and eat stewed pork rice (滷肉飯)…before it’s too late!?

A guest piece by Afra, freelance writer and podcaster [Jordan: I highly recommend this show!] with working experience in tech and crypto. Personal site here.

DeepSeek’s winds have already been blowing for some time, but this particular gale seems to have real staying power.

President Trump characterized DeepSeek as “a wake-up call,” Scale AI CEO Alexandr Wang called it “earth-shattering,” and Anthropic CEO Dario Amodei deigned to come on ChinaTalk to discuss fears of powerful AI within “authoritarian systems of government.”

On Chinese social media, the discussions took on a life of their own, with the most popular use case being the calculation of one’s Ba Zi (八字) and astrological chart, using the social media tag “AI玄学” (AI Mysticism). Users weren’t just seeking their personal fortunes — they saw the nation’s destiny itself shifting through DeepSeek’s emergence. These conversations are a swirling mix of collective jubilation, national pride, and gleeful satisfaction over America’s “China envy,”1 often accompanied by playful banter.

Yet amidst this discourse, a deeper and more resonant question emerges: could this be a sign of China’s technological ascension? Is this evidence that Guoyun (国运) — the nation’s long-awaited destiny — has finally arrived?

First, what is Guoyun 国运？

The term 国运 combines two characters: 国 (guó, “nation/state”) and 运 (yùn, “fate/destiny/fortune”). This concept emerged from traditional Chinese cosmological thinking, where the destiny of the state was seen as intertwined with celestial patterns and dynastic cycles.2 This term, once confined to the ornate dialogue of period dramas set in imperial China, has begun to surface with increasing frequency on my social media timeline.

The Guoyun narrative around DeepSeek began when Feng Ji 冯骥, creator of the globally successful game “Black Myth: Wukong,” declared it a “national destiny-level technological achievement.”

The discourse gained momentum when Zhou Hongyi 周鸿祎, Chairperson of Qihoo 360, positioned DeepSeek as a key player in China’s “AI Avengers Team” against U.S. dominance. This sentiment echoed across media, with headlines like “Is DeepSeek a breakthrough of national destiny? The picture could be bigger” and “DeepSeek triggers U.S. stock plunge; can it really change the nation’s destiny?"

For Chinese netizens, discussions about politics on social media are often marked by subtlety and veneration with trepidation (for reasons that require little explanation). However, during the 2025 Chinese New Year, the discourse expanded far beyond politics and DeepSeek into a cacophony of cultural euphoria —a wave of self-congratulatory enthusiasm that evolved into something larger culturally. This included the movie Nezha 2, which shattered box office records and surpassed Inside Out 2 to become the highest-grossing animated film of all time (with patriotism-fueled consumption boosting the box office performance), TikTok refugees flooding Xiaohongshu, and advanced Unitree robotics performing during the Spring Festival Gala. These achievements seemed to occur against a historical backdrop where technological and cultural advances carry deeper significance about China’s rightful place in the cosmic order.3

The Guoyun discourse extends beyond tech leaders, media commentary, and social media posts.

President Xi Jinping has woven the concept of destiny into official rhetoric, though carefully stripped of its more superstitious elements. Speaking at the 19th Academician Conference of the Chinese Academy of Sciences in May 2018, Xi declared, “Innovation determines the future; reform concerns national destiny. The field of science and technology is the area most in need of continuous reform 创新决胜未来，改革关乎国运。科技领域是最需要不断改革的领域.” This statement aligns with his broader techno-nationalist vision, explicitly linking technological advancement to China’s strategic future.

A 2024 People’s Daily article discussing Xi’s thoughts emphasized that “cultural confidence is a major issue concerning national destiny 坚定文化自信，是一个事关国运兴衰...的大问题"。

This rhetorical shift signals a carefully calibrated blend of traditional Chinese concepts with modern governance — a bridge between ancient ideas of dynastic cycles and contemporary aspirations for technological supremacy.

Beyond superstition: is this a collective myth-making or post-pandemic yearning for certainty?

It would be a mistake to dismiss this discourse as mere superstition or propaganda.

The COVID-19 pandemic marked a watershed moment in Chinese society’s relationship with national destiny. To me, Zero COVID became a mirror polished to cruel clarity, reflecting a China I no longer recognized. During the rigid cycles of lockdowns and reopenings, I didn’t see my parents for two years, my grandmother was hospitalized, and my cousin was confined to his university dorm for three whole months culminating in a severe mental breakdown. Friends lost loved ones due to a lack of timely treatment options. Back then, seeing how waves of people wanted to “run (润)” from China, I thought for the first time that I might never return to China, and that I might become part of the Chinese diaspora forever.

COVID created a collective trauma that many Chinese are still processing.

But this experience has paradoxically reinforced a certain earnest faith in China’s future among ordinary citizens. The optimism in the discussion of Guoyun might represent a complex emotional response to the uncertainty and trauma from the COVID era — a blend of traditional fatalism with genuine aspirations. Having weathered the pandemic’s disruption, many ordinary Chinese seek reassurance about the future through familiar cultural frameworks. ‘National Destiny’ provides exactly that — it’s a narrative that contextualizes current struggles within a larger, ultimately triumphant story. It’s therapeutic.

The discourse around 国运论 (guóyùn lùn, or “national destiny theory”) reveals parallels to America’s historical myth-making. Perhaps the most striking similarity between China and the US is their unwavering belief in their own exceptionalism and their destined special place in the world order. While America has Manifest Destiny and the Frontier Thesis, China’s “national rejuvenation” serves as its own foundational myth from which people can derive self-confidence. Through countless repetitions across state and social media, this narrative has become deeply ingrained in China’s national consciousness.

The wounds behind techno-nationalism

Where myths nurture the national consciousness, technology has become the battleground where China’s historical narrative demands its vindication. The roots of China’s techno-nationalism run deep, drawing emotional power from China’s “century of humiliation.” U.S. actions — chip controls, the attempted TikTok ban, tariffs, investigations of Chinese scientists, and suspicions of Chinese espionage — rekindle the historical trauma of humiliation.

For decades, China has been portrayed as a mere copycat or thief of Western innovation. Each technological breakthrough now serves as vindication, a refutation of that dismissive narrative — this shame has never truly been resolved. As Kevin Xu elaborated on DeepSeek’s open-sourced nature, “It’s all for the validation and approval,” — a sharp acknowledgment that when Chinese engineers share their code with the world, they’re not just demonstrating technical prowess but seeking to heal a wound in the national psyche:

In the Chinese open source community, there is this thing that I would call open source “zeal” or “calling” (开源情怀)

Most engineers are thrilled if their open source projects — a database, a container registry, etc-- are used by a foreign company, especially a silicon valley one. They’d tack on free labor on top of already free software, to fix bugs, resolve issues, all day all night. It’s all for the validation and approval.

Implicit in this “zeal” or “calling” is an acute awareness that no one in the West respects what they do because everything in China is stolen or created by cheating. They are also aware that Chinese firms have been taking for free lots of open source tech to advance, but they want to create their own, contribute, and prove that their tech is good enough to be taken for free by foreign firms -- some nationalism, some engineering pride.

So if you want to really understand why DeepSeek does what it does and open source everything, start there. It’s not a political statement, not to troll Stargate or Trump inauguration, or to help their quant fund’s shorts on NVDA (though if that were the case, it’d be quite brilliant and savage)

The drive to prove oneself on behalf of the nation is expressed vividly in Chinese popular culture. I couldn’t stop thinking about Illumine Linga (临高启明), an open-source collaborative novel that has captivated China’s engineering community and become a phenomenon of its own. The story follows modern Chinese engineers who time-travel to the declining Ming dynasty, right before China was conquered by the Manchus, bringing industrial equipment and technical knowledge. They gradually industrialize Hainan and Guangdong provinces before expanding outward with the ultimate goal of establishing global hegemony.4

Though ostensibly just fiction, Illumine Linga pulses with the heartbeat of China’s “Industrial Party” (工业党) — that loose constellation of engineers, programmers, and technically-minded patriots united by an almost religious faith in technology as destiny’s instrument. The novel serves as a sharp allegory for contemporary aspirations: technological mastery as the path to national resurrection and global respect.5

In the Western intellectual tradition, technology and data have undergone phases of detached scrutiny — viewed first as tools of emancipation, and later as vectors of control. Foucault’s panopticon mutated into Zuboff’s surveillance capitalism; Wiener’s Cybernetics birthed both Silicon Valley and Snowden’s disclosures. This academic back-and-forth assumes a fundamental premise: technology can theoretically exist as a neutral substrate awaiting ideological imprint.

However, in my impression, China’s techno-discourse never evinces such “purity.”

From its inception, technology has been semantically encased in the shell of techno-nationalism. In China’s history textbooks, Qian Xuesen’s missiles for the Two Bombs, One Satellite program were never just missiles, but brushstrokes in the narrative of “standing up again.”6 Yuan Longping’s hybrid rice strains didn’t merely feed millions; they were genetic correctives to the “Century of Humiliation,” each harvest a quiet refutation of the colonial-era belief that China couldn’t innovate.

On Chinese New Year’s Eve, a fake response to the “national destiny theory” attributed to Liang Wenfeng circulated widely online, with many believing and sharing it as authentic. This response claimed that DeepSeek’s open-source decision was merely “standing on the shoulders of giants, adding a few more screws to the edifice of China’s large language models,” and that the true national destiny resided in “a group of stubborn fools using code as bricks and algorithms as steel, building bridges to the future.” This fake statement—notably devoid of wolf warrior rhetoric—spread virally, its humility and relentless spirit embodying some values people hoped Chinese technologists would champion. Meanwhile, the real Liang Wenfeng remained silent after DeepSeek’s rise. A month later, he appeared on CCTV sitting beside Tencent’s Ma Huateng at Xi Jinping’s symposium for top business leaders.

The public’s fascination with Liang showed no signs of waning. In Silicon Valley, his previous interviews were swiftly translated into English and meticulously analyzed, while in China, his rise also inspired mystical interpretations—during the Spring Festival holiday, Liang Wenfeng’s ancestral home in Zhanjiang, Guangdong transformed into an impromptu tourist attraction, drawing feng shui masters eager to study the geomantic properties of his family residence.

Humans have always sought ways to calculate the incalculable. Perhaps that’s what makes the conversation around Guoyun so captivating: it’s not just about predicting the future, but about sense-making in China’s present.

Gary Wang spent the past decade developing business and product strategy for Silicon Valley technology companies, with a focus on enterprise software, the industrial internet of things and AI. He has a degree from HKS and worked in China. The views expressed here represent only his own.

About a decade ago, the best forecasts heralded a promising manufacturing future, in the United States and globally, with the advent of the fourth industrial revolution (also called “industry 4.0,” the “industrial internet,” or “industrial internet of things” aka IIoT). The belief was that the falling cost of cloud computing, sensor costs, and machine learning — coupled with new connectivity technologies such as 5G or IPv6 — would lead to a revolution in manufacturing productivity and ultimately higher GDP growth.

Despite these promising forecasts, multiple data points indicate that US manufacturing has largely stagnated. Analysis from the New York Federal Reserve reveals that both total factor productivity and labor productivity have been flat from 2007 to 2022. Meanwhile, US share of global manufacturing value add fell from nearly 25% in 2000 to an estimated 15% today in 2024. The UN Industrial Development Org projects US share of global manufacturing value add will fall to 11% in 2030, while China may account for 45% of global output.

This decline comes after multiple presidential administrations’ efforts to revitalize American manufacturing — from the Obama-era policies such as the Advanced Manufacturing Partnership or the Manufacturing USA initiative, to the Biden administration’s Inflation Reduction Act, and now the Trump administration’s desire to reshore manufacturing via tariffs and other policy tools.

Off-shoring and free-trade agreements go only so far in explaining this decline. And the present debates over US industrial policy — sparked by the advent of emerging technologies (generative AI, quantum computing) as well as intensifying competition with China — perhaps focus on the wrong things.

The real questions US policymakers must grapple with: why did the United States fail to capitalize on technology that was already available to make its manufacturing base more competitive?

Put another way: why have the promises of the IIoT revolution failed to materialize in the United States?

This piece makes a few key arguments:

• The “industrial internet of things” is not an industry. It’s a set of disparate technologies that all need to be adopted together to create value.

• The free market will not always optimize adopting a broad set of technologies for an entire ecosystem of industries. The underwhelming results of today’s industrial internet is a case in point.

• China’s industrial policies to “win” the fourth industrial revolution offer lessons for policymakers in the United States to consider.

• When it comes to revitalizing manufacturing, or ensuring American leadership in AI or quantum computing, policymakers need to craft policies to develop entire value chains and tech ecosystems — not myopically focus on just one strategic technology (eg. advanced semiconductors).

  ChinaTalk is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.

What is IIoT?

IIoT refers to the interconnection of machines, devices, sensors, and systems which are connected on the internet in performing industrial tasks.

Take one of IIoT’s leading “use cases” (ie. applying tech to solve a business problem): predictive maintenance. Sensors connected to a piece of factory equipment, such as a boiler, can measure temperature or vibration. When combined with machine-learning algorithms, manufacturers stand to save millions by predicting when a machine would fail, and then proactively maintaining the machine before a failure occurs — thus reducing factory downtime and increasing productivity.

Another use case: gathering GPS data from truckers could enable machine-learning algorithms to optimize the routes of commercial trucks (saving fuel costs). When paired with data on customer demand (say, Pepsi sales in a city), manufacturers could save billions by optimizing their inventory costs to ensure that the optimal amount of Pepsi reached store shelves at just the right time.

The use cases are endless: deploying robots on the production line, using cameras and AI to automate quality inspection for finished products, creating a “digital twin” of an entire production process for optimization, and much more. All of these use cases required cloud computing, real-world historical data, and connectivity. As a practitioner who has worked with technology companies on their strategy for delivering industrial IoT to manufacturing companies, I can attest to the level of industry enthusiasm for IIoT during this time (as well as the numerous operational challenges).

How off were the IIoT forecasts?

In 2015, McKinsey forecast $1.2 to 3.7 trillion in economic value created per year by 2025 from IoT technologies in factories. Assuming technology vendors alone capture 5% of the value created — a very conservative benchmark — that’s $60 to $185 billion in revenue. The International Data Corp in 2017 forecast that manufacturers would spend $102 billion in the industrial internet, meaning vendors selling IIoT technologies should see comparable revenue figures. Accenture and World Economic Forum joined the hype, intoning that the “Industrial Internet will transform many industries, including manufacturing, oil and gas, agriculture, mining, transportation and healthcare. Collectively, these account for nearly two-thirds of the world economy.” These market forecasts led the Congressional Research Service in 2015 to predict, “The current global IoT market has been valued at about $2 trillion, with estimates of its predicted value over the next 5 to 10 years varying from $4 trillion to $11 trillion.”

These forecasts were off by multiple orders of magnitude. Today, to my knowledge, there is only one publicly listed company in the United States solely focused on IIoT: Samsara, with $1.4 billion in revenue, growing at a healthy ~40% year over year. (Palantir in 2024 reported $700 million in revenue from US private-sector firms, some of which include manufacturing — but the majority of Palantir’s business is with governments.)

General Electric and Siemens both tried to become technology companies by developing their own cloud platforms and AI applications to digitize the manufacturing sector. A series of New York Times headlines, though, tells the saga of GE’s attempt to capture the purported massive opportunity of the industrial internet of things:

• 2011: GE Wants Software for the ‘Industrial Internet’

• 2012: Looking to Industry for the Next Digital Disruption

• 2013: General Electric Adds to Its ‘Industrial Internet’

• 2015: GE Navigates, Carefully, the Industrial Internet of Things

• 2016:  GE, the 124-Year-Old Software Start-Up

• 2018: GE Makes a Sharp ‘Pivot’ on Digital

And finally, later in 2018:

Siemens positioned its industrial internet cloud platform, Mindsphere, as its next growth vector. Today, Mindsphere has been rebranded to “IoT insights hub,” and the last time Siemens company leadership talked about Mindsphere on their earnings call with equity analysts was in 2022, indicating a retrenchment in expectations (unlike when they spoke about Mindsphere on earnings calls with analysts in 2015,  2016, 2019, and 2020; what industry leaders tell Wall Street indicates where they think their companies’ growth will come from).

Why were the predictions so wrong?

IIoT is a cluster of disparate technologies that have to work together to create value. It’s not one technology. Consider the aforementioned predictive maintenance use case. To realize value, a factory owner needs to adopt six or seven different technologies from different vendors.

• There’s the company providing sensors (sometimes with software) for the machines to gather data for analytics.

• Many factories have historically not been connected to the internet, so a company like Verizon needs to get involved to set up an in-plant 5G connectivity network. (Leading analysts have estimated there are only a handful of 5G industrial projects in the United States, compared to likely thousands in China.)

• A company like Cisco has to provide the networking equipment to enable internet connectivity in the factory.

• A cybersecurity company needs to ensure the sensors and machines, now that they’re connected to the internet, are secure from cyberattacks.

• A cloud-computing company, such as Microsoft or Amazon, needs to provide the compute and storage for the customer to develop AI algorithms to analyze the data generated by the sensors. These cloud-computing companies often provide the AI algorithms for customers to customize themselves (assuming they have the in-house data science talent) to analyze the data from factory equipment.

• A company needs to integrate these disparate systems together — usually a system integrator like Accenture or Wipro.

The factory owner has a finite budget, must negotiate with six different vendors (each with their own pricing and profit models, none of whom necessarily coordinate their selling activities) — but still must realize a high enough return on investment (ROI) to justify solving this one use case. Imagine a consumer buying a car — but instead of buying from an OEM like Tesla or General Motors, you have to negotiate individually with the tire company, the engine manufacturer, the seat belt maker, the company making the infotainment display, and every other component manufacturer.

The nature of the physical world makes this coordination problem even more complex:

• Algorithms aren’t immune from false positives. What happens if the algorithms incorrectly predict a machine will break down, but a maintenance technician has already been dispatched to make repairs? That reduces ROI.

• Machine algorithms need to be trained on historical data of when the machine has broken down before — but for many factories, maintenance records aren’t digitized; if available at all, they’re paper logs of when a technician fixed a machine.

• Third, from the perspective of the technology vendor, sales cycles to manufacturers often are usually one to two years — since customers will pilot the technology for one set of machines (one use case) in one factory, measure the cost or productivity savings, and then decide whether they want to scale the technologies to multiple use cases across multiple factories. Factory budgets are managed locally, not globally — meaning a vendor has to sell to a manufacturer’s factory site in, say, the United States, then Brazil, then Germany, and so on.

All of these factors help to explain why venture capitalists — with few exceptions — have not invested in startups tackling industrial IoT, as well as why it’s been hard for existing vendors to scale their business. Even McKinsey admitted in 2021, “To date, value capture across settings has generally been on the low end of the ranges of our estimates from 2015, resulting from slower adoption and impact. For example, in factories, we attribute the slower growth to delayed technological adoption because many companies are stuck in the pilot phase.”

What has China done?

While IIoT hasn’t lived up to its potential in the United States and elsewhere in the West, China has leaped ahead in the fourth industrial revolution: there is no other country in the world that can boast of legions of “dark factories” — ie. factories where entire manufacturing processes are automated.

How has China done it? By focusing on technical challenges and market-coordination problems.

First: Chinese policymakers at the highest level — eg. the State Council — crafted policies to solve known technical challenges which threatened to hold back Chinese manufacturer’s adoption of IIoT technologies.

For example, in the predictive maintenance use case, there is a known problem of “asset mapping” — ensuring all the physical and digital assets in a factory can be identified in a common taxonomy to enable machine-learning analytics and then workflow automation (sending a technician to repair a robot, changing the workload of robots working together if one robot is breaking down, etc.). Specifically, if factory owners want to predict when a robot arm will break down, they need a comprehensive way to uniquely identify the specific robot, the specific arm of that robot, the specific sensor that may be attached to the robot, the specific 3D model of the robot’s arm, and then map all of these physical and digital assets together. Without a common taxonomy, it’s impossible to automate the analysis of sensor readings from the robot arm (eg. its grip strength) and then trigger a workflow to fix the robot arm while enabling the production process to continue seamlessly, that is, in a “lights out” fashion.

China’s State Council, in a 2017 planning document — “Guidance for Deepening the Development of the Industrial Internet ‘internet + promoting manufacturing” 深化“互联网+先进制造业” 发展工业互联网的指导意见 — specifically called for implementing networking connectivity and “identity resolution system” 标识解析体系 to solve this problem, using a combination of known technologies and standards such as IPv6, software-defined networking, 5G connectivity, time-sensitive networking, and passive optical networking. The technologies mentioned in this document were available in China (and the United States) in 2017. An identity resolution system (the English equivalent term would be a digital “tracking system”), when combined with advanced networking technologies, solves this predictive-maintenance problem because then a piece of software — such as a predictive-maintenance application for robots — can automatically locate the robot arm that’s emitting sensor data indicating a breakdown, match that to the 3D model that specifies how the robot arm should function, detect issues with the robot arm, and then trigger a workflow to remediate. Dozens of physical and digital systems are involved in solving this problem.

Of course, the free market can solve this problem as well — but it runs into the same issue mentioned above: coordination of multiple vendors with multiple technologies and standards that all have to work together. No wonder that, in 2024, 5G adoption in the US manufacturing sector was at 2%. After all, a factory doesn’t realize any business value from just deploying 5G by itself, if the rest of the technology stack (sensors, algorithms, applications, cloud computing, security, etc.) isn’t also deployed.

Second: China targeted industrial policy to solve known market-coordination problems that would hold back IIoT adoption.

For example, consider the problem of sub-scale platforms. To better understand what this is, I’ll first lay some foundation on key terms:

A platform is any technology in which an underlying resource, such as computing power (eg. Amazon Web Services), is offered to customers as a software component to build a fully functional piece of software. In the IIoT case, “industrial internet of things platforms” are cloud platforms that allow manufacturers to (1) access compute and data storage, (2) enable data to be sent from physical machines to the cloud, and (3) secure the network and data from machine to cloud. An IIoT application is a packaged piece of software with algorithms and an end-user interface that solves a business problem.

The consumer analogy is how the iPhone is a platform and Google Maps is the application that runs on the platform, using its compute and storage. Manufacturers need the IIoT platform, and they must either (1) build the IIoT application themselves (which is difficult since manufacturers often don’t have the in-house talent), or (2) buy a prepackaged application from a vendor.

The sub-scale platform problem occurs when, in a market, there are too many platform vendors who can’t make enough money to scale their business due to intense competition and operational execution issues (identified above) and when there aren’t enough applications to actually create value for the customer, the manufacturer. The IIoT market in the United States has faced precisely this problem, especially because digital-platform markets tend toward winner-take-all or oligopoly competition dynamics (eg. iPhone vs. Android; the four major cloud-computing platforms: Amazon, Google, Microsoft, and now Oracle), and platforms make money only if application vendors build on the platform.

BCG, in a 2017 report titled “Who Will Win the IoT Platform Wars,” identified over 400 IoT platforms in the market due to the excitement of the industry at that time. But few of these platforms really grew to any significant scale, with some notable failures (see GE’s attempt above) because of the technical and operational issues. As a result, there were few IIoT application vendors building prepackaged software. There too many platforms they could choose to build on, and the lack of platforms at scale meant there were too many technical challenges that were unresolved. The value of the platform is to solve the underlying technical issues so an application developer doesn’t have to. In the IT world, a software developer doesn’t have to worry about which type of server or networking equipment is in the data center to build a cloud application. The same is true for a software developer on mobile: they don’t have to worry about the specific type of camera lens on the phone when building their app.

As a result, there are few if any IIoT applications at scale (Samsara being a notable exception). For example, there is no packaged software application that a factory own can buy to predict when any robot it chooses to deploy will breakdown today, or for any other type of equipment (of which there are literally thousands) in a factory.

Meanwhile, China’s State Council, in the same 2017 policy document, designed policies to solve the sub-scale platform problem in IIoT:

By 2020, form the industrial internet platform system, supporting the construction of approximately 10 cross-industry, cross-domain platforms, and establishing a number of enterprise-grade platforms that support companies’ digital, internet-enabled, and AI-enabled transformations. Incubate 300,000 industry-specific, scenario-specific industrial applications, and encourage 300,000 enterprises to use industrial internet platforms for research and development design, production manufacturing, operations management, and other business activities. The foundational and supportive role of industrial internet platforms in industrial transformation and upgrading will begin to emerge.

到2020年，工业互联网平台体系初步形成，支持建设10个左右跨行业、跨领域平台，建成一批支撑企业数字化、网络化、智能化转型的企业级平台。培育30万个面向特定行业、特定场景的工业APP，推动30万家企业应用工业互联网平台开展研发设计、生产制造、运营管理等业务，工业互联网平台对产业转型升级的基础性、支撑性作用初步显现。

Like most industrial policies in China, the State Council’s high-level policy guidance becomes operationalized in provincial- and city-level policies via funding and other incentives. For example, Jiangsu 江苏 province set a goal of establishing 1,000 “smart” (aka enabled by cloud, AI, advanced connectivity, etc.) factory workshops in 50 provincial-level factories by 2020.

What can the United States learn?

If we’re serious about revitalizing US manufacturing or maintaining leadership in emerging technologies such as AI and quantum computing, here are some things US policymakers should consider:

1. The free market, while efficient for specific markets, may not optimize for transforming entire sets of industries. The technologies for the industrial internet of things were available in the United States — but due to technical and market-coordination challenges, adoption has lagged behind that of China. AI and quantum are foundational technologies that may require an even greater level of market coordination to overcome operational and technical obstacles compared to that of the industrial internet of things.

2. Industrial policy needs to move beyond tax incentives, tariffs, and subsidies to make calculated bets on specific technologies, with deep technical expertise incorporated early on in the policy process. For example, in AI, the policy debate has focused exclusively on semiconductor subsidies and export controls — but there is limited if any discussion on how to make the AI data center itself easier to build and operate. High energy costs and energy availability due to the limits of the utility grid are known technical and business challenges to data center capacity today. Ultimately, the total cost of using AI to make predictions, optimize processes, and create value (eg. cost of inference) is not just the cost and efficiency of the chips, but the entire data center stack, including energy costs.

3. Successful commercialization of a set of technologies creates its own positive feedback loop, which reinforces first-mover advantages. Since China has a significant head start in digitizing its manufacturing base via IIoT technologies, Chinese vendors likely have more real-world data (by deploying more sensors), which enables firms to perfect their machine-learning algorithms, which will further improve manufacturing productivity in China relative to the United States. Robot adoption is a key example: when adjusted for labor costs, China uses 12 times more robots than the United States. This deployment of industrial robots at scale further advantages Chinese manufacturers and the entire technology stack associated with robotics (eg. operating systems for robots, robot supply chain, AI software to control the robots, software integrating robots into production processes, etc.). Recent reports of the Chinese government and enterprises mass-adopting DeepSeek only add urgency for more innovative industrial policies in the United States. Therefore, to achieve policy goals such as restoring US manufacturing or maintaining US leadership in quantum or AI, the United States must support companies to actually buy and use these technologies themselves.

While China may have “won” the initial round of the IoT platform wars, it isn’t too late for the United States, with smart policies and leadership, to win the broader industrial-technical leadership competition with China. While some may object to “picking winners and losers,” without urgent policy action, there may only be losers left to pick from.

Why is Trump appeasing Russia? What lessons can we learn from the battlefield in Ukraine? How will AI change warfare, and what does America need to do to adapt?

To discuss, we interviewed Shashank Joshi, defense editor at the Economist on a generational run with his Ukraine coverage, and Mike Horowitz, professor at Penn who served as Biden’s US Deputy Assistant Secretary of Defense for force development and emerging capabilities in the Pentagon.

We discuss….

• Trump’s pivot to Putin and Ukraine’s chances on the battlefield,

• The drone revolution, including how Ukraine has achieved an 80%+ hit rate with low-cost precision systems,

• How AI could transform warfare, and whether adversaries would preemptively strike if the US was on the verge of unlocking AGI,

• Why Western military bureaucracies are struggling to adapt to innovations in warfare, and what can be done to make the Pentagon dynamic again.

This episode was recorded on Feb. 26, two days before the White House press conference with Zelenskyy, Trump, and JD Vance. Listen now on iTunes, Spotify, YouTube, or your favorite podcast app.

Jordan Schneider: Shashank, it seems you had a lot of fun on Twitter this week?

Shashank Joshi: I was in a swimming pool with my children on holiday in the middle of England and didn’t notice until 18 hours after the fact that the Vice President of the United States had been rage-tweeting at me over my intemperate tweets on the subject of Ukraine. I provoked him into this in much the same way that he believes Ukraine provoked the invasion by Russia.

Jordan Schneider: What does it mean?

Shashank Joshi: It means the Vice President has far too much time on his hands, Jordan.

This is a pretty significant debate. Fundamentally, this was about whether Ukraine is fated to lose. His contention is that Russian advantages in men and weapons or firepower meant that Ukraine’s going to lose no matter what assistance the United States provides.

My argument was that while Ukraine is not doing well — I’m not going to sugarcoat that, I’ve written about this and it’s made me pretty unpopular among many Ukrainians — it’s not true that advantages in manpower and firepower always and everywhere result in decisive wins. Indeed, Russia’s advantage in firepower is much narrower than it was. The artillery advantage has closed. Ukraine’s use of strike drones — which we’ll talk about later — has done fantastic things for their position at the tactical level.

On the manpower side, Russia is still losing somewhere in the region of 1,200-1,300 men killed and wounded every single day. While it can replenish those losses, it can’t do that indefinitely. I’m not saying Vance is completely wrong — I’m just saying he is exaggerating the case that Ukraine has already lost and that nothing can change this.

My great worry is this is driving the Trump administration into a dangerous, lopsided, inadequate deal that is going to be disastrous for Ukraine and disastrous for Europe. I’m worried profoundly about that at this stage.

Michael Horowitz: Quantity generally sets the odds when we think about what the winners and losers are likely to be in a war. Russia has more and will probably always have more. But there are lots of examples in history of smaller armies, especially smaller armies that are better trained or have different concepts of operation or different planning, emerging victorious. Most famously in the 20th century, perhaps Israel’s victory in 1967.

Jordan Schneider: We have three years of data. It’s not like you’re playing this exercise in 2021. You’re doing this exercise in February of 2025. By the way, Mr. Vice President, your government actually has a ton of the cards here to change those odds and change the correlation of forces on the ground, which just makes the argument that this is a tautology so absurd coming from one of the people who is in a position to influence and who has already voted for bills that did influence this conflict.

Shashank Joshi: Wars are also non-linear. You can imagine a war of attrition in which pressures are building up on both sides, but it isn’t simply some mathematical calculation that the side with the greatest attrition fails. It depends on their political cohesion, their underlying economic strength, their defense industrial base, and their social compact.

The argument has been that although Russia feels it has the upper hand — it has been advancing in late 2024 at a pace that is higher than at almost any time since 2022 — there’s no denying that to keep that up, it would have to continue mobilizing men by paying them ever higher salaries and eventually moving to general mobilization in ways that would be politically extremely unpalatable for Vladimir Putin. War is not just a linear process. It’s a really complicated thing that waxes and wanes, and you have to think about it in terms of net assessment.

Michael Horowitz: That’s especially true in protracted wars. I’m teaching about World War I right now to undergraduates at Penn. One of the really striking things about World War I is if you look at the French experience, the German experience, and the Russian experience in particular, given the way that World War I is one of the triggers for the Russian Revolution, how their experience plays out in World War I is in some ways a function of political economy — not just what’s going on on the battlefield, but their economies and the relationship to domestic politics and how it then impacts their ability to stay in and fight.

Jordan Schneider: America has levers on both sides of the political economy of this war. There was a point a few weeks ago when Trump said he was going to tighten the screws on Putin and his economy. The fact that we are throwing up our hands and voting with Putin in the United Nations, saying that they were the aggressor, just retconning this entire past few years is really mind boggling. There was a line in a recent Russia Contingency podcast with Michael Kofman, where he says “The morale in Munich was actually lower than the morale I saw on the front in Ukraine,” which is a sort of absurd concept to grapple with.

Michael Horowitz: If you were to mount a defense here, what I suspect some Trump folks might say is that they believe this strategy will give them more leverage over Russia to cut a better deal. That involves saying things that are very distasteful to the Ukrainians, but they think as a negotiating strategy, that’s more likely to get to a better outcome.

Shashank Joshi: That’s right, Mike. Although they’ve amply shown they are willing to tighten the screws on Zelenskyy. If you were looking at this from the perspective of the Kremlin, would you believe General Keith Kellogg when he says, “If you don’t do a deal, we’re going to ram you with sanctions, batter you with economic weapons"? Or do you listen to Trump’s rhetoric on how we’re going to have a big, beautiful economic relationship with Russia and we’re going to rebuild economic ties, lift sanctions?

You’re going to be led into the belief that the Americans are really unwilling to walk away from the table because the Vice President and others are publicly saying we don’t have any cards, that the Ukrainians are losing, and if we don’t cut a deal now, then Russia has the upper hand. It puts them in a position of desperation.

My big concern is not just that we get a bad deal for Ukraine, it’s that the idea of spheres of influence appeals to Trump, dealing with great men one-on-one, people like Kim Jong Un, Vladimir Putin, Xi Jinping — and that what will be on the table is not just Ukraine, but Europe. Putin will say, “Look, Mr. President, you get your Nobel Peace Prize, we get a ceasefire, we do business together and lift sanctions. And you can make money in Moscow, by the way. Just one tiny little thing, that NATO thing. You don’t like it, I don’t like it. Just roll it back to where it was in 1997, west of Poland. That would be great. You’ll save a ton of money here. I’ve prepared a spreadsheet for you.”

That is the scenario that worries us — a Yalta as much as a Munich.

Jordan Schneider: We have a show coming out with Sergey Radchenko where we dove pretty deep into Churchill’s back-of-a-cocktail-napkin split. At least Churchill was ashamed.

It’s so wild thinking about the historical echoes here. I was trying to come up with comparisons, but the only ones I could do were hypotheticals. Like McClellan winning in 1864, or — I mean, Wendell Willkie was actually an interventionist. There was some Labor candidate that the Nazis were trying to support in the Democratic Party in 1940, but he never made it past first base. Has there ever been a leadership change that shifted a great power conflict this dramatically?

Shashank Joshi: From the Russian perspective, that’s Gorbachev. Putin would look back at glasnost, perestroika, and Gorbachev at the Reykjavik summit as moments where a reformist Soviet leader sold the house to the Americans and threw in the towel.

Michael Horowitz: You also see lots of wars end with leader change, with leadership transitions, when wars are going poorly for countries and you have leaders that are all in and have gambled for resurrection. If you think about the research of someone like Hein Goemans back in the day, then you have to have a leadership transition in some ways to end wars in some cases if leaders are sort of all in on fighting.

Jordan Schneider: The Gorbachev-Trump comparison is a really apt one because it really is like a true conceptual shift in the understanding of your country’s domestic organization as well as role in the world. Gorbachev, for all his faults, at least had this universalist vision of peace, trying to integrate in Europe — he wanted to join NATO at one point. But going from that to whatever this 19th century mercantilism vision is, is really wild to contemplate.

Shashank Joshi: The other thing to remember is Gorbachev’s reforms eventually undid the Soviet empire. They undid its alliances and shattered them. In the American case, the American alliance system is not like the Soviet empire. France and the UK are not the Warsaw Pact. We bring something considerably more to the table. It’s a voluntary alliance. It’s a technological, cultural alliance. These are different things.

Subscribe now

I worry sometimes that this administration or some people within it — certainly not everybody — views allies just as blood-sucking burdens. What they don’t fully grasp is how much America has to lose here. I want to say a word on this because Munich — and I heard this again — the FT reported recently that some Trump administration official is pushing to kick Canada out of the Five Eyes signals intelligence-sharing pact.

Now okay, the Americans provide the bulk of signals intelligence to allies. There’s no surprise about that. But if you lost the 25% provided by non-US allies, it will cost the US a hell of a lot more to get a lot less. It will lose coverage in places like Cyprus, in the South Pacific, all kinds of things in the high north, in the Arctic in the Canadian case. This administration just doesn’t understand that in the slightest.

Michael Horowitz: Traditionally what we’ve seen is regardless of what political hostility looks like, things like intelligence sharing in something like the Five Eyes context continues — in some ways the professionals continue doing their jobs. If you see a disruption in that context, that would obviously be a big deal.

Jordan Schneider: Just staying on the Warsaw Pact versus NATO in 2025 today, America plus its allies accounted for nearly 70% of global GDP during the Cold War. The economic outflows that were needed to sustain Soviet satellites eventually bankrupted the USSR. America isn’t facing anything resembling that situation by stationing 10,000 people in Poland and South Korea.

Michael Horowitz: We are in a competition of coalitions with China, and it is through the coalition that we believe we can sustain technological superiority, economic superiority, military power, et cetera. Look at something like semiconductors and the role that the Netherlands plays in those supply chains, that Japan plays in those supply chains. There are interconnections here. We have thought that we will win because we have the better coalition.

Shashank Joshi: That’s an interesting question to ask more conceptually — does this administration want a rebalancing of its alliances or does it want a decoupling? You could put it in terms of de-risking and decoupling if you want to echo the China debate here. Does it simply want more European burden-sharing? But fundamentally the US will still maintain a presence in Europe, underwrite European security, and provide strategic nuclear weapons as a backstop. That is what many governments are trying to tell themselves.

The more radical prospect is that whilst there are some people who envision that outcome — Marco Rubio, Mike Waltz (the National Security Advisor), and John Ratcliffe (the head of the CIA) — the President and many of the people around him view things in considerably more radical terms. It’s more of a Maoist cultural revolution than a kind of “I’m Eisenhower telling the Europeans to spend more.”

Jordan Schneider: There’s this quote from Marco Rubio that’s really stuck with me from a 2015 Evan Osnos profile where he talks about how he has not only read but is currently rereading The Last Lion, which is this truly epic three-part series. The middle book alone is most famous, which is what Rubio was referring to, where Churchill saw the Nazis coming when no one else did and did everything he could in the 30s to wake the world up and prepare the UK to fight.

Rubio is referring to this moment by comparing it to how he stood up to the Obama administration when they were trying to do the JCPOA nuclear deal with Iran. To go from that to having to sit on TV and blame Ukraine for starting the war, I think is just the level of cravenness. There are different orders and degrees of magnitude.

Shashank Joshi: You have to think about this not in terms of a normal administration in which people do the jobs assigned to them by their bureaucratic standing. You have to think about it like the Kremlin, where you have power verticals, or an Arab dictatorship where you have different people reporting up to the president. Think of this like in Russia, where you have Sergey Naryshkin, the head of the Foreign Intelligence Service, who may say one crazy batshit thing, but actually has no authority to say it. In which Nikolai Patrushev may say another thing, in which Sergey Lavrov may lay down red lines, but they have no real meaning because there’s a sense of detachment from the brain, the power center itself. Ultimately, it’ll still be Putin who makes the call. I think it’s a category error if we try to think about this administration as a normal system of American federal government.

Michael Horowitz: I will say, I can’t believe I’m now going to say this, but let me push back and say that there’s a lot of uncertainty about what the Trump administration wants to accomplish here, given the way they have embraced the notion that Trump is a master negotiator. To be professorial about it, in a Thomas Schelling “threat that leaves something to chance” way, or like madman theory kind of way, they think that there’s a lot of upside here from a bargaining perspective.

Most of Trump’s national security team is not yet in place. We just had a hearing for the Deputy Secretary of Defense yesterday. Elbridge Colby, who’s the nominee for undersecretary, has a hearing coming up, I think either next week or the following week. So a lot of the team is still getting in place.

Jordan Schneider: The thing about Trump 1.0 is there weren’t wars like this. You had two years of sort of normal people who were basically able to stop Trump from doing the craziest stuff. Then the COVID year was kind of a wash. But Trump 2.0 matters a lot more, it’s fair to say, over the coming four years than it did 2016-2020.

Shashank Joshi: It’s much more radical. In the first term, John Ratcliffe had his nomination pulled as DNI because he was viewed as inexperienced and not up to the job. Today, John Ratcliffe looks like Dean Acheson compared to the people being put into place. We have to pause and make sure that we recognize the radicalism of what is being put into place around us.

When you look at the sober-minded people who thought about foreign policy — and I include amongst this people I may disagree with, like Elbridge Colby, who will be probably the Pentagon’s next policy chief — what is the likely bureaucratic institutional political strength they will bring to bear when up against those with a far thinner history of thinking about foreign policy questions?

Jordan Schneider: I haven’t done a Trump-China policy show because I don’t think we have enough data points yet. But what, if anything, from the past few weeks of how he’s thinking and talking about Russia and Ukraine, is it reasonable to extrapolate when thinking about Asia?

Shashank Joshi: Two quick things. One is I see significant levels of concern among Asian allies. The dominant mood is not, “Oh, it’s fine, they’re going to just pull a bunch of stuff from Europe, stick it into Asia and it’ll be a great rebalancing.”

Number two, I think this is important: there is a strong current of opinion that views a potential rapprochement with Russia as being a wedge issue to drive between Russia and China, the so-called reverse Kissinger. Jordan, you know much more about China than I do. I’m not going to comment further on that, but I will say I believe it is an idea that is guiding and shaping and influencing current thinking on the scope of a US-Russia deal.

Michael Horowitz: You certainly have a cast of officials who are pretty hawkish on China, which will be a continuation in some ways of the last administration and the first Trump administration. I think the wild card will be the preferences of the president. There was a New York Times article a few days ago that talked about Trump’s desire for a grand bargain with China — his desire to do personal face-to-face diplomacy with Xi as a potential way to obtain a deal.

Now I think the reality is that every American president that has tried to do that kind of deal, whether in person or not over the last decade, has found that there are essentially irreconcilable differences. There’s a reason why there is US-China strategic competition and why that has been the dominant issue in some ways of the last several years and probably will be over the next generation. But Trump may wish to give it a shot — and it sounds like, at least from that article, that he might.

Jordan Schneider: We’ve also had every administration in the 21st century try to start their term by trying to reset relations with Russia. “Stable and predictable relationship” was Biden’s line. Maybe this stuff is just a blip, but I think Shashank’s right. We’re in really uncharted territory.

Paid subscribers get access to the rest of the conversation, where we discuss…

• AI as a general-purpose technology with both direct and indirect impacts on national power,

• Whether AGI will cause instant or continuous breakthroughs in military innovation,

• The military applications of AI already unfolding in Ukraine, including intelligence, object recognition, and decision support,

• AI’s potential to enable material science breakthroughs for new weapons systems,

• Evolution of drone capabilities in Ukraine and “precise mass” as a new era of warfare,

• How China’s dependence on TSMC impacts the likelihood of a Taiwan invasion,

• Whether AGI development increases the probability of a preemptive strike on the US,

• How defense writers and analysts help shape policy and build bureaucratic coalitions,

• Ukraine as a real-world laboratory for testing theories about warfare, and what that means for Taiwan’s defense.

Jordan Schneider: Let’s talk about the future of war. There is this fascinating tension that is playing out in the newly national security-curious community in Silicon Valley where corporate leaders like Dario Amodei and Alex Wang, both esteemed former ChinaTalk guests, talk about AGI as this Manhattan Project-type moment where war will never be the same after one nation achieves it. What’s your take on that, Mike?

Read more

Off all the wild moves we’ve gotten out of this Administration so far, basic science funding could be the dumbest and hardest to reverse.

does a great job with the basic plot.

Noahpinion

What happens when we gut federal science funding?

Back at the beginning of his first term in 2017, Trump briefly tried to cut federal science spending. I wrote a pair of posts for Bloomberg saying that this was a bad idea. Here are some excerpts from the first of those…

Read more

a month ago · 252 likes · 68 comments · Noah Smith

I’d like to spotlight the newest NSF directorate, Technology, Innovation and Partnerships (TIP) created by the CHIPS & Science Act, that has been particularly hard-hit by DOGE. The idea was to supplement the world-class basic research that NSF does with more use-inspired and translational research with higher technology readiness levels. I’ve been following this directorate since its creation, recorded a panicked emergency pod when for a hot minute Senate Commerce almost killed it, and have been really impressed with its work so far.

TIP helped stand up NAIRR, has done a fanstastic job helping catalyze regional innovative hubs, and is the only org I’ve seen in government actually be strategic about workforce development. My personal favorite its new APTO program, which is creating the data and intellectual substrate necessary to really do smart S&T and industrial policy. For more of what TIP has been up to, check out their Director’s annual letter here. I’d also encourage DOGE to have a read of the TIP’s roadmap for the next few years and try to spot stuff that America doesn’t need.

The NSF is not perfect. IFP has some excellent proposals on how to incorporate novel funding strategies like lotteries that need faster adoption. But IFP also recently wrote up how the NSF showed its mettle, and was able to move faster than the NIH for COVID-related grants. TIP in particular has collected some of NSF’s most forward-thinking talent and is experimenting with novel programs and funding strategies faster than anyone else in the NSF mothership.

American basic research is our golden goose and the envy of the world, building the basis for scientific innovations that make us richer, live longer, and make us more powerful. Our universities attract the best minds in the world which is an enormous boon to the country, and absent radical intervention will continue to do so. While the NSF could use reform, we are criminally underfunding R&D already, and firing the most forward-thinking junior staff in the directorate singled out by national security heavyweights as critical to competing with China is an error this administration should correct.

Try Picking on Someone Your Own Size

DOGE should really try taking on some government programs that aren’t already running lean, creating the future, preventing pandemics and saving lives. The real discretionary bloat isn’t malaria bednets and fundamental physics research but F-35s and carriers. A real push at a few deadweight DoD programs could deliver way more savings than whatever you can squeeze from NSF and USAID and likely make for a more effective force.

From Jennifer Pahlka:

The only way the DoD was really going to change was through major budget cuts — something that forced people’s hands into new ways of working, into true prioritization, into processes that took less time because they were less burdened by the trappings that come with enormous budgets. I began my comment with an apology to the senior Air Force official sitting next to me, a caveat that I meant no disrespect, and wasn’t arguing for less military might — in fact, what I wanted was a more capable military. To my surprise, he piled on. “She’s right,” he said. “But it has to be much deeper than anything we’ve seen before. We had to cut during the last sequestration, and it was around 15% off the top of everything, which doesn’t force meaningful choices. It needs to be like half.”

To get at wasteful DoD programs and acquisitions regulations this administration would have to do the hard work of wooing Congresspeople into taking votes that would more substantially impact their districts. I hope that Trump 2.0’s staff has the stomach and topcover for this sort of work that could yield real long-term dividends for the country, not just grabbing the lowest hanging political fruit which really even have long term fiscal relevance like cutting probationary employees, foreign aid, and basic R&D.

From a ChinaTalk episode coming out on Monday with Mike Horowitz, former Biden DoD official, and The Economist’s Shashank Joshi:

Jordan Schneider: And I think this is like one of the many shames of the Trump imperial presidency. He has enough control of Congress to do this well and could even get some Dem votes for real defense reform!

Mike Horowitz: Let me muster a point of optimism here. If you look at Hegseth's testimony, his discussion of defense innovation is very coherent. He has takes that are not structurally dissimilar to the ones that we have been making.

There is a potential opportunity here for the Trump administration to push harder and faster on precise mass capabilities, on AI integration, and on acquisition reform in the defense sector. Because the president right now seems to have a strong hand with regard to Congress. Whether the president's willing to use political capital for those purposes is not clear. How the politics of that will play out is unclear. But if the Trump administration does all the things that it says it wants to do from a defense innovation perspective, that may not be a bad thing!

Shashank Joshi: My concern is also that you have people who are good at radicalizing and disrupting many businesses and sectors and fields of life. But the skills that are required to do that are different to the skills in a bureaucracy like this. Because, just because you were able to navigate the car sector and the rocket sector, doesn't mean you know how to cajole, persuade, and massage the ego of a know-nothing congressman who knows nothing about this and who simply cares that you build the attributable mass in his state, however stupid an idea that is, and who wants you to sign off on the 20 million dollars.

I worry that they will either break everything, and I'm afraid what I'm seeing DOGE do right now with a level of recklessness and abandon is worrying to me as an ally of the United States from a country that is an ally, but also that they will just not have the political nous [British for common sense] to navigate these things to make it happen. Just because Trump controls Congress and has sway over Congress doesn't mean that the pork barrel politics of this at the granular level fundamentally change. You need operatives, congressional political operatives. A tech bro may have many virtues and skills, but that isn't necessarily one of them.

Here’s to hoping! Howabout a Washington quote to send us off, from a 1775 letter sent to General Schuyler: “Animated with the Goodness of our Cause, and the best Wishes of your Countrymen, I am sure you will not let Difficulties not insuperable damp your ardour. Perseverance and Spirit have done Wonders in all ages.”

The Death of AI Safety: Moving Past the Pantomime

Tim Hwang is a writer and researcher. Relevant to the NSF topic above, he also hosted a great podcast series on metascience you should check out!

The AI Action Summit, which closed just over two weeks ago in Paris, will be remembered as a historically important gathering — though not how many of its organizers, attendees, and contributors anticipated. Rather than cementing AI safety as a priority for transnational collaboration, it turned into a memorial service for the safety era.

Billed as the successor to the high-profile gatherings of leaders that took place in the United Kingdom in 2023 and Korea in 2024, the Summit was originally intended to build on the frenetic activity that has taken place over the last few years to create international machinery for collaboration on AI safety issues. This has included an agreement on statements of principles, the formation of AI Safety Institutes around the world, and a blue ribbon “IPCC-style” report on safety issues.

This Summit’s lasting moments, however, came not from the success of “open, multi-stakeholder and inclusive approach[es]” on safety championed by the official declaration from the event, but instead dramatic declarations of national primacy unshackled by safety concerns. Vice President JD Vance’s speech made little accommodation for either safety or internationalism, declaring that the United States was “the leader in AI and our administration plans to keep it that way,” and that he was not here to talk about AI safety but instead “AI opportunity.” Macron touted a massive €110 billion fund to back AI projects in France, and the United States and United Kingdom declined to sign the Summit’s declaration language. A wildcat “Paris Declaration on Artificial Intelligence” backed by private industry hit the Summit for failing to back a “strong, clear-eyed, and Western-led international order for AI.”

A sense of stuckness prevailed in the side conversations and events taking place throughout Paris. At the AI Security Forum, a slow carousel of speakers ran through very much the same tropes and ideas that had dominated the discourse for years. Shakeel Hashim captured a feeling widely held — that the Summit was a “pantomime of progress” rather than the genuine article.

The photo ops were taken, and the keynotes were done, but the old gestures and governance rituals — which had seemed so potent just a few years ago in Bletchley Park — are now odd anachronisms in the harsh light of 2025.

This isn’t just a vibe. The “AI safety community” has always nurtured a shared, but often unspoken, agreement that public-minded technical expertise and international cooperation were the most promising pathways to promote good global governance of the technology.

But the safety community made a historically bad bet. The wheels were already coming off multistakeholder, international governance in the world at large even as the safety community began to invest in it seriously in the mid-2010s. Resurgent nationalism, great-power competition, and the fecklessness of international institutions have limited options for global governance across many domains, and AI has been just another one of the casualties. This isn’t just about Trump winning: these changes in the international system are structural, and the domestic shifts in places like France and the UK would have led to a very similar result even if Harris had pulled it out last year.

The safety community was also profligate in the use of its attention and social capital. The political influence of fair-minded technical experts turned out to be a rapidly depleting resource, wasted away as one “high-profile letter from very concerned scientists” and “dramatic demo of hypothetical model threat” followed another to little effect.

Against such a backdrop, it’s no wonder that AI safety in 2025 feels ever more like pantomime. We’re still frantically pulling the same levers, even as the whole constellation of forces that move nations in general and technology policy in particular have rearranged.

We need to be asking hard questions. What are historical models for technological safety and stability in a world of fierce, unrestricted nationalism? What happens when scientific evaluation has lost its ability to persuade the policymaker? How do you slow down or stop a technological race-in-progress?

The real intellectual work is now rebuilding a theory for safety that takes these uncomfortable realities into account and builds as best it can around them.

Mood Music

Can economic warfare really work? What can we learn from the 21st century historical record of American sanctions policy?

To find out, we interviewed Eddie Fishman, a former civil servant at the Department of State and an Adjunct Professor at Columbia. His new book, Chokepoints: American Power in the Age of Economic Warfare, is a gripping history of the past 20 years of American sanctions policy.

In this show, we’ll talk about…

• The evolution of U.S. sanctions policy, from Iraq and Cuba to Iran and Russia,

• How Reagan’s deal with the Saudis turned the dollar into an economic chokepoint,

• The incredible success of sanctions against Iran, and how that playbook could have been used to punish Russia,

• Historical lessons in enforcement that are relevant for export controls on China today,

• The role of great civil servants like Stuart Levey, Daleep Singh, Victoria Nuland, and Matt Pottinger in building state power,

• Institutional challenges for economic warfare and the consequences of failure to reform,

• Strategies for writing groundbreaking books about modern history.

Watch below or listen now on iTunes, Spotify, YouTube, or your favorite podcast app.

Financial Chokepoints

Jordan Schneider: Let’s start with the Bosphorus. How does this little corner of our beautiful planet explain the evolution of sanctions?

Edward Fishman: The Bosphorus is the epitome of a maritime chokepoint. It is a narrow strait between the Black Sea and the Mediterranean Sea. Throughout history, maritime chokepoints like the Bosphorus have been critical for strategic power. Sparta was able to win the Peloponnesian War because they won a battle around the Bosphorus and blockaded it, ultimately starving the Athenians into submission. Athens had relied on the flow of grain through the Bosphorus to feed its population — that was really the whole purpose of ancient Athens’ maritime empire.

Historically, these chokepoints have been geographic features. But now, as a result of globalization, there are chokepoints in the global economy that are not geographic — the most critical of which is the U.S. dollar. This is why the book is called Chokepoints.

For thousands of years throughout history, the only way to block a maritime chokepoint like the Bosphorus was a physical naval blockade. What’s changed is that in the wake of hyperglobalization in the 1990s, the U.S. acquired the ability to block chokepoints like the Bosphorus just by weaponizing its control of the U.S. dollar.

Today, the director of OFAC, the unit at the Treasury Department that oversees sanctions policy, can sign a few documents in her office and blockade a chokepoint like the Bosphorus. This actually happened on December 5, 2022, when the G7 oil price cap went into effect. The Bosphorus was backed up with dozens of oil tankers, because Turkish maritime officials were so nervous about violating the terms of the price cap that they didn’t want the ships to cross. It took OFAC days of very intensive diplomacy with Turkish authorities to persuade them to allow the ships to cross.

Jordan Schneider: You open this book with some wild contrast. Historically, you needed triremes. Now, all you need is a piece of paper from the Treasury Department to clog up the strait in Turkey halfway around the world.

Like you, Eddie, I was a sanctions nerd in college. I wrote my thesis about the origins of the UN and did papers on sanctions policy. I remember very vividly reading this literature arguing that sanctions are useless and don’t have any big impact. There was this great quote from George W. Bush in your book where at some point in the 2000s, he said, “We’ve sanctioned ourselves out of any influence” when it came to Iran’s nuclear program. You put the spotlight on one civil servant who takes that as a challenge and through ingenuity, creativity, and a whole lot of elbow grease, is able to discover and leverage a whole new lens of American power. Let’s briefly tell the story of American sanctions pre-Stuart Levey before we discuss Iran’s nuclear program.

Edward Fishman: When Stuart Levey came in as the Treasury Department’s first undersecretary of terrorism and financial intelligence in 2004, the most recent big case of sanctions that the U.S. had was a 13-year sanctions campaign against Iraq from 1990, when Saddam originally invaded Kuwait, until 2003, when George W. Bush launches the invasion of Iraq. That embargo required full UN backing and was implemented by a 13-year naval blockade. You had literally a multinational naval force parked outside of Iraqi ports inspecting every single oil shipment going in and out of Iraq.

The lesson from this situation was that sanctions didn’t work — Saddam didn’t come to heel. He seemed to be just as aggressive, if not more so. Over time, this embargo wound up leading not only to humanitarian problems in Iraq, which are very well documented, but also significant corruption. Saddam was siphoning away oil money under the nose of the UN.

By the time Levey comes in, sanctions had been seen as something that had been tried and failed against Iraq, and in fact had paved the way for the U.S. invasion of Iraq. In many ways, the 2003 invasion of Iraq was a direct result of the perception that sanctions had failed.

When Levey started working on the Iran problem around 2004, the prospect of even doing an Iraq-style sanctions campaign against Iran was off the table because there was no way to get the UN Security Council to agree to that at the time. Bush’s comment about having sanctioned ourselves out of influence with Iran was a result of the fact that without the UN, the U.S. thought that the only type of sanctions we could impose were primary sanctions, like an embargo where U.S. companies can’t buy things from Iran or trade with Iran. The only issue is we had had an embargo in place since the mid-90s, so there wasn’t any trade to speak of between the U.S. and Iran. The two avenues of sanctions were closed off — sanctions through the UN had been discredited by the 90s, and the other, primary sanctions on Iran, had already been maxed out and had been for a decade by then.

Jordan Schneider: The other seminal piece of sanctions in American 20th-century history is the embargo on Cuba. That is the same story — we cut off trade with this country, yet Castro’s still there in 2004, some 50-odd years later. It’s interesting — if you go back even further, there was this real hope after World War II where the UN at one point was even going to have its own air force. The idea was that sanctions were going to be this incredible tool to deter bad actions by different actors around the world because the U.S. and the Soviet Union were friends and we would all police the planet in a happy-go-lucky way. That was not how the Cold War ended up working out.

In 2004, Stuart Levey started to understand that he can leverage the dollar’s role in global financial flows. Eddie, can you tell the story of how the U.S. dollar became globalized in this way?

Edward Fishman: Bretton Woods, the conference that set the rules of the road for the post-World War II economy, happened in 1944. It put the U.S. dollar at the center of the global economy and established the dollar as the global reserve currency. It made the dollar as good as gold — the dollar is convertible for a fixed rate of $35 per ounce of gold.

At the same time, it explicitly prioritized the real economy and trade over finance. John Maynard Keynes, who was one of the architects of the Bretton Woods system, said that capital controls were a very important part of the system. For the first 30 years of this new global economy that emerged after World War II, you had the dollar at the center of the world economy, but it wasn’t a particularly financialized world economy. Most states had pretty significant capital controls, and banking was a very nationalized and, in some ways, even just a regionalized type of business.

By 1971, the U.S. dollar had been losing its value for quite some time and we were running significant deficits because of the war in Vietnam. Ironically, this is when Richard Nixon unilaterally took the dollar off of the gold peg. The dollar was still at the center of the world economy, but it was no longer tethered to gold. Exchange rates were now set by the market instead of by government fiat.

In the years after that, the capital controls of the Bretton Woods system fully erode and the dollar winds up becoming even more integral to the world economy as we see financialization take off from the ’70s through the Clinton era. You get to the point where we have a foreign exchange market that is turning over seven or eight trillion dollars every single day, which is by far the largest of all financial markets.

Jordan Schneider: How did oil come to be traded in U.S. dollars?

Edward Fishman: The dollar’s role in trading oil is arguably the most important chokepoint for a number of the key sanctions campaigns of the 21st century.

After World War II, the U.S. was a large oil producer and a big exporter. The 1973 Arab oil embargo shifted our perspective, and the U.S. realized just how vulnerable it was to being cut off from Middle Eastern oil.

In 1974, Richard Nixon — who was wallowing under the political pressure of the Watergate scandal and massive deficits that we had no reasonable way of plugging — sent his treasury secretary, Bill Simon, to make a deal. Simon was a former bond trader, a New Jerseyite, a chain smoker...

Jordan Schneider: A chain-smoking New Jersey native, described by a peer as, “far to the right of Genghis Khan.”

Edward Fishman: He’s a really colorful figure. The book includes a photo of him testifying before Congress with a giant plume of smoke around him.

Bill Simon tried to think about how to plug these deficits using his financial background as a bond trader. He proposed cutting a deal with the Saudis such that, not only do they agree to keep pricing oil in dollars into perpetuity, but they actually take the dollars they earn from selling oil and reinvest them in U.S. government debt — they basically plug our deficit with the money that the U.S. is paying them for oil. He wound up taking a flight to Jeddah in the summer of 1974 — getting copiously drunk en route.

The deal worked. He cut a deal with the Saudis in which they agree to recycle their petrodollars into U.S. Treasuries. This agreement largely still exists to this day. Oil, by and large, is priced in dollars no matter who’s buying it or selling it.

Chokepoints in the global economy are typically formed by the private sector. They kind of develop naturally as businesses evolve. However, there are important moments when government intervention becomes critical.

Simon’s original deal in 1974 solidified the petrodollar, but then a few years later, as the dollar continued to slide in value, oil exporters and OPEC started getting upset because the weakening dollar was in turn reducing the real value of their oil earnings. Jimmy Carter’s Treasury Secretary, Michael Blumenthal, actually went back to Saudi Arabia and cut a new deal in which he agreed to give Saudi Arabia more voting shares at the IMF in exchange for Saudi continuing to price oil in dollars.

Jordan Schneider: Why did the Saudis even cut the deal in the first place?

Edward Fishman: The Saudis got two things. First, they got access to US military equipment, which was pretty beneficial to them. Second, which I think is more of a direct part of this deal and one that’s more easily provable through historical documents, the Saudis were able to buy U.S. government debt in secret outside of the normal auctions. Instead of participating in the public auctions for U.S. Treasuries, they had their own side deal where they could buy Treasuries. That was a big benefit to them because they were able to lock in prices and also do so without facing potential political opprobrium.

Jordan Schneider: That’s crazy.

Edward Fishman: It’s a remarkable turning point in the financial and economic history of the 20th century. There was a real shot that oil could have been priced against a basket of currencies, which in some ways makes more sense. For these countries in the Middle East and OPEC members, their entire economy basically depends on generating oil revenue. If you want stability and predictability, you don’t want to take exchange rate risk. But people like Bill Simon and Michael Blumenthal intervened and were able to get the dollar enshrined as the key part of the oil market.

The Iran Sanctions Formula and JCPOA Diplomacy

Jordan Schneider: Let’s talk about 2006, when Stuart Levey was trying to figure out how to make sanctions work against Iran. Can you explain his light bulb moment during the January 2006 trip to Bahrain?

Edward Fishman: Levey realized other countries hadn’t stopped doing business with Iran — only the U.S. had, and that’s why the sanctions weren’t working. But he realized that he could use access to the dollar as a lever to pressure foreign banks.

Typically, when you’re trying to get other countries on board for sanctions, you would go negotiate with their foreign ministry and say, “We think what Iran’s doing is bad. You should impose your own sanctions on Iran.” That was the paradigm before 2006. What Levey realizes is that he can go directly to the CEOs of foreign banks, bringing declassified intelligence demonstrating how Iran uses their banks to finance their nuclear program, and funnel money to terrorist proxies like Hamas and Hezbollah. To start, he could just present the facts and potential reputational concerns would often persuade these banks to exit Iran. In more extreme circumstances, when banks wouldn’t go along with him, he could threaten their access to the dollar to try to get them out of Iran.

What Levey really pioneered was the direct diplomacy between him as a Treasury official and his team at the Treasury Department with bank CEOs. You might ask, how did Stuart Levey get meetings with CEOs of banks all around the world? He was lucky — right when he had this epiphany, Hank Paulson, who had been the CEO of Goldman Sachs, came in as Treasury Secretary. Paulson is arguably the most well-connected banker in the world at the time. Hank winds up opening a lot of doors for Stuart and getting him meetings with ultimately more than 100 of the key banking CEOs around the world.

Jordan Schneider: Interestingly, you have to convince all the banks to get on board, because even the slightest institutional leakage would allow Iran to sell as much oil as they want.

How did Levey and his team go about convincing the Russians, the random Chinese banks, the Azerbaijani banks, and all of these other banks?

Edward Fishman: What Levey succeeds at doing between 2006 and 2010 is getting the big name-brand global banks to exit Iran. By and large, there are a few stragglers like BNP Paribas. Most of the big main global banks are out of Iran by 2010, though there are still some banks in places like the UAE, Turkey, and other countries doing business with Iran.

What winds up happening at that time is Congress, which has very little faith in Barack Obama’s willingness to come down hard on Iran — namely because Obama had very explicitly run for president in 2008 saying he wanted diplomacy. He even exchanged letters with Ayatollah Khamenei.

Even Iran hawks that are on the Democratic side of the aisle, like Bob Menendez, don’t really have much confidence that Obama is going to be tough on Iran. Democrats and Republicans basically form almost a coalition against the Obama administration on Iran sanctions and wind up passing progressively harsher sanctions legislation.

The key part of these sanctions laws, the first one called CISADA (the Comprehensive Iran Sanctions Accountability and Divestment Act of 2010), is that they require the Obama administration to impose what’s called secondary sanctions. That’s not sanctions directly on Iran, but sanctions on Iran’s business partners — for instance, the UAE or Turkish bank that I mentioned before.

Levey was a Bush appointee retained by the Obama administration (he’s one of only two very senior officials, along with Bob Gates, who’s kept on). He uses this law with the mandatory secondary sanctions as a significant cudgel. He goes to places like Dubai and talks to banks saying, “Look, if you don’t get out of Iran, I will be forced by American law to impose sanctions on you. You will lose access to the dollar and all of your assets will be frozen.” That threat is very significant. When the choice is between Iran and the United States dollar, it’s a pretty easy choice for most banks around the world.

Secondary sanctions had been tried before in the mid-90s, but the U.S. effectively wound up blinking and not imposing secondary sanctions on Total, the French oil company that had been investing in Iran’s oil sector. Even the George W. Bush administration decided not to impose secondary sanctions. This tool was very controversial. You can imagine it didn’t go down well with other countries. If you’re an American diplomat and you go meet with one of your counterparts abroad and say, “Sorry, we have to sanction your biggest bank if they don’t stop doing business with Iran” — that just feels like mafia diplomacy, not something that goes down very easily.

One of the virtues of Obama being so beloved around the world was the success of sanctions on Iran. Obama built international consensus that Iran’s nuclear program was a problem.

Jordan Schneider: We also had multilateral sanctions from the UN alongside U.S. action. What did that end up doing for the Obama psyche and the global push to limit Iran’s oil revenue?

Edward Fishman: Obama successfully got a major UN Security Council resolution done in the summer of 2010, right alongside when CISADA, the secondary sanctions law, passed Congress.

Jordan Schneider: In the Medvedev era, mind you.

Edward Fishman: Yes, exactly. Historical contingency matters — the fact that Medvedev was president of Russia at the time meant that Russia didn’t veto UN Security Council Resolution 1929. In retrospect, the benefit of that resolution wasn’t so much the specific sanctions it imposed on Iran. Rather, it explicitly drew connections between Iran’s banking system and energy sector with its nuclear program. This meant when Obama officials traveled the world to tell foreign banks and their governments that they’d be forced to impose sanctions if they didn’t stop doing business with Iran, they could credibly say they were just complying with UN Security Council Resolution 1929 and that international law was on the side of the United States. The legitimacy that Obama’s sanctions campaign derived from the UN was ultimately very significant.

Jordan Schneider: Iran was completely unprepared for this. They literally took out ads in newspapers in Austria to beg for help financing their nuclear program.

Edward Fishman: Exactly. This speaks to assumptions about how the global economy worked at the time. People just trusted that banking networks wouldn’t be weaponized. Iran really thought that they could publicly advertise these fundraising activities with no issue. Foreign banks weren’t aware of what Iran was doing and weren’t particularly worried about being penalized for it. They probably viewed sanctions as something that were unlikely to happen to them — and if they did happen, they could just be chalked up as a cost of doing business.

Jordan Schneider: Let’s talk about the penalties. One of the remarkable accomplishments of the Treasury Department, which the export controls regime on China over the past few years hasn’t been able to do, was the billion-dollar fines thrown on violators — $2 billion on HSBC, and almost $10 billion on BNP Paribas. How did this work?

Edward Fishman: This is a very important part of the story and one that often goes unnoticed. It’s not that sanctions didn’t exist before this period in the early part of the 21st century — it’s that the cost of violating them wasn’t particularly high.

One of the most important strategic legacies of the campaign against Iran pioneered by Stuart Levey is conscripting banks to be frontline infantry of American economic wars. This wasn’t because banks decided that this was morally righteous, it was because they realized that violating sanctions was existentially dangerous for their businesses.

Between 2010 and 2014, Standard Chartered wound up getting fined about a billion dollars, HSBC was fined $2 billion, and BNP Paribas was fined $9 billion. In each case, the New York Department of Financial Services actually threatened to withdraw banking licenses from each of those banks, which would eliminate their ability to do business in the United States. That was a sword of Damocles hanging over these banks — U.S. law enforcement probably could have extracted even bigger fines.

We’re still living with that legacy today. The reason that financial sanctions in particular are so powerful is a confluence of two factors.

1. The dollar is essential to international commerce. Trying to do business across borders without access to the dollar is like trying to travel without a passport.

2. The U.S. actually can weaponize the systemic significance of the dollar because banks are afraid of going against American government dictates.

Jordan Schneider: The political economy of it is also different than whacking Nvidia or Synopsys, becauce those three banks are foreign. It is one thing to threaten with extinction some hoity-toity French bank that sponsors the French Open and has been doing business with Iran forever. It’s another to threaten a major contributor to America’s national competitiveness, employment, and growth.

Compare the death sentence of being cut off from the New York Federal Reserve versus mere fines in the case of export controls. With Huawei, there were some cases where they threatened to put executives in jail. Over the past few years, the types of companies that the Biden administration has gone after have often been random Russians in Brooklyn smuggling chips into Russia and China. Whereas the Obama administration was trying to put teeth behind big economic warfare efforts by throwing down billion-dollar fines.

Edward Fishman: Is it possible to conscript tech companies in the same way that banks are conscripted? My own view is yes. If the fines were harsh enough and if the enforcement were strong enough — because the other fact we haven’t talked about is it wasn’t just fines for these banks, it was also independent monitors. The Justice Department sent in people to oversee compliance reforms for several years thereafter.

It is possible, though politically challenging, on one hand to be subsidizing American semiconductor companies to the tune of 50-plus billion dollars, and then on the other to say we’re going to take that money back because you’re violating export controls. It is possible.

One thing I would mention though is that with the BNP fine and the HSBC fine, those took many years to come to fruition. These were years and years of bad behavior that then eventually led to giant fines. It is possible that someone right now at the Justice Department is working away at a major export control violation case that we’ll learn about maybe in a couple of years.

Jordan Schneider: You mentioned “Mafia diplomacy” as a sort of derogatory term for sanctions tactics. There are a lot of moments in this story where gentlemanliness appears to be very important to Obama.

After the invasion of Crimea, around the Maidan revolution, Obama had a call with Putin where he warned that “Moscow’s actions would negatively impact Russia’s standing in the international community.” Putin’s response was basically like, “I don’t know, man, it’s hard to take you seriously.”

Why was Obama’s demeanor so helpful in the case of Iran?

Edward Fishman: Obama was very attuned to international law, or as you put it, gentlemanliness. You could argue he was very lawyerly in his approach. With respect to the Iran sanctions, I think it actually wound up being helpful because the secondary sanctions against Iran were beyond anyone’s imagination.

We haven’t talked yet about the oil sanctions, which were put in place in 2012. The U.S. successfully reduced Iran’s oil exports from 2½ million barrels a day to 1 million barrels a day over about a year. This is explicitly a unilateral U.S. sanction.

Would that have worked as well had Obama not been as attuned to diplomacy and invocations of international law? I’m not so sure. You may have seen more challenges from places like China and India and maybe more obstinance. I do think it was helpful in some regards.

Subscribe now

Looking at all the various examples of economic warfare that I talk about in the book, this is in some ways the most remarkable because of how unlikely it is to succeed. But it works.

One big exception from the financial sanctions during the Stuart Levey era is the Central Bank of Iran. The Central Bank of Iran is not under sanctions because it’s the repository for all of Iran’s oil revenues. The Obama administration was really nervous that if they sanction the Central Bank of Iran, other countries won’t be able to pay Iran for its oil. All of a sudden you’ll have all of Iran’s oil go off the market overnight, you’ll have a giant spike in oil prices, and everyone will be in a world of hurt.

Senator Bob Menendez, who was the key Iran hawk in the Democratic Party...

Jordan Schneider: For international listeners, Menendez is now in jail for having taken gold bars from Egypt. But anyways, continue, Eddie.

Edward Fishman: It’s a wrinkle in the story. Then Mark Kirk, who’s his Republican counterpart, who also wants to do a naval quarantine of Iran — the two of them basically say, “We don’t care, Obama, we’re going to sanction Iran’s central bank.” That amendment passes 100 to 0 in the Senate.

Obama is left with figuring out how to make this work. They come to a compromise with the Hill in which they agree to sanction the Central Bank of Iran, but they create two exceptions. One is an exception for countries who every six months significantly reduce their purchases of Iranian oil. For instance, if you’re a Chinese bank, you’re exempt from this — you can pay the Central Bank of Iran so long as China as a whole every six months reduces its overall purchases of oil from Iran. This gives a glide path for Iranian oil sales to decline over time and winds up working marvelously, luckily with the ramping up of shale production in the U.S.

The other exception put in place in 2012 says you can pay the Central Bank of Iran if you’re a Chinese refinery or bank, but those payments have to go into an escrow account that stays inside China and can only be used for bilateral trade between China and Iran.

This actually gives Chinese entities an incentive to comply, because keeping this money in China is going to boost Chinese exports to Iran — there’s nowhere else that the Iranians can use the money.

The one-two punch of these gradual oil reduction sanctions and the escrow accounts leads to a situation where Iran’s oil sales collapse by 60% by volume and it effectively has zero access to its petrodollars. Within 18 months, about $100 billion of Iran’s oil money gets trapped in these overseas escrow accounts. This is the context in which Iran’s economy really goes into free fall. Hassan Rouhani, a dark horse presidential candidate in 2013, won the Iranian presidency on an explicit platform of trying to get the sanctions lifted.

The remarkable thing about this oil sanctions regime is it’s probably the most effective oil embargo we’ve seen in modern history. It’s done unilaterally by the U.S. — no other countries are fully bought into this. It doesn’t involve any sort of naval strategy at all. There’s no quarantining of oil ships or anything. It is just using these threats of being cut off from the dollar to coax banks in places like China and India to comply with American dictates.

Jordan Schneider: This is going to be the poster child for decades of history books in that it actually created political change. It both drove home economically, causing hyperinflation and really hitting growth, and then got you a new slate of politicians who some would argue really wanted to make a deal. Looking back 15 years later, what’s your take on JCPOA and how we should think about the lessons from how the Obama administration used the leverage that they created with this oil embargo?

Edward Fishman: The JCPOA is the high point of American economic warfare in the 21st century in that you actually see sanctions leading to the outcome that the United States had set out, which was to get a peaceful resolution to Iran’s nuclear program. You can quibble about whether the terms of the JCPOA were stringent enough. However, there’s pretty good consensus that sanctions were the critical unlock to that deal.

Democrats say that sanctions were the key to getting the deal. Republicans say that sanctions were working so well that if we had only kept them in place longer, we would have gotten an even better deal. Within really a 10-year period, we flip that consensus from sanctions don’t work to sanctions are this magic bullet that just ended Iran’s nuclear program without firing a shot.

The key lesson here is that you need both economic leverage to make sanctions work and a clear political strategy. Having a clear political strategy, which was to get a nuclear deal with Iran, wound up being very important because you wind up having the international community grudgingly go along with the sanctions. They don’t voluntarily go along — they kind of have to be dragged along, including even the Europeans. But it would have been much harder to bring them along if there hadn’t been a political strategy, if it had just been bludgeoning Iran with economic pain without any sort of political end game in mind.

Responding to Russia (2014 vs. 2022)

Jordan Schneider: Let’s transition from the success of Iran sanctions to the failed response to the annexation of Crimea. What was different about how Obama and the world responded to Russia’s invasion in 2014?

Edward Fishman: Too often we tell our histories in silos — U.S. policy toward Iran vs. U.S. policy toward Russia. One thing I wanted to show in my book is that all of these sanctions campaigns are intertwined because ultimately these are the same decision makers at the table in the Situation Room across multiple issues.

The timeline is interesting here — the U.S. signed the original Iran nuclear deal, which froze Iran’s nuclear program, on November 24th, 2013. On the same exact day, hundreds of thousands of protesters descended upon the Maidan in Ukraine to protest Viktor Yanukovych’s deal with Putin.

The Ukraine crisis really does wind up taking the Obama administration by surprise. It’s not like the Iran nuclear program, which played out over the years as a slow-burning crisis. The Ukraine crisis and the Crimea annexation happened very quickly, with the U.S. constantly playing catch up. This parallel is important because right when Obama officials are scrambling to figure out what to do about Putin’s annexation of Crimea, they’re fresh off this giant victory where they just froze Iran’s nuclear program basically just by using sanctions.

It became natural for Obama officials in February-March of 2014 to say maybe sanctions could work against Russia. It’s a harder problem with Russia for several reasons. Russia has a much larger economy than Iran — in 2014 it was the 8th largest economy in the world and the world’s largest exporter of fossil fuels. Europe is completely dependent on Russian energy to heat their homes. Natural gas pipelines crisscross the continent between Russia and Europe.

Putin is creating facts on the ground as the U.S. is trying to scramble to put together sanctions. The annexation of Crimea happens within weeks of the “little green men” showing up in Crimea — they appear at the end of February and the annexation is formalized in middle of March. Shortly thereafter, Putin starts sending little green men into the Donbas, Ukraine’s industrial heartland.

Jordan Schneider: Let’s focus on the multilateral dynamic of this because obviously the UN is thrown out when Russia’s doing the thing. I remember very vividly watching the transition of the European actors who were pretty close to shrugging off this whole thing — until all those Dutch people died in the commercial liner that the Russians shot down by accident with their anti-aircraft missile. Can you explain how that changed the dynamic?

Edward Fishman: When Putin annexed Crimea in March of 2014, the U.S. and Europe did go ahead with some sanctions, but by and large they’re individual sanctions on people very close to Putin — his judo partners from childhood who have been elevated to positions of power at companies like Rosneft. Igor Sechin, for instance, the CEO of Rosneft, is sanctioned, but there are no sectoral sanctions, no actual significant economic sanctions on the Russian oil industry or its banking sector.

Obama and European leaders very publicly threatened this in March of 2014, but they don’t do anything. The reason is partly because there isn’t political will, but it’s also because they don’t know what kind of sanctions are tolerable to their own economies. They wind up spending months negotiating and coming up with what they eventually term “scalpel-like sanctions,” which effectively cut off big Russian state-owned enterprises from Western capital markets. It’s using an even narrower chokepoint than the dollar — it’s really just Western financing.

Interestingly, something that doesn’t often get recognized enough, the Obama administration went ahead with these sectoral sanctions, cutting off some big Russian energy companies and banks from U.S. capital markets on July 16, 2014, the day before MH17 was shot down. Obama and his team were getting fed up with the European foot-dragging. They say we need to send a powerful signal to Putin if we’re going to have any chance of deterring a broader invasion of the Donbas.

At the time, the New York Times was publishing headlines like, “Obama goes ahead without the Europeans.” Banking CEOs in the U.S. are incredibly upset because they’re saying this is just going to lead to a flight from the dollar to the euro and all our competitors in Frankfurt and London are going to benefit at our expense.

The next day, Putin’s proxies in the Donbas shot down a commercial airliner using a Russian-made Buk missile. They killed almost 300 people, by and large Europeans, most of them Dutch. All of a sudden the political aperture just widens completely in Europe. The Europeans are suddenly not only ready to match the U.S. sectoral sanctions of July 16, but actually go beyond them — they wind up cutting off all of Russia’s state-owned banks from the European financial system. The real core sectoral Russia sanctions are put in place after MH17, really from late July 2014 through September 2014 when Russian and Ukrainian leaders agree to the first Minsk agreement, the first ceasefire in the conflict.

Jordan Schneider: There are two parts that made me get upset rereading and reliving this story. One is that the Obama administration had just learned the lesson which Democrats in general have a really hard time with — escalate to de-escalate. It’s such an Obama thing, the same with the debt ceiling, where he was just like, “I’m going to be a nice normal actor and lay out my five demands and okay, we’ll get to two or three.” The Tea Party — this is ancient history now — and the Republicans were like, “No, we want 100% of what we want.” Obama would get scared, then they’d do a debt ceiling fight and he would end up giving way more than he realized he had to.

By the time we got to 2014, he just said “screw you.” He had the playbook with Iran. All the Treasury forecasting about the catastrophic costs of sanctions is overblown. The U.S. had more agency than expected, the euro was not going to take over.

But Russia really got away without serious economic consequences. Why didn’t Obama put the money where his mouth was?

Edward Fishman: In retrospect, there are two things that led to Obama’s overly cautious approach. One was real, genuine concern about the U.S. economy and the European economy. Remember, we’re still in the wake of the financial crisis and the Eurozone crisis is very much a live situation. There are genuine fears from the Treasury Department that you could accelerate a financial crisis in Europe if Russia were to cut off their gas supplies, and that contagion would spread to the US.

The other thing — this is an interesting paradoxical lesson for the Trump people now and people who say Europe needs to pull more of its own weight — Obama was very deferential to the Europeans over the Ukraine crisis. He explicitly wants people like Angela Merkel and François Hollande to take the lead. The negotiating block that came up with the Minsk agreement, the Normandy format, is France, Germany, Russia, and Ukraine. The U.S. doesn’t even have a seat at the table in the negotiations. Obama was saying, “This is in Europe’s backyard. It’s really their problem.”

In retrospect, that caution does not look very wise. Obama should have hit Russia much harder than he did in 2014. One interesting thing though is even though the sanctions put in place that summer — these capital market restrictions, the “scalpel-like sanctions” — are much weaker than the Iran sanctions, in the second half of 2014, oil prices cratered from over $100 a barrel to around $50 a barrel.

While the sanctions were aimed at trying to constrain Russia’s economic horizons as opposed to creating an immediate financial crisis, the sanctions do push Russia to the brink of a complete meltdown. In the winter of 2014-2015, Russia’s economy looks like it’s about to collapse — honestly just as bad, if not worse than Russia’s economy winds up looking after the much more drastic sanctions from February-March 2022.

Subscribe now

The reaction is remarkable. I have some of these quotes in the book. European leaders look at this and say, “This isn’t scalpel-like — this is what we signed up to. We didn’t want to push Russia off a cliff.” Hollande, the French president, actually says, “We explicitly don’t want to push Russia to its knees.” The Europeans, and to a certain extent the United States, got spooked by how impactful the sanctions are because they wind up being accelerated by this collapse of oil prices. Part of the reason why there’s a real frantic desire to get another more permanent agreement, which winds up being called Minsk II in February 2015, is because the Europeans really didn’t want to see Russia’s economy fall off a cliff.

Jordan Schneider: Elections matter and leadership matters. I like that you included so many McCain quotes about the events in both Iran and Ukraine, since he could have been president during these years.

Edward Fishman: One of the key ingredients of the success of Obama’s Iran sanctions is the fact that there’s this bipartisan supermajority in favor of tougher sanctions on Iran. Even if Obama had instincts to be cautious or lawyerly, Congress was passing draconian sanctions laws 100 to 0 over a veto-proof majority. With Russia, you had no sanctions laws at all.

What that speaks to, which becomes more important as our story develops, is that U.S. companies had a lot to lose in Russia. It’s not as much of a political winner for members of Congress and senators to try to layer sanctions onto Russia because they might hurt a company in their state or district. We start seeing that maybe there are domestic political limits to how far the U.S. is willing to go with economic warfare.

Jordan Schneider: Commitment to sanctions is a key factor. Secretary Lew once remarked, “One of the things the Russians would say to me is, ‘We survived Leningrad, we could survive this.’ Their definition of what they were willing to tolerate was well beyond the realm of what we would consider tolerable.”

America’s rich, and the pain that we would end up inflicting on ourselves with sanctions would only be like a half percentage point hit to our quality of life. Whereas Russia is starting from a lower baseline, and sanctions hurt them way more than they hurt us. Yet, we’re not comfortable letting ourselves be pinpricked, even if it’s to save the international order.

You wrote…

“With the loss of the Russian market, Lithuania’s dairy industry teetered on the brink of bankruptcy. When a team of State and Treasury officials met with a Lithuanian dairy farmer outside Vilnius in 2015, they expected her to express frustration. She did, but it wasn’t about her declining business. ‘You should be hitting Russia harder,’ she said.”

It doesn’t come down to economics for a lot of this stuff. There are the political economy games of the Texas senator wanting to help out Exxon or whatever, but it often is a question of moral righteousness. We live in rich countries and we can afford to go without, by and large, way more than that Lithuanian dairy farmer could go without.

Edward Fishman: That’s exactly right, Jordan. One of the macro ironies of the book is, the rise of economic warfare in U.S. foreign policy in the 21st century is partly because military force became politically toxic in the aftermath of Iraq and Afghanistan. As those wars were going south, neither Republicans nor Democrats felt like they could even fight limited military engagements, which is very different from the ’90s when there were all kinds of small wars and U.S. bombing campaigns.

Economic warfare initially is seen as more politically palatable because it’s not hurting Americans — we can sanction Iran out the wazoo and there’s no pain felt at home. But then once you get to Russia and even more powerfully once you get to China, there are real political risks for leaders who impose sanctions on these countries. Even a 10% spike in oil prices or a marginal increase in inflation can become powerful factors in the minds of American presidents and wind up constraining our ability to successfully prosecute economic warfare.

Jordan Schneider: That’s a great point. In the 90s, you had the Taiwan Straits crisis where Clinton threw a carrier there and things calmed down. You had Mogadishu, you had Yugoslavia. But there’s this moment in 2014 where the Ukrainians asked, “Can you give us Javelins, please?” The Europeans said no. Blankets don’t win wars, bullets do.

This is the heartbreaking thing — if Russia believed that the U.S. and NATO were really going to put their money where their mouth was in arming the Ukrainians for war number one, maybe they would have been more concerned — not only about the economic impact, which they clearly underpriced, but also the military impact. We have had hundreds of billions of dollars of armaments go to help Ukraine. It was totally reasonable for Putin, based on the track record of the Obama and Trump administrations, to not expect that to be the response when it came to 2022.

Edward Fishman: Looking at the real error of U.S. policy toward Russia, it’s not necessarily anything that happened in 2014 because we were dealing with a completely novel problem, an unexpected crisis. There was no playbook for sanctions on Russia. This is one area where it’s important to be empathetic to Obama and his top team because it wasn’t easy what they had to deal with. The sanctions they did put in place in 2014 wound up being really impactful — Russia’s economy effectively collapsed that winter.

The bigger indictment on American policy is what happened after February 2015 when the Minsk II agreement was signed. After that, the Obama administration took its foot off the gas on sanctions, basically saying they’re just going to maintain what they have in place. Russia very publicly interferes in the 2016 election. Obama had threatened Putin with drastic sanctions if he continued to interfere. Putin continued to interfere, and the sanctions Obama put in place in December right before he left office were really minor. That’s a bad signal.

Then you have four years of the Trump administration in which Trump does nothing on Russia sanctions. It’s a logical lesson for Putin to draw, both from the last year and a half of Obama and all four years of Trump, that he basically got away with the annexation of Crimea at a reasonable cost. That’s just speaking of the U.S. — Europe is even worse. In 2015, after the annexation of Crimea, a consortium of companies signed the Nord Stream 2 pipeline deal to double the amount of gas that Europe would get from Russia. Putin was completely within reason to assess that the West does not have the stomach for a real economic war.

Jordan Schneider: Unlike in Crimea, the U.S. sees this coming in 2022 and has months to try to get its ducks in order, to try to do everything it can to dissuade Putin from trying to take Kyiv. What happened then?

Edward Fishman: When Biden comes in, there’s a real debate amongst his advisors about what to do. Russia had accumulated all of these misdeeds that had gone unanswered. Biden himself, when he was vice president, wanted to arm the Ukrainians. He was the most hawkish member of the top Obama team on Russia, always in favor of tougher military steps to help the Ukrainians, always in favor of tougher sanctions.

There was real debate about what to do. Should they come in right away with really tough sanctions? Biden’s conclusion was that we were still reeling from the COVID pandemic, we had climate change to deal with, and China was the biggest geopolitical issue on his radar. They tried to have what they called a “stable and predictable relationship” with Russia — which is hilarious in retrospect, as “stable” and “predictable” aren’t things you necessarily ever ascribe to Putin’s Russia.

They came out of the gate in April 2021 with a modest increase of sanctions, saying, “Here’s some sanctions to repay you for all these bad things you’ve done over the last six years. But after this, we want stability and predictability.” Putin gets a summit with Biden, which he’s very happy to get. Then he pens a rambling 5,000-word essay about why Ukraine’s not a real country and should be part of Russia in the summer of 2021 while he’s in lockdown. He masses over 100,000 troops around Ukraine’s border that fall.

It becomes quite clear that Putin has designs on Ukraine. In what is probably the biggest intelligence success of the 21st century, the US intelligence community gets Putin dead to rights. They figure out exactly what his plan is, to the point where Biden starts warning American allies privately in September and October 2021 that an invasion is coming. Very soon thereafter, he starts making public warnings that invasion is coming and tries to use the threat of swift and severe consequences, particularly very dramatic economic sanctions, to deter Putin from invading Ukraine.

Jordan Schneider: Let’s talk about how they tried to build that coalition and signal those sanctions in the lead-up to the ultimate invasion.

Edward Fishman: A stroke of luck for the Biden administration was having Daleep Singh, who had played a significant role in the 2014 sanctions. He’s one of the top financial minds in Washington — a city that doesn’t have many people with deep financial markets expertise. Daleep is an exception. He was in the perfect role to orchestrate a sanctions campaign as the Deputy National Security Advisor for International Economics, overseeing the organs of the US Government that do economic warfare.

In late 2021 and early 2022, Daleep builds relationships with his fellow G7 counterparts: in Brussels, Bjoern Seibert, and in London, Jonathan Black. They start getting into the nitty-gritty of what kind of sanctions they might impose if Putin were to invade. This preparation is important not just for being ready to do something real if Putin pulls the trigger, but also for making the threat of deterrence more credible. Russia has a world-class intelligence apparatus — if all you had was Biden wagging his finger saying “You’re going to face really strong sanctions if you invade,” but there’s no actual bureaucratic movement in these capitals creating sanctions ready to go, Putin would probably assess it was a bluff. The preparation that Daleep Singh and his counterparts in Europe and Japan do is very important.

Jordan Schneider: I love how they were doing this like in secret, but also in public. They weren’t being super hard about using classified communications — they were just calling each other on their phones because they actually want the Russians to be listening and believe they are going to put real sanctions on them.

Edward Fishman: That’s exactly right. They view the preparations as important from both a practical standpoint and a signaling standpoint.

By the time we get to the moment of decision in late February, it becomes clear after Putin and Xi Jinping meet in early February that an invasion probably won’t happen until the Beijing Olympics wraps up — Putin doesn’t want to spoil Xi Jinping’s party. By that time, you have a very extensive menu of sanctions options. Most importantly, you have what’s called the Day Zero package — the raft of sanctions that would go into effect as soon as Putin invades.

The compromise is made because inflation is at a four-decade high and there are concerns about oil prices potentially spiking. Biden says they’re going to maximize sanctions on Russia but not aggressively target its oil sales, which is tough because Russia’s economy depends on hydrocarbon exports. The strategy of the Day Zero sanctions is to implement maximalist sanctions on Russian banks — Sberbank and VTB, the two biggest banks in Russia — as well as Russia’s access to foreign technologies. They took the Foreign Direct Product Rule that had been imposed on Huawei in 2020 and recast it to cover the entire Russian economy. They take something that had been previously employed on just one Chinese company and apply it against an entire state.

The tragedy of the situation is that Putin invades and very quickly — similar to that moment in July 2014 after MH17 was shot down — there’s a giant shift of the Overton window in Europe. Everyone becomes gung-ho for very aggressive sanctions after Putin invades and we start seeing just how horrible this war is and how imperialistic Putin’s goals are. Hundreds of thousands of people protest on the streets of places like Berlin, and there’s a massive political movement in favor of stronger sanctions.

Within 24 hours of the invasion beginning, the Day Zero package that Daleep Singh and his colleagues had worked months on looked much too weak and actually undershot the political moment. Within that first weekend of the war, the United States and the G7 agreed to go much further and actually sanction Russia’s central bank directly — something that was seen as too politically radical to even consider in the lead-up to the invasion. Putin clearly agreed because he had left half of his central bank reserves completely exposed to Western sanctions.

Jordan Schneider: This goes back to the mafia diplomacy concept. Ironically, Putin expected the West to be more gentlemanly and concerned about the centrality of the dollar and euro to global trading. Once the war started and the Overton window shifted — which everyone had a hard time foreseeing — things changed. Looking back, it seems silly that they didn’t anticipate massacres when Russia invaded. While sanctioning their central bank was an option, there remained questions about whether they could get the money out, and if they would even believe the threat before it happened. The actual deterrent value we had during those months remains an open question.

Edward Fishman: Clearly, we would have been better off had the U.S. and Europe created more aggressive sanctions plans in advance. This could have strengthened deterrence and weakened Russia’s economy and warfighting capability more quickly, directly helping Ukraine on the battlefield. There were significant costs to underestimating how willing political leaders would be to implement tough sanctions in the U.S. and Europe. But going back to your earlier point, Jordan — from a deterrent standpoint, would that preparation have overridden Putin’s lesson from 2014 and the seven or eight years of basically allowing Russia to get off scot-free after annexing Crimea? Putin had likely already sized this up in his head by then, and I’m not sure we could have changed his mind.

Jordan Schneider: Here’s a crank idea — why didn’t the Treasury Department go long on oil if they were worried about it spiking up to $250 a barrel? Couldn’t you just do the math that way?

Edward Fishman: This is a point I make toward the end of the book — the U.S. is much better at imposing economic penalties than deploying capital for strategic reasons. That would be a very creative use of government resources, but it’s not a bad idea. If we had the flexibility to do something like that in a strategic manner, sure. We do use things like the Strategic Petroleum Reserve to stabilize the oil market. In March 2022, the Biden administration released 180 million barrels of oil to try to stabilize the market.

Jordan Schneider: They did eventually act, but it took too long, and the Department of Energy people are complaining that the caves might crater in. Reading through your book, I can only imagine how frustrating it must be for these officials working around the clock to get the whole world to ramp up sanctions, and they can’t even get their own government to release oil for arguably the biggest crisis in at least 50 years.

Edward Fishman: Many of our institutions are built on the assumption that we live in a peaceful, predictable world, and we don’t always get our act together in time for crisis. This isn’t unique to the 21st century — it’s been true throughout American history.

Jordan Schneider: Here’s another crank idea for you. In the winter of 2023, everyone was terrified that oil prices were going to spike. Did anyone discuss geoengineering solutions, like spraying sulfur in the air over Europe to save everyone’s energy bills?

Edward Fishman: There are a number of tragedies in this story, one being that you decided to become a podcaster instead of a sanctions nerd. Had you gone down this path, maybe we would have benefited from your creativity in the U.S. government.

Institutional Dysfunction

Jordan Schneider: The people you profile, whom you clearly admire for their incredible feats of civil service, were creating new concepts and regimes unimaginable back in 2004 while operating under such constraints in such a dysfunctional system. They made enormous family sacrifices, which you mention several times. We did a show called “Is the NSC Unwell?” where we opened with Jake Sullivan being awake at 4 AM on a Tuesday during a home invasion because he was dealing with Ukraine issues.

Having the idea is the easiest part. Sure, I can suggest geoengineering to fight the impact of Russian oil, but transforming a clever idea that checks all the economic, institutional, and diplomatic boxes into reality is unbelievably difficult. Multiple times in your stories, there are eight-month delays for things that everyone should have immediately approved on day one.

Edward Fishman: We need a government that’s purpose-built for the age of economic warfare. That’s the premise of my book — we are living in an age of economic warfare. Sanctions, tariffs, and export controls are how great powers compete today and will compete tomorrow. This is a secular trend we’ve seen throughout the 21st century, yet we haven’t changed our government to actually fight and win these economic wars.

There’s nothing like the Pentagon for economic warfare. During my short stint at the Pentagon working for then-Chairman of the Joint Chiefs of Staff Marty Dempsey, I noticed that military force has one agency and a clear chain of command up to the Secretary of Defense. With economic power, you’ve got numerous agencies involved — the Treasury Department, the Commerce Department, the State Department, the Energy Department. Much time is spent just coordinating the interagency process.

Ideally, we would have a dedicated department with clear leadership for economic statecraft or economic warfare. Some governments have moved in this direction — Japan now has a cabinet-level minister for economic security. The U.S. hasn’t innovated like that. There’s a core budgetary problem where agencies like TFI (Office of Terrorism and Financial Intelligence) at Treasury, which Stuart Levey led, or BIS at the Commerce Department, haven’t seen significant budget increases despite their missions growing exponentially.

Subscribe now

Jordan Schneider: This theme comes up repeatedly in these stories and with the chip export controls. When cabinet-level officials disagree without presidential direction saying “We’re doing X, not Y, get with the program,” things stall or take longer. Cabinet members are congressionally approved; their words carry weight. When Janet Yellen believes a sanction would harm global inflation and the American economy, Jake Sullivan must call Mario Draghi to persuade her because Biden won’t act without her support. Everyone has different priorities, and without a central authority or an engaged president, you end up with stasis — allowing Russia to make an extra $200 billion they shouldn’t have throughout 2023.

Edward Fishman: Exactly. The Draghi call is one of the more remarkable episodes in the book. After the political aperture expanded during the first weekend of the Ukraine invasion in 2022, making central bank sanctions possible, the G7 agreed. Then Janet Yellen raised concerns, requiring a call from Mario Draghi, Italy’s leader and former European Central Bank chair, to personally assure her it was acceptable.

Regarding China, much of why your podcast is amazing has been its in-depth coverage of chip export controls. Looking back to the first Trump administration, export controls were deployed against Huawei instead of sanctions largely because Treasury Secretary Steven Mnuchin opposed a tough China policy. In early 2019, after the arrest of Meng Wanzhou 孟晚舟, some administration officials suggested sanctioning Huawei and putting them on the SDN list. Mnuchin refused, so they defaulted to putting Huawei on the entity list, which Wilbur Ross controlled as Commerce Secretary. The whole export controls landscape might have been very different with a more hawkish Treasury Secretary during the first Trump administration.

Jordan Schneider: You have this wild anecdote from Matt Pottinger, former ChinaTalk guest who became Deputy National Security Advisor towards the end of the Trump administration.

Pottinger noted that at one point, Bolton decided not to tell Trump about arresting Meng Wanzhou. Pottinger interpreted Trump’s rhetoric as supporting a tough stance on China.

“Pottinger told his Commerce colleagues that Trump was pursuing a two-pronged strategy. On the one hand, the president was seeking to preserve his personal relationship with Xi Jinping and the appearance of pursuing warmer ties. But as for officials in the bureaucracy, Trump ‘wants us punching as hard as we can.’ In effect, Pottinger was telling the Commerce officials to take Trump seriously, not literally — to tune out the verbal concessions that Trump made in public and keep a default position of being ‘tough’ on China.”

Presidents, even those not in their 70s, only have maybe 5% of their day for these matters. This leaves an enormous amount to be sorted out by empowered appointees and cabinet members, which explains how we ended up with export controls instead of sanctions on Huawei — quite remarkable in retrospect.

Edward Fishman: The first Trump administration has been characterized as super hawkish on China, but examining the record shows Trump himself wavered between being very hawkish and totally obsequious to Xi Jinping. The policy was shaped by different factions: people like Pottinger and Bob Lighthizer were tough on China, while Mnuchin and Gary Cohn wanted to return to the early 2000s approach — the Hank Paulson school of U.S.-China relations. These factions took advantage of opportunities when Trump leaned their way to advance their policies. Trump didn’t take a more consistently hawkish line toward China until his final year in office, when he believed Xi Jinping had lied to him about COVID, destroying his re-election chances. We’ll likely see similar dynamics in a new Trump administration — Trump vacillating while different factions capitalize on moments when he’s more receptive to their proposals.

Jordan Schneider: You close the book, Eddie, with the idea of an impossible trinity.

“We don’t yet know when the Age of Economic Warfare will end, but we can envision how. The trade-offs facing policymakers in Washington, Beijing, Brussels, and Moscow can be thought of as an impossible trinity consisting of economic interdependence, economic security, and geopolitical competition. Any two of these can coexist but not all three.”

Walk me through the 20th and 21st centuries — what different trade-offs did states make, and where are we landing now in 2025?

Edward Fishman: Let me explain why I ended the book this way. While I wrote a narrative history because I believe individuals can shape history — remove certain individuals and history would have gone differently — there are also structural reasons underlying the age of economic warfare. Consider this statistic: Barack Obama used sanctions about twice as much as George W. Bush, Trump used them twice as much as Obama, and Biden uses them twice as much as Trump. This suggests both individual agency and structural factors matter.

The geoeconomic impossible trinity I developed explains why this is happening. You can only have two of these three elements simultaneously — economic security, economic interdependence, and geopolitical competition. During the Cold War, we had economic security and geopolitical competition in a bipolar order between the U.S. and Soviet Union, but at the expense of economic interdependence — there was no meaningful economic relationship between them.

When the Cold War ended, geopolitical competition disappeared. China and Russia transformed from adversaries to potential friends, and we invested significant political capital bringing both into the liberal international order, including the WTO and other key international bodies. Without geopolitical competition, we could embrace economic interdependence without sacrificing economic security.

Today, we maintain economic interdependence while geopolitical competition has returned full force, resulting in lost economic security. This affects all major powers — the United States, Japan, European Union, China, and Russia. None feel economically secure, leading them to invest heavily in protecting themselves from rivals’ sanctions, export controls, and tariffs. To regain economic security, we must either end geopolitical competition, which seems unlikely, or significantly reduce economic interdependence. My view is we’re heading toward a significantly less interdependent global economy in the years ahead.

Jordan Schneider: You end the book with some dark words,

“Without the ability to channel geopolitical conflict into the economic arena, great powers could once again find themselves fighting on an actual battlefield. The dream of economic war, for all its downsides, is that it can be an alternative to a more violent kind of war. Someday the age of economic warfare might end, but we might miss it when it’s gone.”

Care to elaborate on this idea?

Edward Fishman: We face very significant stakes in our economic decisions today as we head toward a less interdependent global economy. This could manifest in two ways. First, a world economy where the U.S. and its allies deepen their connections. We might have less trade with China and Russia, but more with Canada, Mexico, the European Union, and Japan. Janet Yellen in the Biden administration called this “friendshoring.” Bob Lighthizer proposed this in a recent New York Times op-ed, suggesting the U.S. and other democracies create a bloc with low internal tariffs and high tariffs on everyone else.

The alternative is deploying sanctions, tariffs, and export controls arbitrarily against friends and foes alike, creating a chaotic breakdown of the global economy. We’d be forced into autarky by default, without long-term economic agreements with allies or adversaries. This scenario frightens me most because history shows that when states can’t secure resources and markets through free trade and investment, the temptation for conquest and imperialism rises.

President Trump’s talk about seizing Greenland for its mineral resources echoes Hitler’s pursuit of Lebensraum. Hitler feared being cut off from European trade after Europeans sanctioned Mussolini for seizing Abyssinia. If economic interdependence unravels into every country for itself rather than friendly blocs, we could see a return to great power war.

Jordan Schneider: Dark. I’ll refer folks back to our two-part episode with Nicholas Mulder on The Economic Weapon, which told that whole 1920s and 1930s story of how Imperial Japan and Nazi Germany developed their autarkic, resource-hungry vision. While racial ideology played a role, they were clearly terrified about accessing enough oil, minerals, and resources to remain great powers.

Researching Modern History

Jordan Schneider: Let’s shift topics. Tell me about writing history of the past 20 years. You don’t have everything declassified, you’re doing interviews, and history seems to be happening in WhatsApp groups. What was it like both as a former civil servant and then interviewing all these people to piece this recent history together?

Edward Fishman: As you know, Jordan, since we shared some classes, I studied history and in a parallel universe might be a university historian. After college, I went into government work and realized that in this era, many decisions bypass formal processes. Even back in the 2010s, decisions were made through informal communications, in coffee shops, never written down, through WhatsApp groups. This has only accelerated since I left government.

Contemporary history plays a crucial role because documentary records won’t be as valuable in 30 years as they were previously. They might even mislead — often the package going into an NSC meeting doesn’t reflect what’s actually discussed or decided. Many decisions happen outside formal meetings entirely.

This experience convinced me that the best approach was to follow Thucydides’ method — write contemporary history, documenting the times you live in, striving for impartiality. What you lose in documentary records, you gain by talking to people who were actually present. Thanks to my government experience and non-partisan reputation, I accessed everyone crucial to this story — Democrats, Republicans, and current civil servants.

Future historians will surely build on and improve the story told in Chokepoints when they access all documents. However, I hope the insights derived from my access to these people and my insider government experience will prove durable.

Jordan Schneider: Did you send Nabiullina an email?

Edward Fishman: No, I didn’t speak to Elvira Nabiullina, unfortunately. One wrinkle in the story is that I was sanctioned by the Russian government in 2022, before I even started writing. I’m currently banned from any travel to Russia.

Jordan Schneider: She’s got an open invitation to ChinaTalk. I’d love to hear her side of the story.

y through declassified documents showing what really happened — I’d bet most of the narrative around U.S. policy holds up. Rather, I hope we’ll see Chinese, Russian, or European versions of Chokepoints. While I capture those stories to some extent, the book focuses on the United States. If counterparts in those systems wrote similar books, we’d have a much more complete picture.

Jordan Schneider: Eddie and I were classmates at Yale, studying ancient history together. I love how you say you’re walking in Thucydides’ footsteps — let’s say we’re doing the same with ChinaTalk. For both of us, Donald Kagan’s classes were among the most formative in thinking rigorously about politics, history, and warfare. Any memories or reflections about his impact in the classroom?

Edward Fishman: One sad aspect of publishing this book is that Don died a couple years ago and won’t have the chance to read it. Of all my teachers, he had the biggest impact, shaping my career in many ways. He even influenced how I teach my class at Columbia on Economic and Financial Statecraft — I use his exact seminar format, with students debating each other’s papers weekly.

The main lessons I learned from Kagan that influenced the book include understanding the role of contingency in history — people and their decisions matter. While many history books focus on impersonal forces, Kagan taught me that structure sets context but free will and decisions can change history’s course. That’s why I focused on the people creating these policies.

Second, chronology matters. You must understand historical decisions within the knowledge available at the time. We tend to judge past decisions with hindsight, but understanding what people knew then reveals more about how history unfolds.

Finally, history itself matters. Kagan said, “Without history, we are the prisoners of the accident of where and when we were born.” Beyond clichés about repeating history, understanding what our predecessors did right and wrong helps us live better lives today.

Jordan Schneider: Another lesson coming through your book is that while we can debate grand strategic decisions, like Biden’s approach, the most human agency appears one or two levels below. Having someone from Goldman Sachs who understands the global insurance market enables implementing policies that might not otherwise be conceived. While we criticize civil servants in today’s America, it’s important to recognize that you can expand government’s effectiveness by empowering the right people to make decisions and analyze questions thoughtfully. For anyone at a career crossroads, read Eddie’s book and understand that your future choices matter.

Edward Fishman: I appreciate that, Jordan. If there’s one takeaway, it’s that government officials’ decisions truly matter. The protagonists I highlighted — Stuart Levey, Adam Szubin, Dan Fried, Matt Pottinger, Daleep Singh, Victoria Nuland — if you remove them from their situations, you’d have very different policies. We were fortunate to have them in those positions. Having more people with diverse skill sets willing to serve in government increases the odds of having the right person in the right place at the right time.

Mood Music—Iranian Pre-Revolution Psychadelic Rock

A friend and past ChinaTalk guest Walter Kerr is trying, ala Fast Grants, to step in the gap to provide funding to the most effective organizations most impacted by the USAID funding cutoff. He and some partners have launched the The Foreign Aid Bridge Fund. I donated a few grand this week and think you should too.

Walter runs Unlock Aid, new think tank that has done some great work to make USAID a more efficient and effective organization. Have a listen to him on ChinaTalk (iTunes, Spotify, YouTube).

On Monday, Xi Jinping hosted a symposium of top business leaders, signaling increased support for China’s private sector.

While concrete policy details have yet to emerge, the meeting featured plenty of strong rhetoric. Premier Li Qiang gave a markedly pro-stimulus speech, arguing, “We must make every effort to enhance consumption… and strive to open up a chain where consumption drives investment, industrial upgrading, employment, and income growth.”

The guest list had plenty of familiar figures, like Alibaba’s Jack Ma, BYD’s Wang Chunfu, and DeepSeek’s Liang Wenfeng. But to understand the implications of this stimulus soirée, we have to look at the lesser-known attendees as well — this event assembled propaganda czars, Politburo tech strategists, and neoauthoritarian academics to mingle with tycoons of semiconductors, chemicals, agriculture, 3D printing, and more.

Today, we’ll introduce some of the deeper cuts — eight from industry and four from government — to illuminate what Xi’s new alliance means for the emerging technology race.

Thanks to Ray Wang for writing the profile on Yu Renrong. All other profiles were authored by Lily Ottinger.

Industry leaders

Wang Xingxing 王兴兴 (Unitree Robotics 宇树科技)

Wang Xingxing is the founder and CEO of Unitree Robotics, which developed China’s most popular robot dog.

Wang’s eccentric personality, technological optimism, and unconventional approach to talent have prompted comparisons to DeepSeek CEO Liang Wenfeng. According to an April 2024 interview:

“I think by the end of 2025, there will be at least one company in the world that can produce a relatively general-purpose robotic large model.
…
In the future, humanoid robots could reshape every industry, including manufacturing, services, market production, agriculture, mining, and construction. Looking ahead to the ultimate possibilities, I believe that governments could fully deploy 100,000 humanoid robots, designate a piece of land, and build a brand-new city there. They could complete the infrastructure and provide housing for free. At that point, ordinary people wouldn’t even need to work — robots could sustain everyone. This is entirely possible.
…
Eventually, they might even be able to create robots the size of cells. … At that point, governments would definitely need to introduce regulations to prevent an uncontrolled explosion of robots — after all, they could end up consuming all available resources.”

Wang is also a believer in the potential of AGI, but thinks that large language models are the wrong way to get there. From a 2024 panel:

“I believe that embodied intelligence is the only path to achieving AGI. Currently, LLMs lack physical presence and thus have an insufficient understanding of the physical world. This is why many top AI researchers advocate for developing world models. Tesla's autonomous driving also relies on collecting real-world data for training rather than using virtual data, because virtual data lacks sufficient real-time interaction with the physical world. I think embodiment is actually a crucial pathway to AGI, and AGI is very likely to emerge from robotics companies.”

How does Xi’s symposium fit into this vision? According to Wang, hard-core innovation requires a skillful PR strategy. “If your technology is not innovative, it is meaningless. Of course, you cannot express this innovation directly. It is better not to go beyond the public's cognition too much, otherwise I think I would be scolded to death.” The publicity campaign goes beyond shaking hands with Xi — Unitree’s latest humanoids performed during the 2025 CCTV New Year’s Gala, which is of course the world’s most-watched television program.

For more on China and robots, see our features on industrial robotics and humanoid robots.

Xu Guanju 徐冠巨 (Transfar Group 传化)

Billionaire Xu Guanju is the founder and chairman of Transfar Group, a publicly traded conglomerate dealing in chemical manufacturing, supply chain logistics, agricultural biotechnology, and even finance.

Xu spent a decade as chairman of the Zhejiang Federation of Industry and Commerce, advising the local government on behalf of the private sector, and helping to shape the “Zhejiang model” of economic development that incubated both DeepSeek and Unitree. He also served as a member of the CPPCC (China’s top advisory body) and was elected as a deputy to the 13th National People’s Congress (NPC).

Subscribe now

Xu famously built his chemical empire on a humble foundation. After borrowing startup funds from friends and family, Xu began producing liquid soap in a small workshop with his father in 1986. According to one well-known story, Xu initially had to contract a chemical engineer due to his lack of expertise. This contractor proved frustratingly indispensable thanks to a mysterious powdered thickening agent he added to finish each batch of soap. Eventually, Xu agreed to pay 2000 RMB (~10,000 RMB today) for access to this “trade secret,” only to learn that the powder was just ordinary table salt. Xu’s father responded to this revelation by exclaiming, “I spent all this money to send you to school, and for what?” Xu said later, “This incident made me realize that technology is the primary productive force.”

Peng Fan 彭凡 (KOCEL 共享装备)

Peng Fan is the chairman of KOCEL Group, the machinery manufacturing company that produced the hydropower turbine blades for the Three Gorges Dam.

In 1983, Peng moved to the remote northwest region of Ningxia to work at the state-owned Great Wall Foundry​. He climbed the ranks from an ordinary casting technician to plant manager. The foundry underwent privatization and restructuring in 2003, with Peng leading the effort to transform it into KOCEL.

Peng has a postgraduate degree in casting engineering, and under Peng’s leadership, KOCEL embraced cutting-edge “intelligent manufacturing” techniques. He championed the adoption of industrial 3D printing, obtaining more than 500 patents related to advanced casting techniques.

Peng’s team cracked several difficult engineering problems, including the manufacture of heavy gas turbine casings and hydropower turbine blades. Under his guidance, KOCEL has grown into a leading global supplier of high-end equipment components.

He also served as a delegate to the National Congress of the CCP in 2012 and 2017.

Jiang Bin 姜滨 (Goertek 歌尔股份)

Jiang Bin is the chairman of Goertek, a company he co-founded with his wife, Hu Shuangmei 胡双美, in 2001. Goertek is one of the primary manufacturers of Apple’s AirPods and Vision Pro headsets.

Jiang Bin was born in 1966 in Shandong province. He earned a bachelor’s degree in engineering from the Beijing University of Aeronautics and Astronautics and later an MBA from Tsinghua University​.

With Jiang as chairman, Goertek grew from a small acoustics firm into a global supplier of microphones, speakers, sensors, and other hardware. The company has filed more than 29,000 patent applications and is now the world’s top supplier of micro speakers, MEMS acoustic sensors, and AR/VR headset components.

Apart from Apple, Goertek’s clients include Meta, Amazon, Google, Samsung, and Sony — and in turn, the company has been criticized for relying too much on the patronage of these foreign tech giants. In 2024, the company announced an investment of US$280 million to build new production capacity in Vietnam.

Jiang is currently serving as a deputy to the 14th NPC and regularly participates in government-organized industry forums. He frequently uses these platforms to promote metaverse technology. Jiang’s current net worth is reportedly more than US$5 billion.

Yu Renrong 虞仁荣 (Will Semiconductor 韦尔半导体)

Yu Renrong (虞仁荣) is the founder and chairman of Will Semiconductor, one of the top 10 fabless chip companies in the world. It is also one of the largest global image sensor providers, ranking only behind Sony and Samsung.

Yu grew up in a small town in Ningbo, Zhejiang, but his humble background did not prevent him from demonstrating his brilliance. His academic abilities brought him to one of the best high schools in his hometown, and he graduated with an Electonic Engineering degree (known at the time as the Department of Radio 无线电系) from Tsinghua University.

After gaining engineering experience at Chinese IT pioneer Inspur and sales experience at Yuelong Electronic Technology, he founded his first company (Beijing Huaqing Xingchang Technology and Trade) in 2006.

Subscribe now

He then founded Will Semiconductor in 2007, which both designs advanced semiconductors and sells electronic components. After its listing in 2017, Will Semiconductor rapidly expanded through acquisitions, including of OmniVision Technologies, CelePixel, Cerebrex, and Synaptics’s Asian Touch and Display Driver Integration business.

Today, Will Semiconductor's market value is reportedly more than US$27 billion. Yu is also recognized as China's richest chip tycoon (中国芯片首富), with a personal net worth of $6.4 billion in 2024.

Qi Xiangdong 齐向东 (Qi-Anxin 奇安信)

Qi Xiangdong is the founder of Qi-Anxin Technology Group (奇安信科技集团), which provides cybersecurity services to government agencies, commercial actors, and critical infrastructure facilities.

After serving as Vice President of Yahoo China from 2003 to 2005, Qi co-founded the consumer antivirus software company Qihoo 360 (奇虎360), which was added to the BIS entity list in 2020.

Qi is deeply integrated into China’s cybersecurity establishment, and he’s currently a member of the 14th National Committee of CPPCC and Vice Chairman of the ACFIC. He is also a vocal supporter of AI safety regulations, frequently expressing concerns about AI-enabled hacking, deepfakes, and the black box problem.

Leng Youbin 冷友斌 (Feihe Dairy 飞鹤乳业)

Leng Youbin is the CEO of Feihe (literally, “Flying Crane”), China’s leading infant formula and dairy company. He transformed Feihe into a top domestic competitor in a market once dominated by foreign brands.

Under Leng’s leadership, Feihe worked to cultivate trust among Chinese consumers in the wake of the 2008 milk scandal, investing in high-quality milk source bases and product R&D. Feihe has also implemented agricultural IOT practices, including collars that monitor the vital signs of dairy cows for early disease detection.

Just like how the CEO of Chobani served on the Homeland Security Advisory Council in the US during the Biden administration, Leng served as a deputy to the 13th NPC and as the vice-chairman of the All-China Federation of Industry and Commerce (ACFIC).

Liu Yonghao 刘永好 (New Hope 新希望)

Liu Yonghao 刘永好 is the founder and chairman of New Hope Group, one of China’s largest agricultural companies. New Hope manufactures livestock and aquaculture feed, farms and processes meat and dairy products, and even dabbles in finance and real estate.

Born in 1952, Liu began his career as a teacher. In 1982, he started a poultry breeding operation in rural Sichuan with his brothers. By 1992, their company was among the largest non-governmental conglomerates in China. Today, Liu’s net worth is more than US$6 billion.

Under Liu’s leadership, New Hope expanded from animal feed into a broad agrifood empire. He also co-founded China Minsheng Bank in 1996, one of China’s first private banks, and served as its vice-chairman​.

Liu’s political activities include serving as a committee member of the CPPCC, a deputy to the 12th NPC, and vice-chairman of the ACFIC. In the past, he’s publicly spoken out against Xi’s detentions of businesspeople.

Government Officials

Shi Taifeng 石泰峰 (PB Member, United Front Work Chief)

Shi Taifeng is a CCP Politburo member and the head of the Party’s United Front Work Department (UFWD), an organization tasked with monitoring and influencing elite groups outside the Party, including businesspeople, academics, ethnic minority leaders, and the overseas Chinese diaspora​ — which made him a natural choice to receive an invitation.

A trained legal scholar with a Master of Laws from Peking University, Shi spent part of his career as a professor at the Central Party School​. Before rising to national prominence​, he held key regional posts – including governor of Jiangsu province and Party Secretary of first Ningxia province (2017–2019) and then Inner Mongolia (2019–2022).

In Inner Mongolia, Shi oversaw a massive buildout of data centers, 5G infrastructure, and renewable energy as part of a broader strategy of anti-separatist digital governance in the region. His tenure also saw intense crackdowns on the rights of ethnic minorities and an expansion of AI-enabled censorship. In August 2020, for example, the provincial government announced a plan to force Inner Mongolian schools to teach certain subjects in Mandarin (the goal was to transition to an exclusively Mandarin curriculum by 2023, although that was publicly denied at the time). Authorities responded to complaints by blocking China’s only Mongolian-language social media platform, Bainu, and tracking down protesters.

As Ningxia Party Secretary, Shi launched an “innovation-driven strategy” of economic development and warned that not innovating is “a dead end.”

Wang Huning 王沪宁 (PBSC Member, CPPCC Chair)

Wang Huning is a top leader sitting on the CCP’s 7-member Politburo Standing Committee and currently serves as Chairman of the CPPCC. We did a podcast on Wang with Chang Che, who wrote a great profile of him a few years back (iTunes, Spotify, YouTube).

Wang made his name as a scholar and theorist rather than a regional administrator. He was a professor of international politics at Shanghai’s Fudan University, where he was a well-known advocate of “neoauthoritarianism” (新权威主义) and authored widely read books including Analysis of Comparative Politics, Analysis of Contemporary Western Politics (1988), and America Against America. His undergraduate degree is in French.

Over the past three decades, Wang has been the de facto chief ideologue for three consecutive Chinese presidents — Jiang Zemin, Hu Jintao, and now Xi Jinping​. He’s credited with formulating key political concepts like Jiang’s “Three Represents,” Hu’s “Scientific Outlook on Development,” and Xi’s “Chinese Dream” as well as “Chinese-style modernization.” He is the Party’s top ideological craftsman, “using cosmetics to dress up political policies” with unprecedented longevity across administrations.

Wang is an active proponent of AI development, which he of course frames in ideological terms. His presence at the event signals that private sector innovation will be valued as a key component of grand strategy and national power.

Ding Xuexiang 丁薛祥 (PBSC Member, 1st-Ranked Vice Premier)

Ding Xuexiang is a member of the Politburo Standing Committee and the first-ranked Vice Premier in China’s State Council. He’s also one of Xi Jinping’s closest confidants.

Unlike many Chinese politicians, Ding has a technical education. He studied engineering and worked as a materials science researcher and administrator in Shanghai’s science bureaucracy for years​. He is the only trained engineer on the current Standing Committee​. His role at the event was to make the tech entrepreneurs feel less out of place.

Ding is also the director of the Central Science and Technology Commission (CSTC), a high-level CCP body unveiled in 2023 and tasked with coordinating China’s national science and tech strategy. His contributions have thus far been organizational — setting up and leading the new governance structures for innovation and mobilizing resources across government, academia, and industry.​

Here’s Ding’s take on AI regulation:

Ding Xuexiang said that emerging technologies such as artificial intelligence can be a powerful driving force for development, but they can also be a source of risk…. We will not blindly follow the trend, nor will we participate in unrestrained international competition. China has a strong governance and regulatory system and institutional measures, and we are confident that we can manage and use artificial intelligence technology well.

Ding Xuexiang said that global governance of artificial intelligence is a global problem. If countries are allowed to compete in an unorderly manner, the “gray rhino 灰犀牛” is right in front of us. Historically, the United Nations has played a good role in controlling nuclear safety and biosafety, and its successful experience is worth learning. The United Nations should be supported to play a central role, and all countries should participate together to jointly study and formulate powerful and effective rules to ensure that new technologies such as artificial intelligence become “Ali Baba’s Cave” of treasures, rather than “Pandora's Box.”

Li Shulei 李书磊 (PB Member, Propaganda Chief)

Li Shulei is a Politburo member and the head of the Party’s Central Propaganda Department. Hi has described modernization as a Western imposition that has now become a necessity for national power.

Li was a child prodigy, entering Peking University at age 14 and later earning a doctorate in modern Chinese literature. He spent many years as a professor at the Central Party School, writing on Chinese literature and culture​ before serving as propaganda chief of Fujian province. In 2022, Xi Jinping tapped him to take over the top propaganda post​.

In 2023, Li’s Propaganda Department (along with the Cyberspace Administration) issued guidelines to ensure AI-generated content aligns with socialist values and does not undermine social stability. In his words:

“Generative artificial intelligence is one of the most revolutionary and leading scientific and technological technologies at present. We must improve the development and management mechanism of generative artificial intelligence as soon as possible, promote industrial development, technological progress and security in this important field, and achieve benefits and avoid harm and safe use. Cyberspace is not a lawless place or an enclave of public opinion. We must strengthen the construction of the rule of law in cyberspace, improve the long-term mechanism for network ecological governance, and ensure that the Internet always operates healthily on the track of the rule of law.”

As the propaganda czar, Li was invited to the meeting because events like these are more about messaging than substance. Going forward, a key indicator of the meeting’s impact will be whether Li’s propaganda continues to discuss technology in terms of control or instead pivots to emphasizing the economic benefits of tech innovation.

A friend and past ChinaTalk guest Walter Kerr is trying, ala Fast Grants, to step in the gap to provide funding to the most effective organizations most impacted by the USAID funding cutoff. He and some partners have launched the The Foreign Aid Bridge Fund and are looking to give out their first tranche of money Friday. I donated a few grand this week and think you should too.

Walter runs Unlock Aid, new think tank that has done some great work to make USAID a more efficient and effective organization. Have a listen to him on ChinaTalk (iTunes, Spotify, YouTube).

How do patents influence emerging technology innovation? How far could AI and DOGE push our current IP regime? Does it matter that China issues way more patents than the US does?

To discuss, ChinaTalk interviewed ​​Andrei Iancu, director of the US Patent Office under the first Trump administration. Andrei has degrees in aerospace and mechanical engineering, and worked at the legendary Hughes Aircraft Company before going to law school. He is currently in private practice at Sullivan and Cromwell.

Co-hosting today is ChinaTalk editor and second year law student at Duke, Nicholas Welch.

Have a listen on Spotify, iTunes, or your favorite podcast app.

We get into…

• The mounting evidence that China's patent system now dominates America’s, and whether these indicators constitute an emergency in the innovation ecosystem,

• Why some US companies now prefer Chinese courts for patent enforcement,

• The fundamental tension between private rights of inventors and public access to innovations,

• What congressional inaction on patent eligibility means for AI innovation, and the bills that congress could pass to immediately jumpstart emerging tech investment,

• What the current administration could do to help USPTO juice the economy,

• Controversy surrounding the Patent Trial and Appeal Board (PTAB), and whether DOGE could put PTAB on the chopping block,

• How Trump will approach patent law and intellectual property rights, including perspectives on appointments and potential reforms.

Thanks to CSIS for partnering with us to bring you this episode, the first in a three-episode CSIS Chip Chat series.

Legislative Omissions and the Political Economy of Patents

Jordan Schneider: Let’s start off with the central contradiction in patent law. What are the two equities that this whole legal superstructure is trying to balance?

Andrei Iancu: Patent law has existed in the United States since the founding of the country. It’s in the body of the Constitution in Article I, Section 8, Clause 8. This is the only place in the entire Constitution where the word “right” is mentioned other than the Bill of Rights. They thought it was that important. The Patent Act of 1790 was the first law passed by Congress after the country was founded. Since then, it has been a central part of the United States economy.

Patents, and in fact all intellectual property rights, balance the private right of the individual creator versus the public right to access that creation. On one hand, it gives exclusive rights, as the Constitution says, to one’s inventions on the technical side, or one’s artistic creations on the copyright side. In exchange for that exclusivity, the creator makes the invention public, and the public has the right to see it and potentially use it.

This is the quid pro quo. This tension between the private right of protection versus the public’s right of access has existed from the very beginning. Thomas Jefferson was the first head of the patent system beginning in 1790. He was reviewing patents at night, and at that time, he said patents were, “an embarrassment” to the American economic system, because they are a sort of monopoly, which he hated. He was very uncomfortable with removing the public’s ability to freely access ideas. However, around the same time he also said that patents have “given a spring to innovation beyond [his] conception.” That tension in Jefferson embodies the tension that exists in the patent system to this day.

Jordan Schneider: Over the course of your career, what role has each branch of government played in balancing the rights of creators and consumers?

Andrei Iancu: Congress creates the laws, the administration enforces and administers those laws, and the judicial branch has to interpret those laws. It begins with Congress — they have to make the patent laws in the first instance, and they’ve been struggling from the very beginning.

Once the laws are passed, it’s up to the USPTO (Patent and Trademark Office) to enforce those laws and grant patents and trademarks, while the Copyright Office registers copyrights based on those laws. The problem has been that these issues are so complicated, and the tension between the two poles I’ve mentioned is so high, that Congress has left a lot of gaps. They have been incapable of legislating.

For example, the first substantive section of the patent code is Section 101. It defines which types of technologies are eligible to receive patents. It is so complicated it hasn’t been legislated on since 1793. The last time Congress wrote a law to say what technology is in and out was in the 18th century. Now, the Patent Office and the courts are trying to figure out how artificial intelligence and DNA processing fit into this 18th-century statute.

It’s a mess because Congress hasn’t returned to it in 250 years. There is a bipartisan bill to address this issue called PERA (Patent Eligibility Restoration Act). It was introduced in the Senate last year by Senators Tillis and Coons, Republican and Democrat respectively, but the bill didn’t move. We’ll see if they introduce it again. It hasn’t been touched since 1793, and it really is important that they get to it, but the issues are really hard.

Jordan Schneider: Let’s stay on the political economy of this. On one hand, you have the political economy fights over who gets to make more money — is it the generics, the healthcare system, or the biomanufacturing companies? Then you have this philosophical arc happening above all the individual industry fights. I’m curious, Andrei, as the pendulum swings, are the changes within specific industries, or is it a broader national shift over the decades where you go from the system more supporting the patent holders versus the patent users?

Andrei Iancu: Let me make something clear here. There is only one patent system in the United States, and by definition, it has to apply equally to all technologies. You cannot have patent laws for pharma that are different from patent laws for tech. Whatever the laws are, and however you interpret them, they basically have to be the same across technologies. This concept of non-discrimination across technologies is part of international agreements that the United States has been pushing for a very long time. The TRIPS agreement is a multilateral international agreement where all the member states have ratified it, and it’s fundamental to the patent system.

Jordan Schneider: Why is that?

Andrei Iancu: By definition, you don’t know where technology is going to go, and the whole point of the patent system is forward-looking. You can’t start picking winners and losers ahead of time — that’s the main reason. The other reason is one country might want to discriminate in favor of one industry, while another country might want to discriminate in favor of another industry, and it would be completely unworkable. There are other reasons too.

Nevertheless, even though the laws ultimately have to be uniformly applicable to all technologies, there are certain technologies that drive change in the system. For example, the major tension for the last couple of decades has been between big tech and pharma. This is a complete overgeneralization, but by and large, you could say that the pharma industries and life sciences industries want and need stronger intellectual property rights for a variety of reasons. Whereas big tech companies, by and large — again, I’m super generalizing here — tend to want weaker intellectual property rights.

Whoever prevails in that fight for their own corporate interests will affect everybody. That’s the interesting thing. It’s not like tech can demand changes that only apply to tech, or pharma can demand changes that only apply to pharma. Generally, this has been the tension between these big industries, and the pendulum has been swinging back and forth according to who has had more political power in the last few decades.

Nicholas Welch: Maybe we can stay on emerging tech for a moment. I’d be curious for your take on patent policy with regard to biotech. Let me know if this characterization is true, but your approach to patent law differed from your predecessor, Michelle Lee, and your successor, Kathi Vidal, in a few ways. You might be described as very supportive of patent rights, whereas Lee and Vidal could have been concerned with patent quality. Vidal made policies which caused invalidations of patents to rise from 59% in 2021 up to 71% in the first half of 2024.

In the context of emerging tech specifically, do you feel like companies have good reasons to demand patents, or should USPTO be cautious about issuing patents to technologies we don’t quite understand yet?

Andrei Iancu: Let me challenge the premise a little bit. It is true that invalidations in post-grant proceedings at the Patent Office have risen in the last administration with my successor, and I don’t doubt the numbers you cited. But that does not indicate an increase in patent quality. Those two are completely separate things.

I very much am in favor of and always spoke about the importance of issuing and maintaining correct rights, but that goes both ways. Just because you’re invalidating a patent doesn’t mean you’re increasing the quality of the patent system. The Office could be wrong in invalidating that patent, and that is a mark of lowering the quality of the system as a whole.

Recently, the Sunwater Institute published a report that shows the Patent Office errs significantly more in the direction of incorrectly not granting patent rights than incorrectly granting patent rights. While the incorrect grant of patent rights is a few percentage points in the single digits, incorrectly denying patent rights is in the double digits, according to the recent Sunwater Institute report, and I commend you to that study.

The question then becomes, which one is worse for the economy? You don’t want errors either way. But if you’re going to err on one side or the other, which one are you going to choose?

Right now, the Office is erring too much on the side of incorrectly withholding patent rights or incorrectly removing patent rights once they have been issued. That can do tremendous harm to innovation and investment in the United States.

In a free market economy like we have here in the United States, there are very few incentives that enable investment in risky new technologies. By definition, innovation is risky — it’s new. You don’t know if it’s going to work or not. Probably 8 out of 10 new inventions fail for one reason or another, and some of these inventions are really expensive to bring to market.

Drugs are a very good example. It costs, on average, about 2.5 billion dollars to bring a particular new drug to market by the time you do the basic science research, all the human studies, the FDA approvals, the marketing, and so on. It’s super expensive, and a lot of those ultimately fail. Now, in addition to all that, if they succeed, a lot of these innovations are easily replicable. A drug is very easy to replicate — you can just reverse engineer the chemical formula usually, and there you have it.

Because of that risk, high cost to bring it to fruition, and easy replicability if you succeed, what incentivizes the investment community and the innovation community to invest time and resources in these risky new technologies as opposed to investing in something else like the old stuff, the tried and true stuff, or opening another restaurant? In a free market economy, there’s not much incentive without the protection of a strong patent system or strong intellectual property system at large.

We risk losing the investment and innovation engine in the United States. This is a big problem with new technologies like artificial intelligence, biotechnology, and quantum computing — things that are long-term, very risky, and very expensive to bring to fruition. If we don’t maximize our innovative output and investment input into that innovation economy, we’re going to be left behind because we are in a humongous technological race with China and others.

Nicholas Welch: You mentioned in the context of drugs, they’re very easy to reverse engineer — you can just take the pill and analyze it. Semiconductors, on the other hand, are super hard to reverse engineer. When I read through Chris Miller’s Chip War, it sounds like the reason the Soviets couldn’t keep up with our chips is they’d get a chip and it would take them forever to take it apart and figure out how they made it. Once they figured out how it was made, they didn’t have the right tools to make it themselves, and by the time they got that, we were already onto the next generation of chips. Curious for your thoughts about how the patent system is valuable or not to the semiconductor industry, especially because those products are just so complicated to manufacture.

Andrei Iancu: They’re very complicated to manufacture and more difficult to reverse engineer for sure, for the reasons you’ve mentioned. However, it’s really expensive to get them going. Look what’s happening right now with the CHIPS and Science Act — we’re trying to get companies to invest to create new plants in the United States. It’s really hard to get this off the ground. It takes decades to bring one of these plants to fruition and tens of billions of dollars.

It’s one of these things where, yes, it’s more difficult to reverse engineer on the back end — it’s not impossible, but it’s more difficult. The investment risk is so high. I was a supporter of the CHIPS and Science Act because I think the United States needs to do more, and a lot more, to create these new technologies and stay competitive.

However, there was no intellectual property provision really in the CHIPS and Science Act — patents were barely mentioned. If you do not combine this financial investment that CHIPS and Science authorizes with a strong intellectual property system, it’s just not going to happen. The private sector investment will not come along at scale. Some of it will for sure, but at scale it won’t come along to co-invest with the public funds, and you cannot do it with public funds only.

Unfortunately, what I’m saying is turning out to be correct. We are investing in the CHIPS and Science Act, but it’s kind of like putting money through a sieve because it’s not going to take unless you have a robust intellectual property system. You need to strengthen the system and give guarantees to the investors that if they co-invest with the United States together with the public funds and they invest at scale and we create these plants, we will not lose the IP to China or somebody else and we’ll be able to enforce it if necessary.

Jordan Schneider: Coming back to a comment earlier, I’d be curious for your take on the amount of discretion that the Executive branch and the USPTO (US Patent Office) has in particular. How far, without new legislation or some new Supreme Court ruling, could an administration potentially push it in one direction or the other?

Andrei Iancu: The PTO director and the administration have some discretion to move that dial, but not complete discretion. The administrator is always bound by the legislation on one hand and then by the courts interpreting that legislation on the other hand. Within those parameters — what the law is and how the courts have interpreted it — there is always flexibility, and the PTO director can dial those things up.

Just as one example from the very first point we discussed earlier in the program, the PTO director certainly has the ability to institute policies at the PTO, institute examination guidelines, and train the examiners to make sure that we grant correct rights — that we don’t grant patents that should not be granted, and at the same time we don’t deny patents that should actually be granted.

On the back end, when reviewing already issued patents, the PTO director has discretion to dial how many or what types of reviews and patents we will take on and exactly what the review criteria will be, again within the bounds of the legislation and as it’s been interpreted by the courts. Just as Nicholas identified in the statistics earlier, you can see from administration to administration — the last administration, for example, canceled proportionately more patents than my administration did in the first Trump administration.

There’s definitely that bandwidth. Unfortunately, the less Congress legislates, the more discretion there is for the director. I’m not saying that’s a good thing — I don’t think it is. Patents and IP in general are long-term assets. Companies need to make investments in these assets for the long run, and not knowing how some of these rules are to be interpreted is not good for the economy.

I’ll give you just one example. There is a bill called the PREVAIL Act, which would tighten the laws surrounding post-grant reviews (PGRs) and IPRs, which we’ll talk about. These are proceedings where the office reviews whether patents should be left in force or canceled. This bill tightens the laws surrounding those proceedings. I’m a big supporter of that bill because it removes some discretion from the Patent Office and sets in law what those guidelines should be in many ways. The public will have a higher level of predictability surrounding the IP they have to deal with. To me, that would be a significant improvement, but we don’t have that right now. The director has significant discretion on those proceedings.

China’s Patent System

Nicholas Welch: Let’s talk about China. China has their own patent administration system called the China National Intellectual Property Administration (CNIPA).

A talking point we hear all the time is that they grant a lot more patents than the USPTO does — and that’s true. The US has a test for patents based on abstractness, whereas the Chinese authority reviews the invention as a holistic whole and focuses on the technical solution. In 2023, there was a study that said more than 12,000 cases had been granted in China and Europe but denied in the United States on statutory subject matter grounds.

Should we care that China grants a lot more patents than the US, and is there anything we could learn from how China runs their patent system?

Andrei Iancu: Yes, we should care. It is a concern to me that somehow, they’ve created a patent system that seems to be more robust than ours in some respects. China continues to steal IP at extremely high rates in many different ways, but on top of that, they also have their own innovation ecosystem, and they’re maximizing it to the extent possible for their economy. One of the things they’ve done over the past couple of decades is systematically improve their patent system, and in some ways they’ve overtaken us when it comes to some of these protections.

You touched on one of them — the Chinese laws when it comes to subject matter eligibility. In other words, what technologies are subject to patent and which ones are not. Their system is more clear and more robust than ours, which, as I said at the beginning of this conversation, has not been legislated since 1793. The Chinese system, when it comes to subject matter eligibility, is new and fresh.

That absolutely has an impact, particularly for our economy. If we want to maximize innovation in the United States and maximize investment in that innovation in our free market system, we need clear laws. We need to know the rules of the road and make sure that industry believes its investment will be protected by the rule of law. Right now, on things like subject matter eligibility, the rules of the road are unclear. You’re making all these investments and you just don’t know if it will be protected down the road, if it’s in or out of the patent system, or how it will be interpreted. That uncertainty, at least on the margins, is depressing our ability to invest at scale in these risky new technologies. That puts us at a disadvantage with China.

If you look at the number of patents Chinese companies have versus the United States and graph it out, it should be frightening to anyone in the United States that cares about these issues. If you start graphing 20 years ago, the Chinese were at the bottom, barely registering on the scale of patent grants worldwide, and the United States was at the top or among the top. But take that out to the present day — about 10 years or so ago, there’s a hockey stick effect that comes into play with the number of Chinese patents or patents granted to Chinese companies.

The United States is basically flat for the last 20 years — by and large, the numbers of patents to US companies are growing pretty much 2-4%, in line with GDP growth. The Chinese numbers show a hockey stick effect about a dozen years ago. It rises and then overtakes the United States about a decade ago, and now they’re blowing us out of the water. They’re blowing us out of the water across the board, but more importantly, in the technologies that really matter. For example, they’re getting six times as many patents in deep machine learning as the United States. This relative positioning applies in almost all the new areas of technology, and that should be a significant concern.

Jordan Schneider: When I see charts like that, my first mind goes to Goodhart’s law — what kinds of incentives are being set up here? It really comes down to a question of patent quality. To what extent is the amount of patents issued at the national level a good proxy for innovation? Is it better to look at the market cap of Baidu versus Meta to address that point?

Andrei Iancu: It’s a very good point. I don’t think you should just rely on the number of patents to make a definitive determination as to who’s winning the race in a particular area of technology. But it is one indicator, and it’s a really good leading indicator of where the technology will be in a little while.

Regarding patent quality — that’s the talking point that people who want to weaken the United States patent system make, which is, “Okay, they’re getting a lot more patents, but they’re weak patents, they’re bad quality patents.” I say, what’s your evidence for that? Show me the study that on average the Chinese are getting weaker patents worldwide compared to United States companies getting their patents worldwide.

I don’t think there is a reliable study out there that shows that. It’s not like all of our companies are getting the best patents in the world, either.

Jordan Schneider: With seven thousand a week, you’ll get a lot of duds too, you know.

Andrei Iancu: The Chinese are now getting three times as many patents as the United States. If you just eliminate two-thirds of those — okay, we’re even. Are there that many really terrible patents that Chinese are getting? Where’s the evidence for that?

More importantly, it’s one indicator. We need to look at all the other indicators, and all the other indicators point in the same direction, with no exception. Chinese scientists and engineers are authors of more peer-reviewed scientific and technical journal articles than American authors. The Chinese are graduating many more scientists and engineers every year than the United States — not close, many more. The Chinese are beginning to take leadership roles in standard-setting committees at extremely high rates.

Every indicator points in the same direction. We can start to dismiss one or the other for one reason or another, but I don’t think the United States should lose sight and shrug its shoulders and say it’s just fiction numbers.

The evidence bears that we are in a tremendous technological race, and they are making tremendous progress.

We better attend to it right away and pull every possible policy lever to incentivize the maximum innovation output in the United States. Otherwise, we’re going to wake up in a year or two or five, and it’s going to be too late.

Nicholas Welch: Another indicator that supports what you were saying — this is from a Hoover Institution report in 2023 — PRC courts provide injunctive relief in nearly 100% of all successful patent cases. Specialized IP courts in China provide rapid rulings and readily issue injunctions. In fact, US companies often now sue in PRC courts when they have a choice of jurisdictions in order to obtain the injunctive relief no longer available in the United States. That blew me away — if American companies are choosing to sue in Chinese courts over patent infringements or to obtain injunctive relief, maybe we could start reforms and have people sue in United States jurisdictions. That’s huge.

Andrei Iancu: That is right. It’s yet another way we have weakened our intellectual property system. It is very difficult to obtain injunctions for patent infringement in the United States. If you think about what a patent is fundamentally, a patent, like any property right, is the right to exclude. If I have a house, I have the right to have a front door with a lock and exclude people if I don’t want them to come in. It’s a fundamental right of any property right. If I have a field, I have the right to put a fence around it and exclude other cows from coming to graze on my grass if I want it, or I have the right to charge people to come in and graze on the grass or do whatever I would like with my property.

Same with intellectual property. In fact, as I said at the very beginning, the Constitution says that patents and copyrights are meant to give inventors and authors their exclusive right, which means the right to exclude and the right to an injunction. But since the eBay decision from the Supreme Court about 15 years ago, that has been very difficult to obtain in the United States. Whereas in China, for example, or in Germany, it’s almost automatic, the way it used to be in the United States all along, until recently.

That basically diminishes the value of American intellectual property. It makes it less valuable and therefore less useful to inventors and investors as they contemplate what to work on and what to devote their energies to. That is very detrimental to our innovation economy. There is a bill out there, bicameral, bipartisan, called the RESTORE Act, to fix this issue. But again, it was introduced late last Congress and it hasn’t moved yet.

Nicholas Welch: In US courts, when you say it’s harder to get injunctive relief because of eBay, does that mean the alternative is just money damages? A company could say the cost of litigating and even losing would be cheaper overall — as long as a court can’t actually force me to stop using the patent, I may as well just go through the litigation. Is that what we’re referring to here when you talk about weakening patent rights?

Andrei Iancu: Yes, exactly. If you don’t have the right to exclude and all you have is the ability to charge a fee, then you are into what’s called a compulsory licensing system. That’s not really a property right.

Imagine you have your house, Nicholas. The law says in your town that you can have your house, God bless, but you can’t kick anybody out. Then along comes Jordan, and he says, “I’m setting up in your back bedroom. You can’t kick me out, but you know what? I’ll pay rent. I’m going to just live with you. I’m going to come in with my wife and my three kids and my mother-in-law and we’ll just live in your house and we’ll pay rent.”

How valuable is that house? You’re most likely going to move from that town if that’s the rule. Most likely you’re not going to buy that house or make that investment if people can come set up in your house, even if they want to pay rent. No, thank you! If I want to rent it out, I will, but you can’t force me to rent it out.

This is such a fundamental principle for real estate, real property, or personal property. Same thing with personal property — like a watch. I’m going to come and just take your watch, but I’ll give you ten bucks a month and I’ll just take your watch and wear it wherever. Who even thinks about these things? It’s laughable.

But when it comes to intellectual property rights, the American courts recently don’t have a problem saying you don’t have the right to exclude. It makes no sense, and as a result, on the margin it devalues that property right. It’s irrelevant that you can also charge rent on it. Sure, it’s helpful — it’s better than nothing, definitely better than nothing. But it is such a humongous devaluation of that exclusive right that the Constitution contemplated.

Trump and IP

Nicholas Welch: If we don’t have predictability, maybe we can get some predictions from you about the next few years coming up, because there are so many policy winds that blow toward the USPTO office. Trump, for example, has nominated David Sacks as the brand new position of AI and Crypto Czar. Sacks is presumably a “regulate AI less” kind of person. Trump’s pick for assistant AG for antitrust, Gail Slater, has been quoted saying she wants to bring back the so-called “New Madison” approach — no duty to license patents, and also that standard essential patents (SEPs) should get the same protections as other patents, which presumably stands for strengthening the rights of patent holders.

Within the leeway that we have, which seems quite large, how do you think an incoming USPTO director is going to handle some of these big issues for the next few months and years in Trump’s second term?

Andrei Iancu: To quote Yogi Berra, “It's tough to make predictions, especially about the future.” With that caveat, I am definitely encouraged by the IP positions in the new Trump administration. I wrote an article in Fortune magazine in October 2024 touching on many of these points, including the regulation of AI. I predicted back then that a Trump-Vance administration would be a significant improvement when it comes to intellectual property and these new technologies.

Looking at the appointments — there hasn’t been enough time to see what they will actually do — but just looking at the appointments of the individuals, I am very encouraged. I worked with Gail Slater in the first administration. I’m familiar with her positions generally, and what you just mentioned illustrates that this administration, and Gail in particular in antitrust, understands the importance of IP to a growing economy and our technological competitiveness.

The New Madison approach basically says that intellectual property is good for the economy. It’s not an antitrust issue that should be regulated from an antitrust perspective. By and large, all patents, including standard essential patents, should be treated the same way and should be given full enforcement rights. There’s no special provision anywhere in the code or legislature that standard essential patents should be treated any differently.

Back in the first Trump administration, we had a collaboration between the USPTO, the Department of Justice, and NIST to put forward policies like this. We had, for example, the 2019 Standard Essential Patent policy position that basically said what you just articulated — all patents are to be treated equally, and they have all the rights of enforcement, including injunctions, and the laws should apply equally to all patents, including standard essential patents.

The Biden administration came in, and one of the first things they did was take down the 2019 policy, so the United States right now has no policy. I am hopeful that with Gail at DOJ Antitrust and Howard Lutnick coming in at the Department of Commerce, which oversees both PTO and NIST, that we can reinstate those types of policies that are protective of American intellectual property. We’ll see how they shake out in the end, but I am very encouraged by the appointments.

Jordan Schneider: Aside from the general direction that you can reasonably project based on the appointee choices and what they’ve said so far, you have this new energy with AI plus DOGE, which wasn’t necessarily in your time in the Trump administration. How crazy could things get at PTO? What are things you would have never dared to do in your time as a director that the Trump administration 2.0, which is clearly considering pushing the envelope in many different places, could potentially do for better or worse when it comes to the Patent Office?

Andrei Iancu: Let me first say that I don’t think the PTO is first or among the first in sight for the people at DOGE. They have other fish to fry before they get to the PTO. Having said that, it’s really important to understand that the PTO is unlike almost all other government agencies.

The PTO is not quite a regulatory agency that regulates the public or taxes the public or anything else. Unlike most government agencies, the PTO creates rights. The public comes to the office and applies for certain rights — patents and trademarks — and they leave with more rights than when they came in, usually. The PTO issues over 7,000 patents every single week. A very large number of trademarks get registered every week. That is additive to the applicants and additive to the economy. We’re not the taxing authority, we’re not the regulatory authority — we help the economy.

That’s a really important distinction, combined with the fact that the PTO does not operate on taxpayer dollars. PTO operates almost exclusively on user fees, and the PTO examiners are production-based — they have to produce a certain number of units every bi-week. This is very different from most government agencies. That’s why I’m not entirely comfortable discussing action at the PTO within the concepts of DOGE, because I just don’t think it fits within that. It’s entirely possible that the new Trump administration will have some bold actions at the PTO, but I just don’t think it’s going to be within the DOGE framework, which is aimed at, by and large, reducing government and reducing government waste.

Jordan Schneider: Andrei, this sounds like you’re pleading at the pearly gates to be let in.

Andrei Iancu: I am pleading for folks to just understand that the PTO is really a special agency that, by and large, helps the economy. Now, I will say there are lots of improvements that can and should take place at the PTO.

One of the boldest things that I’ve seen people talk about in the context of DOGE is that there should be action with respect to the PTAB. The PTAB is the Patent Trial and Appeals Board. It has two functions — trials and appeals. The appeal side is the bigger side. If you’re a patent applicant and you disagree with the examiner — the examiner just will not give you a patent because somebody else invented first or whatever — then you can appeal to this PTAB. That’s the appeal side.

The trial side works on patents that have already been issued. If a patent has already been issued, somebody else in the public can challenge it and say that it’s invalid, it should have never been issued, and bring it to the PTAB for a trial. The trial side of the PTAB has been very controversial over the last decade. It was created in 2012 by the America Invents Act signed by President Obama.

It’s been really controversial from the moment it began because, by and large, it has the administration, not the independent courts, remove a granted right. That makes people uncomfortable for many different reasons, including the fact that 85% of the patents in these IPR post-grant examinations at the PTAB are already involved in a similar proceeding in court. There are now effectively two proceedings at the same time on the same sort of issues between the same parties — one in the Article III independent courts and the other one at the PTAB.

I have heard folks say, “Hey, DOGE, why don’t you take a look at the PTAB? It’s redundant government waste. You should eliminate the PTAB.” Now, I haven’t seen any serious look at that by DOGE. As I said, I think they’re very busy with other things. But if you’re asking me about one of the wildest things that could happen, I could see that happening at the PTAB. But I still think it’s far-fetched.

Nicholas Welch: Another way the USPTO is unique — the USPTO director is in the executive branch but operates kind of like a quasi-judge. They can review decisions executed by the PTAB judges who aren’t actual judges. In 2021, there was this interesting Supreme Court case, United States v. Arthrex. It was an Appointments Clause case (people should do a lot of Appointments Clause homework because that’s what tanked the Trump Florida documents case). It says that PTAB judges are principal officers, which means if you’re a principal officer instead of an inferior officer, you need to be Senate-confirmed, and they weren’t. As a sufficient remedy to this problem, we can just let the USPTO director review any decision made by the PTAB judges, because a director is a Senate-confirmed officer.

How much leeway or authority does that actually give the director? We now have this director review process for any decision the PTAB makes. The USPTO office is going to be feeling a lot of policy winds from whoever happens to be in the White House. What are the odds that some Chinese company sues a US company in US courts and it goes to the PTAB, and then Trump gets on a phone call with Xi Jinping and says, “Nice company you have there, it would be a shame if Secretary Lutnick told the patent director to not review this” or something like that? The USPTO director is kind of a judge with political valence. Am I understanding the org chart right here? What are the implications of Arthrex going forward?

Andrei Iancu: You’re understanding the org chart well — it’s a very perceptive and good question. Just to back up for a second to set the stage, the PTAB does have judges. You say they’re not real judges, but they are judges. They’re not what we call Article III judges from the Constitution, but they are administrative judges. The administration has judges in various areas. The Social Security Administration has judges to resolve disputes with your Social Security checks, the IRS has its own judges to resolve tax disputes you might have with the government, and so on.

We have these patent administrative judges — we have trademark administrative judges as well in the trademark office. They’re not independent Article III judges. The whole point of the Constitution having three independent and co-equal branches is to create a judicial branch that is independent from politics. Yes, the judges are appointed by the President, confirmed by the Senate once, but that’s it — they’re lifetime appointments. The founders found that to be really important to create independence of the judiciary, to be independent of the political winds. That has worked really well for the United States for the past 200-some years, especially in disputes between private parties over private issues. You want to have this independent judiciary that deals with that so they’re not impacted by politics.

However, the America Invents Act created this new proceeding at the Patent Office, the Inter Partes Review Proceeding (IPR), where you can have private parties fight amongst themselves over private property (a patent) in the executive branch. What the Supreme Court said in the Arthrex case is that because it’s in the administrative branch, the buck has to stop with the political appointees. The administrative branch is politically controlled and it has to be controlled by the people voting in the President and therefore the President’s appointees.

The Supreme Court said that’s what the Constitution demands. You cannot have unaccountable career officials that ultimately issue final decisions from the administration that are not controlled by the politically appointed individuals. Why? Because if that were the case, the public has no control then over the actions of the administration. The people need to vote in or out based on the acts the administration takes. If you don’t have political accountability, you’re missing out on the vote of the public.

The Supreme Court said in Arthrex that the final decisions from the PTAB have to be affected or at least available for review by the politically appointed director. That was in the summer of 2021, at the beginning of the Biden administration when the Supreme Court decision came out. The PTO director in the Biden administration basically said, “That’s what the Supreme Court says, therefore I, the director, will take upon myself as a human being to review these decisions if I am asked to do that."

To be honest, I think it’s very uncomfortable to have a political appointee (who comes and goes and is subject to political pressure) call balls and strikes in private disputes between private parties. This is not a dispute between the taxpayer and the IRS — this is a dispute usually between two corporations fighting over this particular property, a patent.

For me, it’s uncomfortable to have the political appointee making those decisions. I don’t think it’s good for the public in the long run because it’s unpredictable — you don’t know who the next appointee is going to be, and patents are long-term assets. But I understand the Supreme Court point here about the Constitution — since this is an administrative action, you have to have political accountability.

It’s really hard to fit this square peg in the round hole. You’re trying to fit a judicial action — resolving private disputes between private parties over private property — but in a politically controlled administrative agency. It’s very difficult to resolve. This is why this IPR system has been controversial since its institution in 2012. People are by and large uncomfortable with a political agency resolving these disputes. At bottom, in my view, this should be left to the independent Article III courts that the founders created in 1776.

For now, we have the statute, it has to be administered, and we have the Supreme Court decision. I would personally implement that differently. I would try to create separation to the extent constitutionally allowable between the Director and these decisions for many reasons. In the end, I would try to find a way to move as much of this towards the court system as I could, because private disputes between private parties for private property should be handled by the independent courts to the extent possible.

Mood Music

Andrei couldn’t come up with a patent song so we asked Deep Research for some suggestions and it gave us these absolute gems.