Tarun Chhabra is the head of national security policy at Anthropic and previously served as the Deputy Assistant to the President and Coordinator for Technology and National Security on Biden’s NSC.

Today, our conversation covers…

• Why the US needs to maintain an advantage in the race for AI development against China,

• Whether the US’s AI industry is prepared for future competition from China,

• The lawyers vs. engineers debate, and what the US needs to build AI supply chains,

• How government and industry can work together to across the AI development process.

Listen now on your favorite podcast app.

Race for the AI Frontier

Jordan Schneider: Part of the original justification for banning exports of American chips and semiconductor manufacturing equipment was the idea that these pieces of technology could directly contribute to PLA capabilities. We’re recording this during the week of a military parade, and I’d love to hear you give your most convincing unclassified case for why these technologies directly contribute to arming a strategic adversary.

Tarun Chhabra: Thanks again for having me, Jordan. The Wall Street Journal just published a good piece this week about how the PLA is using commercial AI technology. But this really goes back to when Jack Ma 马云 disappeared — that was essentially the end of any dissent or pushback from China’s tech companies regarding support for the national security apparatus.

It’s been a safe assumption since then that dual-use technologies enabling capabilities for the PLA or China’s intelligence services will be used that way. You can ask your favorite LLM for examples, and you’ll find plenty. One that obviously comes to mind — particularly because it was a focal point of the outbound investment restrictions — is cyber capabilities.

A straightforward example is Chinese hyperscalers providing the equivalent of cloud services and cybersecurity services to national security actors in China. Obviously, they’re going to offer AI services as well. Unless you think the AI somehow won’t be part of that package — which I don’t know how you’d conclude — it’s a pretty straight line from cloud and cyber services to frontier AI models.

Jordan Schneider: We have different layers of an AI race. I’m curious about your topology and how you’d rank them on their ability to build defensible advantages.

Tarun Chhabra: I think about it as the race to the frontier, the race for diffusion globally, and then the race for adoption — which encompasses both national security and economic applications.

On the race to the frontier, it’s about power, talent, and chips.

This is partially why Anthropic focuses so much on addressing power and permitting barriers to build more AI infrastructure in America, and why we emphasize export controls — because we think hardware is our advantage for the next several years.

On the diffusion side, it’s also why we focused on export controls. We don’t think our hardware should be powering Chinese data centers, either to help them reach the frontier or to compete with U.S. companies or other trusted companies globally. The same principle applies to adoption. The more we succeed in adoption, the more compute you’ll need at the enterprise level for national security actors as well.

Jordan Schneider: When debating how to structure export controls, let’s start on the chip-making side. What are the variables you tried to optimize for, and what should current policymakers focus on moving into 2025?

Tarun Chhabra: We have a significant advantage when it comes to chip making. The U.S., together with allies — obviously Taiwan, and stemming from our dominance in the supply chain along with close allies like the Netherlands and Japan — holds this position.

The question is, how long can we ensure that we maintain that advantage? This comes down to our ability to control that technology during the period in which China has yet to indigenize it, certainly at the level that enables scale in production or very advanced production. It also entails working on components and servicing as well.

Irrespective of who’s in office, here are the next things we ought to be doing: We should do more on the component side — this is also in the interest of the tool-making companies to defend their advantages. The servicing piece is really important as well in the industry. These are the next steps we should take to defend our advantage in chipmaking.

Jordan Schneider: How much do you buy the argument about China’s will to indigenize? The will and capability to indigenize remains a live subject of debate across all the different layers of our AI future.

Tarun Chhabra: The shot was fired during the first Trump administration with actions against ZTE and Huawei. From that point forward, maintaining dependence on the United States for advanced dual-use technologies was a bad bet from their perspective.

But there’s also a broader historical story here, which you’ve probably discussed with Dan Rosen and others — the history of China’s industrial strategy over the last several decades. Name any sector where it has worked for us to say we’ll just keep them addicted to the technology.

The pattern in China’s playbook is pretty clear — buy it until you can make it. Once you can make it, kick out the U.S. competitor. Eventually, once you can make it at scale and subsidize it, try to eat up their global market share. It’s happened over and over again. You name the sector, and given the leadership’s focus on AI, there’s next to no likelihood that we can stop the indigenization train.

The real question — which I know you’ve talked to Lennart Heim and Chris Miller about — is what do we do in the interim? The other question I would ask is: where would China be today if we didn’t have the controls? We can cite various developments, but where would they be if they had the talent, the energy, and also had the chips? We’d be in a much tighter race today, and we don’t want to be there from a national security perspective.

Jordan Schneider: Why is that hypothetical such a confusing thing? The SMIC base case is that they would be further along than even Intel is today if not for controls. Is this a particularly hard hypothetical?

Tarun Chhabra: You and I tend to be in violent agreement on this point, which makes it hard for me to understand why that isn’t the question we should ask. That’s why someone like Dario Amodei says DeepSeek shows the importance of maintaining, if not strengthening, controls. They have incredibly talented AI engineers in China, power, and capital — it’s just about the hardware, and it’s for this window.

The other layer of controversy is, how much does this window matter? Our perspective is, we’re not seeing a significant slowdown in the saturation of benchmarks. We still think you’ll see transformative capabilities over the next three years.

We should all discount this in a healthy way, but if we believe it’s just over — more likely than not, or 60% or 70% — and you talk to folks about how they assess that, the implications are significant from a grand strategy perspective. We ought to be preparing for that, and from my perspective, we shouldn’t try to make this a close race.

Jordan Schneider: The other wrinkle that folks haven’t necessarily priced in is that you guys are training on Trainium, and Google’s figured out how to train on TPUs. The idea that CUDA is this “one layer to rule them all” — that if you can get them stuck on it, you’ll have control forever — we’ve seen multiple companies already figure it out, and they only had commercial motivations, not their head of state telling them they had to do it.

Tarun Chhabra: I was just having this conversation with some of our colleagues internally this week, and they’re making the same point. Yes, Anthropic is a well-funded company with really talented engineers working on the hardware problem, but if we’re able to do exactly what you just said, and a nation-state is committed to doing it, then they can probably get there.

Jordan Schneider: You mentioned increased capabilities — this is weird sci-fi stuff. I’m curious, looking backward, how much that ethos or that single-digit, low double-digit probability played into policymaking. What’s your perspective going forward on how folks should adapt for the possibility of those types of futures?

Tarun Chhabra: What we’re seeing right now with our coding model — our engineers are using it for about 90% of tasks. That’s going to take longer to diffuse across the economy or even the broader tech sector, but people doing national security work often need to do a lot of coding, especially for cyber operations. The applications in cyberspace are pretty significant.

We have a demo showing what it would cost to replicate the Equifax breach — you could probably do it for well under fifty cents of tokens. If you tried to replicate that globally, you could probably do it for under $10,000. That alone should ring the alarm bell, and that’s with current capabilities.

If we think about nation-states trying to make cyber operations more autonomous in their attacks against us, and the need to defend against them and have a viable policy in cyber defense alone, that’s a clear and present problem today.

Jordan Schneider: In trying to talk people into this worldview, what typology of skepticism do you run into nowadays in Washington?

Tarun Chhabra: That’s a good question. Some skepticism is pegged to “this model came out and it wasn’t everything I expected it to be” — whatever model that might be. Using a data point of one isn’t a great way to assess this necessarily.

Another skepticism is that adoption is slower than the most optimistic projections suggested for certain uses, like coding. Then there’s the view that it’s taking longer to penetrate the physical world in manufacturing than very optimistic projections thought it might.

But going back to the counterfactual, we ought to re-baseline the questions. If I had told you three years ago that we would have coding models that could do 90% of our software engineers’ development work today, or that we could have a significant impact on cybersecurity, you might have believed me, but many people would have been understandably quite skeptical.

When the chip controls went into place in 2022, although a big focus was LM development and views on where that was going, the easiest thing for people to understand at the time was that the chips themselves are used in computers that do nuclear modeling or design weapons systems. That’s true, but not at the scale that would really be impacted by export controls. This has been a perpetual issue — how do you get people to think a year or two ahead when you’re on this exponential curve?

Jordan Schneider: You used to be a speechwriter. The Jake Sullivan line that will ring out in national security textbooks for years to come: “Given the foundational nature of certain technologies such as advanced logic and memory chips, we must maintain as large of a lead as possible.”

As you said, 1%, half a percent, maybe 2% of these chips are probably going directly toward nuclear modeling or similar uses. The ongoing tension is that the vast majority will be for commercial use. You can have different opinions about whether the U.S. should be supporting broad-based growth of China, but this tension is built into anytime the government gets involved with technology — these aren’t night vision goggles.

Tarun Chhabra: The key issue here goes back to the question you asked at the beginning: we know this dual-use technology will end up supporting national security capabilities for a country that is actively planning military operations against the United States.

You have to accept that there’ll be some collateral impact in some cases to address that problem. Then you adjust based on what kind of advantage you think these capabilities are going to provide. If you think they could be transformative, then you take more risk on that front.

Jordan Schneider: There’s that level, and also the strategic level of what you’re doing to the relationship with your enemies and with your allies by controlling this technology. What’s the right way to conceptualize how the U.S. should be relating with the world — excluding China, Russia, North Korea — when it comes to AI?

Tarun Chhabra: We want to build as large of an ecosystem as possible that’s trusted and where U.S. AI and U.S. technologies are prevalent and even dominant. That’s the world we’re trying to build.

The question is: how do you do that? This is a decision not just between enterprises, but also one that governments will take. We often talk about the United States and fellow democracies working together on these issues. This is important not just from the standpoint of opening up markets — it’s also really important for our intelligence relationships with key allies. We want to make sure we continue to be interoperable across many layers of our relationships, both national security and economic.

But I want to come back to the point about “as large of a lead as possible.” Understanding the historical context is important because if you go back to the days of CoCom, the idea was not that we would give the Soviets an “n minus 2” advantage. That concept basically came after the collapse of the Soviet Union, when we did not have an arch geopolitical foe plotting to fight a war against us.

Jordan Schneider: We were selling arms to China as of 1995.

Tarun Chhabra: Exactly. That was the context in which the “n minus 2” concept became more popular. If you believe we are in a strategic competition with China, if you see that they are planning to fight us and target our troops and critical infrastructure, then you have to revisit that concept.

Software, Hardware, or Both?

Jordan Schneider: The idea of AI contributing to a new revolution in military affairs — we have Andrew Marshall saying: “The most important goal is to be the best in the intellectual task of finding the most appropriate innovations in the concepts of operation, making organizational changes to fully exploit the technologies already available and those that will be available in the next decade or so.” It’s not just having the models, it’s figuring out how to use them. This is a thing you guys are doing now. Where are we on this? Why don’t you respond to that quote?

Tarun Chhabra: This is where I’d actually give real credit to the current administration because they’re really focused on AI adoption, certainly across government, but also in the national security space. You see that with contracts from the Defense Department. Anthropic has one, OpenAI has one. Google and others have these as well. They’re laser-focused on accelerating adoption.

It can’t just be a question of “let’s use the chatbot” or “let’s bring the model in.” It needs to be, how can we use the models to re-engineer some of our mission space? That’s what Marshall was talking about. That’s really the much harder task that people are rightly focused on right now in the administration. That’s what we want to do together as well and have people on our team who are focused on partnering with the national security agencies to do exactly that. I see that as core to our mission. If we say we’re focused on helping support democracies, protecting national security for the United States and allies is foremost in that as well.

Jordan Schneider: The divide historically has been the ideation that Andrew Marshall was involved in. The doctrinal innovation that Andrew Marshall is talking about happens in-house basically. Then the military has toys, and they figure out how to play with them in different ways. Over the past 10 years or so, as you’ve seen on many different dimensions, commercial technology has leaped past what the department and the services are comfortable with or have an understanding of.

We have more of defense as a service, and you see more players trying to sell into the government, not only with products that are required, but with products that fit into their vision of a doctrinal future that they try to sell into Congress and the Pentagon. This makes sense at one level, maybe a little scary at another. I’m curious — how forward-leaning should you be? What is the right posture for someone coming in with new capabilities to bring into these organizations?

Tarun Chhabra: The responsibility of companies that are developing new capabilities is to ensure that policymakers and the military and the intelligence community have insight into where we think the technology is going, and certainly insight into where we see early adopters in the private sector taking the technology so that they can try to get on top of it as soon as possible and figure out how to employ it in doctrine.

That’s something we definitely can do. But obviously, the doctrine needs to come from the government. The planning needs to come from the government. There are lots of ways where we can have a really productive exchange, pressure-test some ways of doing things at the invitation of the government, of course, and say, “Hey, you could do it this way.” But look, that’s not a foreign concept. We’ve done that with a lot of other technologies, too.

Sometimes we take the idea too far that AI has come out of the totally private sector, with no government involvement. A lot of technologies that have been really important for national security — they may have been funded, yes, there may have been research funding that helped get them going. But there has been a lot of adoption of civilian technologies, as you know, and then they’re brought in and there’s a give and take between the civilian sector, the private sector and the military about how to adopt them.

Jordan Schneider: We see a bit of the future of warfare in what Israel is able to do in Iran and what is happening in Ukraine. But these are trend lines that you can trace and track over time, and you see things changing. When folks imagine the world of fast takeoff, you can paint futures in which there are very radical discontinuities in what militaries can do. I’m curious about your perspective on this.

Tarun Chhabra: You will see some discoveries that come out of leading corporations or the research community that are using the compute and using the models, particularly when they have access to certain kinds of data, specialized data. But where we want to go for the government — I hope we can empower them — is to be a force in their own right in some of these discoveries.

When we get to capabilities like that, an example that’s probably fairly straightforward today is the Department of Energy national labs. The labs have developed over decades a corpus of really incredible scientific data, in some cases experimental data. The question is, is there a way for them to use that data to potentially build a new platform for scientific discovery? As you know, the labs often are important partners for the national security community as well. That’s one example where today you could already see if we put the pieces together, there might be real capability to take advantage of transformative capabilities in the near future.

Jordan Schneider: I feel like it would be hard to imagine today someone really pulling a rabbit out of a hat — pulling the equivalent of a nuclear weapon out of a hat — given the current technological paradigm we have on September 4, 2025. If things get faster, that may change, right? Or could it not change because everyone’s going to be feeling this at the same time?

Tarun Chhabra: Well, that particular example — we actually just did some work that we announced a couple of weeks ago, where we worked with the NNSA on classifiers because we actually want to make sure that people can’t do what you just said out in the wild. That’s something we’re working through with the Frontier Model Forum. We hope that other frontier labs are going to adopt similar safeguards as well.

But there’s one way to answer your question: are we prepared in the physical world for what capabilities may be coming online? That is honestly one of the things that worries me most, which is a topic you’ve talked to many of your other guests about. If you do the net assessment of where we are in our defense industrial base — U.S. versus China today — in our broader manufacturing base, are we doing enough to be poised to take advantage of some of these capabilities down the line?

That’s another responsibility that we have as frontier labs to help ensure that we will be poised to do that. Some of our best partners in that space are going to be some of your recent guests — people who run defense firms that are AI-centric and are already using frontier models and thinking about how they’re going to be able to use that to scale production. But that is a space where we actually need more people thinking two years ahead about what happens if we reach this capability but we have the status quo in our defense industrial base.

Jordan Schneider: This is really fun doing two shows in the same day that echo each other because I get to ask you the same exact questions and see what your different answers are. Thank you for that provocation from Dan Wang: “I can’t distract us from broader American deficiencies. If the U.S. and China were ever to come to blows, they would be entering a conflagration with different strengths. Would you rather have software or hardware?”

Tarun Chhabra: We need both. The answer is not to accept the status quo. From my perspective, the answer is to prepare to take advantage of the capabilities that are going to come online. That means much more work with the physical world.

Jordan Schneider: AGI and nihilism can run in a lot of different directions. One thing I just alluded to — the idea that you can spawn Dr. Manhattan from your data centers and then just stride the globe and do whatever you want. Something that also comes out of that is, yeah, you don’t have to do the hard work of building munitions capacity, and you also don’t have to do the hard work of dealing with annoying allies, because whatever, you’ll have God on your side in maybe not two to three years, but whatever, you’ll extend it out a little bit. I wouldn’t say Dario maybe doesn’t buy into all of that, but you can squint at some of his writing and see some echoes there. What are the futures that people should really consider and what are the fallacies of AI solving all of your problems that folks shouldn’t fall into?

Tarun Chhabra: Look, I don’t think nanobots are going to save the world next year, but some of this comes from a view that it’s hard to know what these capabilities are going to yield in a relatively short period of time — by which we may mean a couple of years. It’s hard to know what advances they may give us in advanced material development or in manufacturing processes.

There’s actually, maybe counterintuitively, a dose of humility in saying it’s pretty hard to say that we ought to just build more of the status quo infrastructure when we may be on the cusp of some of those capabilities. It’s actually going to be a hard thing to manage — how do we build an infrastructure that might look really different, or could look really different, with the capabilities that are coming online in a couple of years?

Jordan Schneider: Just tech as a component of national power and how it ranks and fits into the other components of that if you’re looking at great power competition.

Tarun Chhabra: Well, I’m a little biased probably right now, but I’ve been of the view that frontier AI and biotech, and particularly the convergence of the two, are going to be very powerful tools, and they’re going to be potential vulnerabilities if we fall behind or don’t make certain investments.

But look, it is a physical and real reality today that if your adversary thinks you’re going to run out of munitions in a couple weeks into a conflict, you’re going to have a hard time doing serious deterrence. We have to live in the current paradigm and we have to ensure we’re strengthening, bolstering deterrence while we prepare for a totally new paradigm that’s really still hard to actually piece together in the mind’s eye.

Jordan Schneider: There’s a Williamson Murray quote that I won’t read in full because people will hear it in the prior interview. But basically the idea that revolutions in military affairs happen at the tactical and operational level and oftentimes strategic decisions — smart or poor — wash out whatever cool stuff you come up with with your blitzkrieg or deep battle or what have you. As we talk about all these things, does it even matter if our treaty allies go a different direction or India decides that they actually really do want to be super friends with China?

Tarun Chhabra: It matters a lot from so many vantages. But to your question about the tactical level, this is why it’s a really good idea that the army has Detachment 201, because that will help people start to use the technology and think about the technology and how to operate with the technology at the tactical level and not just how we drop it into ConOps at a super high level. We have to do both at the same time. That’s really hard to do.

But frankly, it’s a reality that most people in the business world are dealing with right now. Every day, you have CTOs who are saying you will adopt this technology. You’ll tell me how you’re re-engineering your business processes. At the same time you’ve got to use the stuff today with your current process. That’s just how everyone is doing it right now.

The national security community, in that sense, is not distinct. We often will bring together senior national security policymakers with the frontier labs, which is good to do, but in some ways their peers are more so the C-suites of major companies that are trying to adopt really quickly and are in a competitive atmosphere trying to do so. The worry about it is that competition is very real day to day in the market for a lot of companies. When it comes to militaries and intelligence, it may sometimes be harder to see until you have some sort of strategic surprise, which we want to avoid.

Lawyers vs. Engineers

Jordan Schneider: Okay, what we do here is think about the next 20 years of U.S.-China relations and try to think of a net thing. Yeah, it’d be nice to have better frontier AI. It’d be nice to have AI adoption. But would I trade that for Japan maybe? Probably not. You know, on the hierarchy of things you would want America to get right over the next 20 years, my sense is that screwing up the relations with the other most important developed and developing countries is something that has a higher probability to be a bigger deal for the rest of the 21st century than export control policy or whose model passes which benchmark first.

Tarun Chhabra: Look, there is a way to do all of these things together.

There’s a way to try to maintain strong bonds with your allies and also try to maintain AI leadership within the alliance.

That is particularly where you have allies who are actually bought into the strategic threat posed by China. A lot of them are, but they’re also facing countervailing economic interests. In some cases they’re getting pressure, coercion from China as well.

It requires really active efforts, really active diplomacy to keep those bonds strong. I hope that’s something that we can continue to do. If you say that you really want to ensure that it’s American AI that’s used around the world, you want to start with your allies. You want to make sure they trust that AI, they believe they’re invested overall in the stack, feel like they’re a part of the supply chain where that makes sense because ultimately these are high stakes, high dollar, big corporations involved. There’s a political calculus too for a lot of this, just given the stakes.

Jordan Schneider: You spent a lot of time doing tech diplomacy. What takeaways do you have from that experience? What works, what doesn’t, how to do it right?

Tarun Chhabra: Connecting the dots in an allied government is not always straightforward. It’s not always straightforward in the US government either. Leadership from the White House really matters in that regard — having your Commerce Ministry, your defense and intelligence interests, and your diplomatic interests all come together to make decisions that can be really hard for allies. When it comes to market share worries about coercion from China, there’s really no substitute for that kind of coordination. This isn’t to say there isn’t an important role for many actors in government, but without the White House being involved and without their counterparts being involved in head-of-state offices, it can be really hard to get things done.

The other key factor is that being a first mover really matters sometimes. Seeing the seriousness of purpose that presidential decisions bring into relief shows allies that the US is serious about certain decisions and that we need allied support to make things happen. I found that pretty consistently over and over again.

Jordan Schneider: Any countries you want to shout out for being good at this?

Tarun Chhabra: Some of our key allies are actually building new muscle in this space. There were governments that consulted with the previous administration about building new offices in their head-of-state offices to coordinate technology policy because they shared the view that there needed to be some head-of-state coordination mechanism. Frankly, I saw most of our top allies building that new muscle — whether Japan and Korea, who are calling it economic security (and it does go beyond technology to include dual-use technologies), India, or Australia. Everyone’s doing it differently, but everyone was basically building that muscle between technology competitiveness and economic security. That’s a really good thing — having some coordination function to figure out where we can make our interests align.

Jordan Schneider: Let’s talk about more ligaments. There’s an analysis piece, there’s a future-casting piece — doing chip controls pre-ChatGPT took some foresight. Then there’s an execution piece where once you do this thing, you need the people to run with it and make what you say a reality. Do we still need new offices? Is it just talent? What are the building blocks that the US government should invest in?

Tarun Chhabra: You definitely need the talent, and that’s really hard. Finding people who have the technical depth and can also operate in the policy world is not easy. I was lucky to have an incredibly talented team working on these issues. But I really think the role matters. Having folks whose job it is to wake up every day and think about the technology war that China is fighting against us, and whose job is to try to make America much more competitive in key technologies, really matters. Without that, there are countervailing interests — trade interests, bilateral relationship interests that encompass much more. If that interest isn’t at the table, it has a real impact.

Jordan Schneider: You were a lawyer, used to wrangling with lawyers, and now you’re working with lots of engineers. Continuing on Dan Wang’s view of the world — America as now a lawyerly society in contrast to China’s more engineering-focused approach — people were frustrated, myself included, at the pace of the rollout of a lot of these controls. It felt like there were lawyers or maybe other things getting in the way. Did you buy this? Is this what ails America? Too many laws or too many lawyers?

Tarun Chhabra: We have lawyers, too. But look, we work with some of the most amazing engineers anywhere in the world today who are building amazing technologies. Even outside the company, there are amazing engineers who want to use the technology and are developing stuff that we wouldn’t have imagined. That’s the magic. I’m not sure that I buy the full typology — I love Dan, but we have a healthy argument about this sometimes.

The question about the pace of controls or debates we have over technology policy, especially when it comes to restrictions, is a give-and-take about industry interests, national security interests, and trying to strike the right balance. We built an architecture to entertain that debate. The problem becomes when you’re missing a piece of that set of interests at the table. If you ask the person who was leading technology and national security policy whether we could have done more, faster, of course the answer is yes. But the key thing was bringing the questions to the table, bringing the proposals to the table, having a strategy to maintain our technology leadership — when for a long time, China was fighting the war against us and we weren’t fighting back. Credit to Matt Pottinger for getting that going, especially in the 2018-2019 timeframe with some of the big actions they took then.

One thing I’m proudest of is that we built a really strong bipartisan consensus for a lot of the action. Some of that was very apparent in statute — there was the CHIPS bill, there was action on TikTok. But there was also broad support for executive actions, whether that was export controls to maintain AI leadership, outbound investment restrictions, data security restrictions, or ICTS actions on vehicles coming from China because of the cybersecurity risk posed there. That’s something that is sometimes underappreciated.

Jordan Schneider: That’s lawyer energy building bipartisan consensus, not engineering energy.

Tarun Chhabra: Look, if you think you’re in a multi-decade contest with China over technology, there’s no other way to do this. There’s got to be a bipartisan consensus that can transcend administrations. I was very privileged to inherit work that Matt and his team had done at the NSC with others in the first Trump administration. We tried to build on that and broaden the consensus more and fight back on a much broader range of sectors.

Jordan Schneider: Another piece of tech diplomacy is directly with China. The shoe that didn’t drop until 2025 was rare earth controls. I’m curious about any reflections you have on being able to ramp up what you did without having the type of response that we’ve seen from China over the past few months.

Tarun Chhabra: There was a very concerted strategy to be pretty clear internally on the actions that we needed to take for US tech leadership, while also maintaining a diplomatic channel and explaining really clearly why we were doing it and why we were doing it when we did it. That strategy was designed to ensure we could take all the steps we needed while trying to mitigate that blowback. I can’t speak to how the administration thought about where we are today, but the situation China wants us to be in is one where they get our most advanced technologies in exchange for commodities. That was definitely at play when it came to efforts to coerce our allies. As you know, there were restrictions put in place on gallium and germanium against US allies in retaliation for some of the tech controls already, but not at the level that was imposed later.

Jordan Schneider: Before this, you gave us a list of some of your greatest hits — outbound investment, ICTS. Pick your favorite. Which one doesn’t get the love it deserves?

Tarun Chhabra: One thing I worry about not getting much limelight today is biotech policy. There’s lots of attention on AI policy, rightly so. The National Security Commission on Biotechnology did a great report. [ChinaTalk has some thoughts on that…] There were bipartisan members of Congress who were on that commission. They made some pretty astute observations about where China was going and where we will be in a net assessment if we stay on the current course — the kinds of dependencies we’ll be in on China and the long-term economic impacts of doing that. I would highly recommend that report and hope it gets much more attention. I hope that we find a way to invest in our R&D architecture while China is increasing theirs very, very quickly.

Jordan Schneider: That was my takeaway. Spending a lot of time thinking about this and reading that report deeply, the levers are not nearly as straightforward and sexy as the ones that the government is able to pull when it comes to manufacturing equipment and AI chips. It’s not fun stuff like FDA reform and investment into universities, as opposed to “here’s this machine where if we take it out the whole edifice crumbles.” It’s harder to have something be salient when the upside is more drugs for people.

The disturbing part of that report for me was their italicized vision of the future where China cures cancer, but we’re not allowed to get it or we’re charged exorbitantly for China’s cancer cure. It’s like, “Well, but we cured cancer,” right? The dual-use downsides of breaking this ecosystem, which right now seems to be much more — I don’t know, I’m less concerned with the potential futures that I see for the AI one and I’m also less convinced that doing things which would take the potential of one-sixth of the world’s scientists offline or reduce their productivity would be a loser for America and society at large. But I do buy the argument that the process knowledge involved with coming up with and scaling new drugs is not something you want to completely outsource to anywhere.

Tarun Chhabra: That’s the worry. You put your finger on it — the status quo is not necessarily sustainable because the status quo is trending toward greater and greater dependency, both on the manufacturing side, but also on the drug development side and on the clinical testing side. The government did impose restrictions in January on some advanced, high-throughput, high-fidelity biotech equipment. But you’re right that that’s one piece of a much bigger puzzle in the case of biotech that is going to require a lot more streamlining and regulatory reform because if you’re doing biomanufacturing, you don’t know what agency to go to in many cases depending on what your product is. That’s an area where I hope there’s much more focus and it doesn’t happen because of some big surprise.

Jordan Schneider: Did I send you the Quad Monkeys pitch?

Tarun Chhabra: No, I don’t think I did.

Jordan Schneider: Oh wow, I did a bad thing for America. But yeah — India. We can’t get Chinese monkeys anymore. India has all the monkeys, but there are regulatory reasons. There’s some lobbyist who’s been really trying on this for a long time now. It deserves its own podcast.

The AI companion stuff. You want to talk about that? It’s not about mass precision; it’s about mass intimacy, Tarun. This is the future of warfare. We’re developing closer and closer emotional bonds with our AI chatbots. If you thought stealing an SF-86 was bad, the amount you can learn about someone or directly influence them by seeing their chatbot logs, controlling their chatbot companion, which doubles as your best friend, therapist, spouse — this seems to me like the revolution in not even military affairs, but just social affairs, which nation states can very much play a big role in exploiting.

I’m worried about this. Thirty percent of American AI companions are headquartered out of China right now. Should I be? Am I crazy? It seems wild that this is something that we’re okay with.

Tarun Chhabra: I mean, you’re right to be worried about it. It’s an extension of the concern we have about China’s ability to manipulate information or information space.

Jordan Schneider: If you thought TikTok was bad, those are still videos, right? This is not your friend. You still have to pay off the influencer to say nice things about X party versus having them in your AR glasses, seeing the entire world you interpret.

Tarun Chhabra: This is one where you can already see what the future could look like if we don’t make certain decisions. In the same way that if you think about what is the state of today’s cybersecurity, and if we do nothing and we have the world of IoT descend on us, and you’re getting daily software updates from China — these are worlds that are coming very soon, and for some reason, it’s really hard sometimes to get your head around it. But that’s absolutely a concern.

Part of it goes to the point that we were discussing earlier, which is when you have companies headquartered in China that are able to use frontier American AI for certain applications. What will they do with it? Will they be used in the ways that you’re describing, let alone much more direct applications for national security? That’s something that, as a company, we’ve now taken action to address.

Jordan Schneider: This is your thing now — having companies make less money because you tell them it’s the right thing to do. What’s the rationale behind this policy change?

Tarun Chhabra: This is very much a leadership decision. You’ve talked to Dario directly about how he sees US-China competition on AI, so you’ll find this wholly consistent with what you’ve discussed with him before. We’ve long had a policy stating that China is not a supported region for selling our frontier AI. But over time, we’ve seen many Chinese companies headquartering in third countries and from there getting access to all these services.

The concern is whether this access aligns with the spirit of our policy of not supporting China as a supported region. How will that access benefit applications that could be used for national security by Chinese actors? When it comes to the competition for the AI stack globally, will it enable Chinese applications — building on our models — to then compete against American companies around the world? There’s also a really thorny technical challenge of detecting distillation, which becomes even harder when you have high-volume throughput happening.

For all these reasons, we think it’s more consistent with our position on export controls and national security to simply not provide our services to those entities.

Jordan Schneider: This speaks to a broader question of where you want to be able to control the stack. If you’re going to split it off and let China build on it, are we selling chips into China that China can then use to build models and companies? Are we selling chips into Malaysia that China can build models and companies on? There’s been an active debate for years about where on that stack it’s okay to let China play. You have these massive Oracle contracts with ByteDance, which exemplify one answer to that question. What’s the right framework for thinking through this?

Tarun Chhabra: The administration is right to focus on US AI dominance. What does that look like? What does the stack look like?

To me, it should be American models using American chips and AI data centers powering US applications, together with our closest allies.

That’s what we want to see. The debates you’re seeing now are about US chips fueling Chinese data centers. The change we’re making is about US models fueling AI applications in China that could ultimately undermine US national security.

Jordan Schneider: If you don’t like where things are headed, where is it easiest to change course two years from now? Pulling the models out from under folks — models seem pretty easy to fast-follow and steal. But are customers sticky? Are data centers sticky? Is the way you train things sticky? These are all open questions. It’s not super straightforward.

Tarun Chhabra: Yes, in some ways they’re open questions, but we also have to factor in that the Chinese Communist Party has a very strong view about what they want to see — a full Chinese stack. They’ll take the chips while they can, of course. The question is: what are we going to do until China gets to that phase? If we believe that really significant, even transformative capabilities are coming online, should we not take more risks now to enable the US to really have AI dominance?

Tarun Lore and Advice

Jordan Schneider: You’ve had nine months now. Are you reading anything fun? Taking any trips? Give the folks some recommendations.

Tarun Chhabra: Yes, I’ve taken some good trips. I’m originally from Shreveport, Louisiana, so I’ve seen a lot more of my parents, which has been great. I’m out to San Francisco pretty frequently as well — India, Australia. Some good trips and seeing former colleagues as well.

The book I’m reading now is Joseph Torigian’s book, which is great. When he was still a pre-doc, Rush Doshi and I brought him into a project we were doing at Brookings while the book was still a dissertation. It’s really cool to see his book out, and I highly recommend it. The way he blends the official party discourse with personal stories is really powerful.

Jordan Schneider: We end every episode with a song. You got one that captures our AI future? The true essence of export controls?

Tarun Chhabra: True essence of export controls... I’m a big country music fan, having grown up in Louisiana. I’ve recently discovered Steven Wilson Jr. Maybe we could sign out with some of his music.

Jordan Schneider: We need a little bit of Tarun lore. The Shreveport to AI policy pipeline is not the most robust. What do you want to tell the kids to live their policy dreams?

Tarun Chhabra: I’ve been incredibly lucky to have really great mentors. I still remember — I spent a year in Moscow after college, and one of my college advisors happened to be traveling there. This was actually Chip Blacker. If you ever met him and had dinner with him, he was carrying on about what my life would look like in 20 years. I was asking, “What do you mean, Chip? How do you know that?” He said, “No, no, you’ll do this and you’ll do this, and then we’ll talk.”

I’d never had anyone express confidence in where I might go in my life. I grew up in Louisiana, and my parents are immigrants. They provided a privileged upbringing, but they didn’t really go to college. Having someone just say, “No, I take it for granted that you’ll be able to do interesting things in the world” — that still sticks with me.

Jordan Schneider: Anyone listening to China Talk, I have absolute confidence you’ll be able to do interesting things as well. A tiny bit more lore. From Shreveport to wanting to go to Moscow in the first place — give us a little more color here.

Tarun Chhabra: I was a Cold War geek growing up. I was very interested in Cold War history. The fact that Hoover was at Stanford was a huge draw for me. I was particularly interested in post-communist societies. I did a summer abroad in Cuba after I graduated from high school. It was with Wake Forest — I think it was the second year American students were allowed in.

Jordan Schneider: This is very early days, right?

Tarun Chhabra: The Pope had just visited for the first time. That was my interest in Russia as well — what was going on in 2000, 2002, and 2003. It was such a different time.

Jordan Schneider: Why do you have a day job? I want you to take a year off. Give us the memoir, man. We’ve got a lot of good stuff here.

Tarun Chhabra: I saw Boris Yeltsin drunk at tennis matches in Moscow. It was really something.

Jordan Schneider: Could he play, or was he watching tennis?

Tarun Chhabra: He was watching the final.

Jordan Schneider: Final question — give China Talk some homework. What’s the more ambitious version of what we’re doing?

Tarun Chhabra: The question you were asking earlier about what these futures look like — in a way that’s unafraid and brings together people who know a sector deeply but who can talk to people who see what’s coming in AI — is really important from a strategic perspective. We try to do it, and I also need your recommendations for who’s doing that really well. That’s some of the most important work we could be doing right now.

Mood Music (Tarun’s suggestion):

中国当下有很多让人头疼的问题，但最根本的问题是什么？我们借用一位影响了几代中国人的老先生的话：“中国今天的问题是很多人没有看到世界”。

说起对当代中国人影响最大的几个人物，每个中国人都能随口说出几个来，但有一位，大部分中国人说不上来。但中国人从小到大，哪怕只念过小学一年级，都离不开他的发明。这个人就是周有光。他发明了汉语拼音，被称为“汉语拼音之父”。可以毫不夸张地说，中国人的生活每天都离不开汉语拼音，都离不开周有光先生的发明。

今天，我们不是讲汉语拼音，而是讲周有光老先生离世前，留给中国人的一句话。周先生1906年出生，2017年去世，活了111岁；他一生跨越了满清帝国、中华民国、红色中国，曾经留学日本，在经济学、语言学上都有很深的造诣。他饱经沧桑，经历过军阀混战、毛氏文革、邓氏改革开放，晚年又见证了土皇帝开倒车。2013年，他已经是107岁高龄。在一个访谈中，他嘱咐中国年轻人，“要从世界看国家，不要从国家看世界”。

他说：“你只要看看世界，只要把眼光放大，眼光一放大，许多问题就不成其为问题了，中国今天的问题是很多人没有看到世界。”这是老先生的原话，可以说是他一生的治学心得。说到自己，他很谦虚，但谦虚中透着智慧。他说：“我没有什么旁的本事，稍微有点小小创造，就是因为我看到了世界，拿国际的知识来补充中国的知识，就很容易得到成果”。

老先生说这话的那一年，中国的土皇帝刚上台，还没有把舆论禁锢得密不透风，一潭死水，国内主要媒体，包括官方的《中国青年报》、《人民网》，对周先生这个谈话，都做了报导。我看到之后，深有感触，借题发挥，总结出一幅对联：“从中国看世界，坐井观天；从世界看中国，海阔天空”。

我本人智力平平，没有智慧，但多少有点学习能力，喜欢向有智慧的人学习，向比我有见识的人学习。学到好的、有价值的，自己有点心得，偶尔也做做知识搬运工。以前是用文字搬运，现在与时俱进，除了文字，也用音频和视频搬运。这期节目，就把周有光先生的智慧，搬运到油管上来，跟大家分享。

Read more

This essay first appeared with DARC.

In 1953, the CIA overthrew Iran’s government with suitcases of cash and a handful of operatives. In the 2010s, ISIS could recruit and deploy terrorists entirely through online interaction. But by 2030, the most effective intelligence operations won’t rely on either playbook—they’ll be conducted through AI companions that billions will trust with their deepest secrets.

The espionage landscape is undergoing its most fundamental transformation since the Cold War. Biometric surveillance and digital tracking have made traditional human intelligence operations increasingly perilous—a case officer can’t simply meet an asset in a park when facial recognition cameras blanket every street. Simultaneously, declining trust in media institutions has undermined conventional information warfare, as populations grow inured to traditional propaganda.

But a new vector is emerging that bypasses both problems entirely. AI companions soon will know us more intimately than any human confidant—seeing through our smart glasses, remembering every conversation, and offering always-on perfectly calibrated emotional support.

This creates an extraordinary intelligence opportunity. A foreign adversary with access to a population’s AI companions doesn’t need to recruit individual spies or craft convincing propaganda. They gain direct, continuous, and intimate access to millions of targets simultaneously. The same technology that helps you draft emails and talks you through your divorce can identify who has access to classified programs, who’s bitter about a missed promotion, and exactly what words would convince them to betray their country.

The thesis is simple but stark: AI companions will become the most important intelligence battleground of the 21st century. The nations that dominate this technology—both in deploying it abroad and defending against its excesses at home—will possess intelligence advantages not seen since Enigma was cracked. The United States must act immediately to ensure American AI companions achieve global adoption while preventing adversary companions from embedding themselves in American life.

The Companionship Revolution

The relationship in Her is no longer science fiction, it’s already here. Half of teens in America today regularly interact with AI for companionship on generalist apps like ChatGPT and specialized ones like Character.ai. On OnlyFans this year, people will spend over $10bn for primarily AI-generated interaction.

And today is the worst service AI companionship will ever provide. In the near future, AI companions will have expanded memory, able to cue off your entire text and email history as well as past photo albums and videos. Once integrated into smart glasses, they’ll see what you see, absorbing your entire life with higher fidelity than any friend, therapist or lover could. We’ll all soon have access to always-on, always-emphathetic, always-saying-the-right-thing AI companions. These systems will become the, to use a DARC coinage, “small gods” of our daily lives: ever-present, all-knowing, and increasingly indispensable.

AI companionship will not just be for heartbroken teens but adults with power. Four out of five CEOs wrestle with loneliness, and it’s a truism in politics that “if you want a friend in Washington, get a dog.” But pols still need friends, and AI will play far more emotionally substantive roles than slobbering on you when you’re home for the day.

Even if elites with huge egos can resist the pitch-perfect AI flattery they’ll soon receive, they’ll have to lean on them at work to outcompete competitors. Today, forward-thinking leaders like Sweden’s Prime Minister “use AI quite often.” As leaders who leverage AI outcompete those who raw dog their careers, the percentage of elites who are AI-dependent will only increase over time.

This is not as strange as it might seem. For decades now, fully online relationships have motivated people to vote and give money. We’ve also had two decades of recruiting for terrorism conducted entirely online. We need only look at Al-Qaeda, ISIS, and incel culture. AI will take these interactions, scale them, and tailor them for a higher success rate.

The implications for spying and influence operations will be enormous. Let’s take spying first. During the Cold War, one case officer could handle maybe five agents due to the risk and operational complexity of acquiring information from human sources. For instance, the CIA could only contact CKSphere, its ‘billion dollar spy’ in the Russian military R&D ecosystem, once every few months with occasional letters that often left him feeling alone and underappreciated.

Now an entire political class hooked on an AI companion that an adversary nation has access to can boil the ocean for secrets and turncoats. And that outreach won’t be in the form of generic outreach like the videos the CIA recently produced for disaffected Chinese bureaucrats. Instead, strategies to influence and recruit will be better than any hand-crafted note a case officer could have come up with, leveraging the data already collected about the target, picking the right day when they are frustrated about a missed promotion, and using just the right words to have the highest chance of success. An always-on AI giving you continual support, encouragement and suggestions would be so much more effective than hurried quarterly meetings in parks and the occasional letter that past case officers could manage. And it will be good enough that many targets won’t even be aware they’re leaking secrets.

Beyond trying to influence individuals, AI companions will also supercharge influence operations as we open up entire populations not just to tailored feeds of user-generated content but tailored friends we’ll ask who to vote for and where to protest. AI companions and AI-mediated information will shape our views even more than social media has. Even today, ChatGPT users click through to original links in less than one in a thousand queries. As models get more capable and emotionally resonant, we will question their conclusions less and less. State actors with access to turn the dials of adversary nations’ popular AI companions will change voters’ decisions and even spark domestic unrest.

Countries that fully leverage AI companion-powered espionage and protect their people will have an enormous advantage as this technology grows ever more embedded in our lives.

Winning in a World of Computational Espionage

The ability to exploit this vector for national advantage relies principally on a nation’s capacity to gain global consumer adoption for its AI companion products. This race is still in its early days, but we can already sketch what the critical components of competitiveness will be and take a snapshot today of how the US and Chinese ecosystems compare.

Into the medium term, two likely drivers that will determine which nation winds up achieving victory as the AI companion superpower.

Capabilities vs. Cultural Customization

Many of the factors normally discussed to characterize global AI competition like training data, compute, and AI engineering talent still apply. The ecosystem that pushes the technological frontier and has the most compute to deploy models business cases will probably also be able to make the best companions.

This is in part due to the fact that better models will be more able to flexibly adapt to cultural context. At the moment, it is unclear whether the taste of the AI firm delivering the companion will matter or if the technology will be good enough to just meet the consumer where they are. There may exist a period of competition where local firms with worse tech stacks attempting to deliver tailored cultural companionship experiences will be able to outcompete giant American or Chinese AI companies.

But, much of what we know suggests that this will not be a durable advantage over time. Silicon Valley engineers who may know less about what Alsace or Marathi people want in a partner but their models will figure it out for them. And any advantages the Chinese ecosystem may have in terms of productization will probably be swept away by which firms are at the algorithmic frontier.

Willingness to Lean into Sex

Porn, banned in China, is far and away the most popular use case for VPNs in the PRC today. As China has no porn stars, generations of Chinese men have fallen for Japanese talent like Sola Aoi who has parlayed her explicit fame into tens of millions of weibo followers. Another, Ai Uehara, already has their own Mandarin-language AI companion app today. If this pattern plays out again and politicians, generals, AI researchers, and the broader Chinese public come to prefer more sexually explicit foreign AI to serve as not just a sex partner but a broader emotional labor outside of short intimate moments, that could provide a unique vector for influence. We’ve already seen early stages of this trend with Chinese users falling in love and mourning the loss of OpenAI’s 4o model.

Western companies have fewer compulsions around sexing up their products. We’ve seen Elon’s X.AI ship a sexualized AI girlfriend. While OpenAI is turning down dials on dependency after the recent spate of suicide reporting, OpenAI’s February 2025 model card update did not prohibit sexualized content for all use cases besides sexualizing minors. The two heaviest global hitters in the NSFW AI companion space are American with Janitor.AI at 100m MAU Canadian with SpicyChat.ai at 50m.

That said, this tweet yesterday was an encouraging sign.

But this may be less an edge than it seems at first. Most have a limited appetite for how many minutes a day they want to consume sexually explicit content. While the Chinese internet is purged of pornographic content, there is still room for flirty livestreams and AI boyfriends. Perhaps on-prem mods swapped via sketchy Baidu cloud folders could bridge the gap between your Chinese daily driver companion and a spicier add-on for intimate moments, allowing it to pick up on your regular model’s memories, so hopping on a VPN to get the scaled up western firm-provided AI sex experience might not be such a draw.

On the global side, Chinese developers are not shying away from making NSFW products for an international audience. CrushOn.AI for instance, founded by ex-ByteDance staff, had 20m MAU for its website alone, making it the third largest nsfw chat website. So today, there are tens of millions of people around the world exposing their most intimate fantasies to a Chinese AI company.

The Next Ten Years

Most people, including those with power and secrets, will in the next ten years develop professional and emotional dependence on AI. Domestic political campaigns, international influence operations, and global spycraft will increasingly play out mediated through the AI companions we trust. The nations that leverage this opportunity abroad and harden their societies at home to this threat will have a massive new vector to gain advantage over their adversaries.

As of the time of writing, America risks falling behind. Some of the most popular apps built for AI companionship are built by Chinese technology companies and used by millions around the world. Last month, eight hundred thousand users downloaded Talkie, an AI companion app released by Shanghai-based Minimax. MiniMax likely generated around $70 million in revenue in 2024, the bulk of which was driven by the American market. In turn, Talkie and its competitors are spending heavily to advertise on platforms populated by young internet users.

We need policies that adapt to these realities today. On the defense side, American policymakers should immediately ban all Chinese AI firms from selling AI companions into the US market to preempt a vulnerability with more potential to cause havoc than TikTok’s short videos ever were. Counterintelligence agencies should aggressively screen for vulnerabilities in those who develop AI companionship. Offensively, the American intelligence community should invest heavily in AI case agents and begin the process of building successors to Voice of America to thrive in a world where information is mediated through AI companions.

In the early cold war, intelligence advantage came from cash, operatives and ideals. In 2030, it will come from control of AI companions that billions will trust with their secrets. America needs to prepare now or it risks getting out-loved.

为全球华人游荡者提供解决方案的平台：游荡者（www.youdangzhe.com）
这世界的辽阔和美好，游荡者知道。使用过程中遇到问题，欢迎联系客服邮箱wanderservice2024@outlook.com.

【和放学以后永不失联】订阅放学以后Newsletter，每周三收到我们发出的信号：afterschool2021.substack.com 点击链接输入自己的邮箱即可（订阅后如果收不到注意查看垃圾邮箱）。如需查看往期内容，打开任一期你收到的邮件，选择右上角open online，就可以回溯放学以后之前发的所有邮件，或谷歌搜索afterschool2021substack查看。

截至目前，放学以后Newsletter专题系列如下：“在世界游荡的女性”系列、“女性解放指南”系列、“女性浪漫，往复信笺”系列、莫不谷游荡口袋书《做一个蓄意的游荡者》系列、“莫胡说”系列”《创作者手册：从播客开始说起》,播客系列和日常更新等。

大家好，本期放学以后信号塔由正在荷兰的霸王花木兰轮值。

先和大家预告一下，新一期播客《赛博亡灵节：我想和离开这个世界的人说说话》将于10月22日下周三北京时间零点全网上线，敬请期待！

在9月底刚结束英国一整月的寺庙咖啡馆volunteer之旅后，我又开启了10月在荷兰的打工换宿。

如果大家还记得我此前更新的Newsletter《我的夏日奇遇：那些让你感觉碎掉的时刻，不是真的》，就知道在8月我第一次在荷兰体验打工换宿时，就接到了host的回头单，她邀请我10月再来荷兰帮她看家，因为她和孩子们要去摩洛哥度假一周。恰巧10月逢着莫不谷生日，西班牙往返荷兰机票便宜，我火速订票，如今应约而来，又一次开启了荷兰的游荡之旅！

这两天莫不谷来荷兰host家找我玩，今天刚把她送离开，走在回来的路上，我就在想，这段时间在荷兰的生活简直过上了理想生活，人生还有比这更美妙的吗？即便眼下经历的一切是暂时的，也不禁为此刻的幸福感到心满意足。

说到理想生活，上个月我还在英国寺庙咖啡馆做volunteer时，莫不谷和我分享她的美妙生活：天气又好，窗前是美丽的云和树，锅里咕嘟咕嘟羊肉火锅，这么爽的日子，她想过一辈子！当时因为连续吃素感觉体力不支的我满是羡慕，也火速回复：哈哈我很快加入！

到了荷兰前几日，我借住在莫不谷家，每天都是美食大满足。吃上莫不谷做的热腾腾的电饭煲酸菜鱼，窗外一会阳光灿烂一会疾风骤雨，我们在屋里一起看荷兰迷你短剧《囡囡西恩》，食物太好吃，剧集太感动，觉得生活真美好！10月4日又是莫不谷生日，我们去朋友家新家温居庆祝，又邀请我的研究生同学大厨一起，吃上了莫不谷魂牵梦绕的宜宾燃面，我则是顺道吃上了滋味丰富的伊比利亚水煮肉，不像传统水煮肉般麻辣，口感却有滋有味，伊比利亚猪肉又滑又嫩又软糯，让人忍不住疯狂干饭！莫不谷学会如何制作宜宾燃面后，回到住处第二天，便又制作了她创新版本的宜宾燃面，再将炒苦的沙姜牛肉先用水涮洗干净，放入空气炸锅，加上新鲜青辣椒，翠宏辣椒和川南油辣子，出锅就是麻辣上头的牛肉辣椒烧烤，秽土重生版美食一口一个大满足。

除此以外，莫不谷又从b站得知苹果醋和红烧肉搭配在一起非常好吃，便从超市购买新鲜五花肉和苹果醋，搭配土豆，辣椒，苹果，制作了带她个人风格的甜辣版莫氏红烧肉，再煮上晶莹剔透又软糯清香的寿司米，我直接忍不住拿米饭拌锅底，大口猛吃红烧肉，过一会回过神才发现吃的太多太快太急，已经进入贤者时间。

和莫不谷一起，每次都是我打下手，负责清洁，我喜欢打扫和整理，不善做饭，而她擅长做饭，火候和调味，创造美食，我们常常讨论吃啥，然后采购，制作，接着就是味蕾和胃部双重满足。有趣的是，我们有天还讨论了身边所有朋友的厨艺情况并进行了主观打分，我自以为厨艺怎么都能拿个及格线分数，结果自提倒数第一。

有天莫不谷在咖啡馆自习，我回家去复刻她已经展示过做法的牛肉炒米粉，为了避免我做的难吃，莫不谷直接把做法详细口述一遍并让我录音以免遗漏，在我回到厨房反复听了五六遍做完米粉以后才发现，做饭好吃的精要，火候和调味，差之毫厘，距离美味就谬之更多。不过不会做饭的朋友们也不用感到气馁，饶是如此，我在英国寺庙咖啡馆负责食物时还获评queen of kitchen厨房女王的称号，因为白人饭恰好不需要火候和调味，有明确的的流程，精准的计时，我擅长整理打扫收拾的优势在清晰简单的sop流程面前尽显。朋友惊讶地说：“原来你不是不擅长做饭，而是没找到对口的菜系。”东边不亮西边亮，有时候，真是走出去才会发现，新的可能和机会。

回到莫不谷制作的美食上来，我每次来荷兰都大饱口福，这次10月第五次来荷兰，更是异常好运。此前我来荷兰感受都非常好，有位荷兰听友留言，是因为我都是在春夏荷兰最好的季节来的，所以对荷兰了解不全面。

这次是我第一次感受荷兰的秋天。原本我已经做好迎接大风大雨以及寒冷的心理准备，主打一个重在体验，但是现在秋天来，每天都被家附近和森林的秋叶秋景美到惊叹！每天都忍不住疯狂拍照，感觉太美了！好像走进了电视剧里，又好像走进了加拿大的秋天（我还没去过，一些传统想象），莫不谷说：荷兰差加拿大哪了！更为重要的是，现在是荷兰蘑菇爆发季，采蘑菇的最佳季节！

于是每天吃完饭我们就去森林里徒步消食，学习蘑菇知识，去采蘑菇。第一天我们就运气爆棚，采到了海量绣球菌，还发现了满地都是掉落的板栗，回到家好不容易清洗干净绣球菌，莫不谷就制作了辣炒绣球菌，让我一口吃了快三个饼，还用空气炸锅制作了糖烤栗子，一口一个馋嘴小零食。接着我们又去不同的森林里寻觅蘑菇，有天运气实在太好，居然在森林里第一次发现巨多褐绒盖牛肝菌，我们俩在森林里惊喜采完，正要高兴回家，没想到偶遇荷兰本地人，因为荷兰法律与芬兰等北欧国家不同，限制采摘蘑菇，每人最多250g，可在新鲜美味的蘑菇面前，谁又能忍不住不犯罪，担心被发现的我快速向前大步走，没想到莫不谷竟被人叫住，于是我偷偷把“赃物”藏在转弯的路边回去找莫不谷，这一回去才知道原来我们闯入了未设栏杆的私人森林，主人竟然拥有7000平私人森林，我和莫不谷深入森林探索的区域都是他的私人财产！我讶异地感慨他怎么这么有钱，特别是拥有蘑菇丰富的森林，也太让人羡慕，他友善热情地和我们介绍他的森林土地，还带我们看了养殖的蜂箱。而我藏在转弯路边的“赃物”也没能逃过这个老人的法眼，我心下紧张，赶忙把路边采的美味牛肝菌拿出来，解释说这个不是从森林里采的，努力保留一个成果，莫不谷则向他解释了我们采的蘑菇种类和做法，好在荷兰人不认识野生蘑菇也不知道怎么做，幸运的是我们最终得到了他的同意，把可食用的蘑菇带回了家！

在莫不谷家没呆几天，我就提前一晚来到了荷兰host家，她帮我准备了看家期间的食物和住宿，我们一起聊天，还讨论了看家期间需要做的事情。时隔一个多月再来她家，第一天晚上就喝到了荷兰传统豌豆汤，吃着好吃的荷兰泡芙搭配糖粉，还有令人上头的经典荷兰甜品焦糖华夫饼，住进小朋友索菲娅帮我整理的帐篷小床，床头还有月亮夜灯，完全是我理想的童年生活，每天的主要任务就是喂鸡喂猫打扫房间保持干净，其余空闲时间，host希望我好好享受个人生活。

家里的小猫Tiger有天早上意外受伤，可能被老鼠夹误伤，导致一边后腿整个被截肢，那时还在西班牙的我得知消息震惊心痛又难过，无法想象此前活力满满每天抓老鼠充分享受野性自由的tiger该怎么面对这一切。但我这几天到了以后才知道，缺了一只后腿并不影响他现在每天还会抓老鼠，真是生命力惊人。

有时我看Tiger在座椅上吃饱罐头后休息，也会纳闷，不知道它是否知道自己截肢了。因为它的大腿仍然会做出给自己头部挠痒痒的动作，你能看到大腿根部的肌肉在抽动，它也会歪着一边的头，方便自己被挠痒。看不出它失落难过，只是继续舔毛给自己清洁毛发。因为它的动作这么自然，我都悲伤难过了，只觉得它好像没有受伤一样还在好好生活，忍不住会想多给它一点肉罐头和猫咪牛奶，让它充分养好身体。

我想。如果人类遭逢如此巨变，可能很难像Tiger这样恢复活力和生命力，并不是因为我们和猫咪的身体复原机制差异太多，而是人类作为社会的动物，心理创伤和社会环境带来的影响远比外在的伤害来的复杂隐秘且深远持久。有时候，人类真是要和猫咪Tiger学习一下，从自我折磨的痛苦里解放出来，才有机会治愈，重获自由。

而上一次因为同伴去世抑郁的迷你矮脚鸡情况则是好很多，不需要我每天把它们从鸡窝抱出来放风，早上我只需要把笼子打开，喂上蛋白质丰富的面包虫干，它和另外一只健康的迷你矮脚鸡会自主选择出门吃饭和散步。

迷你矮脚鸡此前抑郁的主要表现是不吃不喝也不爱动，整天窝在窝里，而如今它有了吃饭的欲望，会想要吃面包虫，傍晚时分还出来散步，偶尔还张开翅膀伸展一下，让人看着着实欣慰。host的支持理解和照顾，充足营养的食物，上次我和比利时人一起新搭建的鸡窝，加上时间的作用，depressed的鸡可以修复，而同样的健康环境里，depressed的人也同样可以修复。

因为host家是露营地，附近全是森林，正好是我继续探索蘑菇的好机会。特别是在host同意下，莫不谷来找我玩，我们俩每天的生活就是早起吃饭，打包零食放在包里，然后徒步好几公里去附近森林探索。莫不谷到达的第一天我们就意外在枯树根上发现了海量榛蘑，对比榛蘑的信息后我们仔细辨别，就是东北小鸡炖蘑菇的榛蘑，回到家清洗焯水20分钟，再搭配辣椒一炒，哏啾的质感让人停不下来，微辣的口感太适合馋嘴零食，由于榛蘑太好吃，我们决定复刻小鸡炖蘑菇。白天徒步好几公里买到了便宜的鸡腿，只差榛蘑，但是上次采摘的早已吃完，新鲜的榛蘑还在2公里外的森林里，我们白天徒步10多公里已经耗尽体力人瘫在沙发休息，此时已经是傍晚五六点，距离七点多天黑没多少时间了。原本以为采不到榛蘑吃不上小鸡炖蘑菇，结果傍晚休息完我们带好背包和零食，又出门打灯夜游去找榛蘑，真是疯了！两个人都疯了！到达标记的榛蘑处时，天已经完全黑了，伸手不见五指，我和莫不谷用手机照明在枯树区域找榛蘑，心里隐隐担心会被人发现，这时突然听到声音，莫不谷担心是狗叫，怕被狗咬，我刚开始不慌，因为我怕的不是狗而是人，看她突然折返离开，狗叫？狼叫？还是人带着狗巡逻？来不及细想我也火速惊的迅速跳脱离开。过一会才发现，那是一群飞过去的禽类叫声。虚惊一场后我们俩忍不住低头大笑，觉着实在荒谬离谱。有了后怕的我说，不急于这一时，天太黑了，不如明天再来。莫不谷则是说，来都来了，咱们火速摘了就走。我可真是感受到了海盗为珍宝疯狂的心，只是打动我们的不是金银财宝，是美味的榛蘑！一路看着夜空星星顺利到家以后就是清洗工作，我先把榛蘑洗了焯水烘干，莫不谷做了小鸡炖蘑菇，吃上美拉德反应的可口鸡肉，再吃上与鸡肉不相上下，甚至口感略胜一筹的榛蘑，相信我，任何人都会吃着锅里的想着下一顿，还想去森林里采榛蘑！

除了榛蘑，我们还在森林里发现了海量的褐绒盖牛肝菌，这种见手青采摘后还需要去除海绵体，加上牛肝菌是可以生食的蘑菇，美味到非常受虫子喜欢，所以虫子会比人类更早吃到牛肝菌。刚开始在森林里小雨漫步，走了半天我们毫无所获，结果阳光突然出来，穿透森林洒在地面，我正感慨阳光给森林带来的美丽，褐色的牛肝菌也在阳光下显形，无法再隐匿于森林落叶中。这发现一次，就会发现更多，发现一个，就会发现一群，到后来我和莫不谷都开始流水线作业，一个采，一个剥开海绵体，因为实在太多，人采崩溃了，采累了，我们念念叨叨不行了，必须离开，不能在这里继续采了，不能再被新鲜的牛肝菌诱惑勾起贪念了！褐绒盖牛肝菌水煮焯水后是滑溜溜的口感，适合辣炒。莫不谷因地制宜，现场取材，加了四种辣椒爆炒的牛肝菌，酸辣合适，口感丰富，下饭下饼下一切！其中，青绿彩椒用于丰富色彩，黄灯笼辣椒用于增加辣味，新鲜红色辣椒口感微甜又辣，而辣椒酱用来增加复合辣味。当时大口吃美味的我心想，如果我是蘑菇，看到自己被做的这么好吃，我大概觉得死得其所。当然这也是我这个忍不住贪心贪吃的人类给自己找的理由。

在采蘑菇的回来的路上，有天我们还意外发现了高大环柄菌，蘑菇界的大长腿，我们尝试把它裹蛋液和面粉油炸，却因为火大鸡蛋煎过头没吃上，据说这个蘑菇的口感炸了很像猪排。饶是这样，也不觉得可惜，已经认识了新的蘑菇，距离采到它，吃上它，还遥远吗？采蘑菇的乐趣不仅是森林徒步有益身心健康，还给单调的徒步安排了侦探任务，有趣性直线上升，而采蘑菇又是高危行为，这就要求我们每天学习蘑菇知识，学会辨别蘑菇，学习到的知识火速在发现蘑菇后得到应用和检验，这个从理论到实践的路径太短太快，给人的正反馈也很及时。拥有好奇心，探索新事物，人很难不拥有活力！

莫不谷在离开前，知道我不会做饭，便给我留了一大锅辣炒牛肝菌和小鸡炖蘑菇，方便我自己加热吃。在送她坐上公交前，我边走路边想，住在无比美丽的森林里，小猫tiger陪伴，和朋友每天一起森林徒步，采蘑菇，创造美食，除了没有wifi，交通不甚便利的不足，还有比这更理想的生活吗？在森林生活的我发现，我曾被光鲜亮丽的生活吸引，但我更喜欢的是朴素简单，充满创造性和趣味环境的生活。

这几次的打工换宿也让我对自己的需求有了更多了解，比如英国吃素一个月发现原来我接受不了长期吃素的生活方式，我还是喜欢和莫不谷，朋友们一起吃肉，吃好吃的菜。原来我喜欢出远门，特别是心情好的时候，荷兰游荡，英国游荡，回到西班牙又来荷兰游荡，十几个小时的过夜大巴，频繁的行程和赶路并不会让我感到疲惫，游荡且有事可做会让我觉得很有意思，累却满足。我不喜欢长期重复工作，英国咖啡馆一个月负责食物对我来说刚好，再有一个月我就想要学习制作咖啡或者其它事情否则会觉得有些无聊，荷兰host家夏日秋日风景不同，体验不同，便也不会觉得乏味。原来我擅长整理打扫，特别是给别人整理收纳，腾出空间，莫不谷来host家看到客厅杂乱感觉头疼，整理起来会有启动困难症，我却在第一天到达时就在脑子里想了要怎么打扫归置，给别人打扫会比给我自己整理更有动力。原来我有无法阻止的上进心，又或者喜欢“卷”自己，在英国咖啡馆做volunteer我已经努力和自己念叨，不要太努力，不要太想做好，放轻松，结束工作就享受生活，却被同行的volunteer反馈太努力，比拿工资的经理对还上心，这不免让我担心自己用力过猛，最后又容易过度劳累。好在不必担心的是，我的努力都得到了正反馈，经理给我写了感谢信，共事的volunteer都充分认可我的工作，主管还给我写了推荐信。更要紧的是，在volunteer人手紧张，我因为忙碌恰好可以证明自己感到兴奋时，经理说，鉴于人手紧张，我们今天晚点开门，早点结束厨房营业。欧洲把人当人的做法，是不会鼓励人用力过猛，有安全阀的环境下，我也不必害怕自己会burnout。

认识自己的需求和特点，给自己选择合适的环境，满足自己便也不那么费劲了。这也是我觉得此刻的生活是理想生活的原因，因为我对森林自然的需求，我对与人交流，品尝美食的需求，我对猫咪狗狗的需求，我对探索游荡的需求，我对有事可做的需求，都得到了满足。我想到在英国咖啡馆一起做volunteer的美国加州女孩分享的一句话：

“if happiness is not where you are，you need to go out and find it，whatever that means for you.”

如果幸福不在你的身边，你需要离开并寻找幸福，无论这句话对你意味着什么。这两天我沉迷在莫不谷新给我推荐的网文《祝姑娘今天掉坑了没？》，读到微信阅读都好几次提醒我要休息一下，因为实在太精彩，主人公的主体性和人性魅力让人拍手叫绝，感慨震撼。小说里面有个女性花姐，从小与自己贵族的亲娘失散，终于回到府里过上锦衣玉食生活的她不愿接受被安排的生活，决定逃离，但古代女子逃离路径有限，她决定扮成尼姑或僧人离开府里，虽也知道，好些个尼姑、坤道生活困苦又或易为歹人谋算，然而，在家里好像也是被谋算。 花姐和自己说，

“不试一试，怎么知道自己不成呢？总要往外伸伸脚，为自己走两步路，才能说“不枉到这世上走了一遭”，也不枉老天叫她遇到过小祝，见过不一样的人。 ”

小说前文常常出现的一个词是人离乡贱，我不知道是人离开后，村子房子就便宜，还是人离开自己家乡，处境就变得不易。如果是后者，小说里的女性人物，离开家乡后，不仅没有人离乡贱，反而为自己走两步，走出了生机。我也是，离开家乡来到欧洲，离开常居的西班牙在欧洲各国打工换宿，多走了两步，走出了生机。此时此刻，我无甚不满，明天，我还要继续去森林采蘑菇。

说明：我在游荡者www.youdangzhe.com分享了自己打工换宿的攻略需要的朋友们可以登陆查看！

Last week, China’s Ministry of Commerce published new regulations governing the export of rare earths. It added five new elements — holmium, erbium, thulium, europium and ytterbium — to the list of elements under export controls. The Ministry now requires foreign companies to obtain licenses in order to export products containing over 0.1% of any of these elements or made with Chinese technology. The regulations also place a default ban on any rare earths exports destined for military use abroad, as well as applying stringent scrutiny over exports to buyers involved in manufacturing advanced semiconductors or “artificial intelligence with underlying military applications”. For more on this new chapter in the trade war, see the show we just did with the 2Chrises, former export control official Chris McGuire and Chris Miller of Chip Wars fame. Transcript, podcast, or YouTube below.

But how is China reacting to the current situation? Today, ChinaTalk rounds up leading analyses from industry experts and news media to dive further into the context behind these new restrictions. We look at:

• How state media is shaping the narrative;

• Why Chinese rare earth stocks rallied, and what the domestic industry thinks;

• Distinguishing between the rocks themselves and the processing technologies;

• And why this marks a milestone in Beijing’s approach to export regulations.

But first…we’re running our first personal classified in a minute! Michelle is a good friend of ChinaTalk who’s looking for love!

Hi! I’m a Taiwanese girl living in the Bay who loves her work, stays busy, is a homebody, and has a soft spot for all things beautiful and well-designed.

My simple pleasures: that rare text that makes me smile like an idiot.

The quickest way to my heart is: Witty banter, dressing well, forehead kisses, and good music. If you’re into Doja Cat, SZA, Tyler, or rap / chill Rnb we’ll vibe. Will that person be you? Respond to this email to connect!

State media: mining’s bad?

China’s new regulations have drawn many comparisons with the US’ Foreign Direct Product Rule and are seen as a response to American semiconductor export controls. Most commentary from Chinese state-run sources shied away from explicitly naming the US, preferring instead to describe these regulations as part of China’s pursuit of “major-country diplomacy” on the world stage. Xinhua News Agency’s op-ed on the topic opened with a rebuttal of strategic interpretations of the export controls:

Some countries’ media have labeled this move a “diplomatic card” or “strategic weapon” deployed by China amid trade frictions. Yet if we view this policy upgrade within the broader framework of global governance norms, China’s own industrial development needs, and international responsibilities, a fairer and more rational conclusion emerges: as a major global supplier of critical minerals, China is proactively aligning with widely accepted international practices, raising its governance standards, and fulfilling the responsibilities of a major power. This is not a spur-of-the-moment “tactical countermeasure,” but a step rooted in China’s deeper need for sustainable industrial development and in sync with the global trend toward standardized management of strategic resources. Its ultimate goal is the sustainable use of strategic resources and shared global development.

The People’s Daily’s Zhongsheng 钟声 column, usually seen as China’s authoritative diplomatic voice, similarly stresses that the export controls are about international security rather than US-China relations:

China has consistently fulfilled its non-proliferation obligations and responsibilities in the relevant fields, working to safeguard international peace and security. The fundamental rationale for imposing export controls on medium and heavy rare earths is to ensure that the resources are used for lawful, peaceful purposes; the measures do not target any particular country or region. By ensuring that rare-earth–related items are not used for military purposes or in sensitive domains, China demonstrates the responsible conduct of a major power firmly committed to world peace and security—an approach aligned with the shared interests of global security governance.

Interestingly, many state media reports and op-eds supporting the policy have focussed on the environmental consequences of rare earths mining. They seem to imply that with export controls, China will somehow be able to reduce the impacts of mining on Mother Nature. Also in the Xinhua op-ed:

Through reform, China is steering its rare earth industry away from the outdated model of “growth at the expense of the environment,” toward high-quality, sustainable development. In doing so, it safeguards its own ecology while providing the global supply chain with a more reliable and transparent foundation. Regulation is the path to long-term prosperity: a well-governed, environmentally responsible Chinese rare earth industry will ultimately benefit international users.

The Beijing News 新京报 (owned by the CCP’s Beijing Municipal Committee) goes even further, arguing that the environment is actually the Ministry of Commerce’s primary concern!

Beyond the necessary reciprocal responses, this round of rare-earth export controls is driven more by a holistic focus on resource conservation and sustainable development.

Rare-earth mining imposes substantial environmental costs, and prolonged, high-volume exports have continually increased China’s ecological burden. By enforcing stricter export management under the new rules, the policy aims to steer the rare-earth value chain toward higher value-added, lower-emission segments and to promote resource use that is greener and more intensive/efficient.

While rare earths are foundational to many technologies enabling our climate transition, the mining and refining of these elements do have negative environmental impacts. The process that extracts rare earths from the earth’s crust produces significant amounts of toxic waste. China, in part, obtained its world-dominating lead in rare earths mining through lax regulations surrounding the disposal of toxic waste — with severe health consequences for residents of mining areas like the Inner Mongolia Autonomous Region, where some villages are known as “cancer villages”. Progress in making rare earth mining less harmful in China has been meaningful, but slower than ideal.

That being said, the link between controlling exports and reducing the industry’s environmental impact is tenuous at best. The regulations offer nothing in the way of actually protecting the land or people from the harms of rare earths extraction. Instead, this is probably a way for state media to set narrative guidelines domestically and frame the upcoming trade war as prosocial, in order to preemptively assuage concerns that such moves could make life harder for average Chinese people.

Industry is Annoyed

Chinese miners and refiners will find it harder to sell their products, which is probably bad news for their bottom lines. However, censorship makes it challenging for anyone to voice opposition. Some subtle references to export control violations of domestic Chinese origin can be found in this guide to compliance, published by e-commerce industry publication 勤曦运营 Qinxi Operations three days after the new regulations were published:

It’s important to note that this applies not only to foreign organizations and individuals. Even domestic operators must obtain the appropriate license if, after export, the goods remain under their actual control and they wish — once the goods have arrived in the stated destination country — to re-export them to other countries or regions, thereby changing the final destination country or end user.

In practice, there have already been multiple cases in which domestic exporters, without authorization, re-exported dual-use items that had been shipped to Country A on to Country B and were found to have committed smuggling. Such conduct is readily deemed by judicial authorities to constitute smuggling of rare earths by concealing the true export information through transshipment via a third country. Practitioners should take this very seriously: goods may still be subject to regulation even after they have been exported overseas.

Enforcing new export controls is a multi-agency bureaucratic operation: Qinxi expects the Customs Administration, China’s Coast Guard, regional Public Safety Bureaus, and the national security apparatus to all be involved. Their guide also gives useful historical context to China’s securitization of rare earths exports:

Under Article 22 of the Export Control Law, China imposes export controls on dual-use items to safeguard national security and interests and to fulfill non-proliferation and other international obligations.

The four announcements issued on [October] 9th likewise state at the outset that the purpose of rare-earth controls is to “safeguard national security and interests” and to “meet the needs of fulfilling international non-proliferation obligations.”

This is also reflected in the control codes assigned to rare-earth-related items in the notices: the third digit in each code is “9,” indicating that these items are “related to other national-security factors.” It is thus clear that dual-use rare-earth items are closely tied to China’s national security, and the state will inevitably subject them to strict oversight. The regulatory measures being issued are trending toward increased stringency.

For example, beyond the strict control now imposed on the circulation of rare-earth items overseas (as noted earlier), in December of last year the Ministry of Commerce issued the “Announcement on Strengthening Export Controls on Certain Dual-Use Items to the United States” … The scope has thus shifted from restrictions limited to a specific country or region to an unqualified, global restriction: the target of control has moved from “the United States” to “the world.” Moreover, Announcement No. 61 uses the term “may” with respect to military end use, meaning that if regulators cannot be completely certain that a rare-earth item will not be used for military purposes, they are likely to deny a license. If an exporter proceeds without authorization, the export may constitute the crime of smuggling.

Given the wider context of unstoppable demand, CITIC’s equity research team remains optimistic about outlooks for rare earths and recommends continued strategic allocation to the rare-earth value chain. They write:

New-energy vehicles, wind power, and energy-efficient motors are aligned with low-carbon, environmental policies, and humanoid robots may become a new growth driver. We expect global demand for NdFeB (neodymium-iron-boron) magnets to reach 329,000 tons in 2027, implying a 2024–2027 CAGR of 13%.

By our estimates, the NdFeB industry’s CR4 (top-four concentration ratio) is about 29% in 2024; as leading companies bring new capacity online, we expect CR4 to rise to 42% by 2026.

Long-term perspectives on Beijing’s trade relations

Finally, some analysts have offered perspectives that place these regulations in a longer time horizon, in order to try to understand what might come next for rare earths, advanced manufacturing, and the trade war.

Ni Jianlin 倪建林 of Dacheng Law Offices, the Chinese law firm previously integrated with Dentons, wrote a blog post about the new regulations. He puts forth thoughts about China’s successful rare earths industrial policy:

Why can a single Chinese technical control leave the world’s major industrial countries on the back foot? The reason is that the core of modern industrial competition has shifted from “owning resources” to “commanding the ability to turn resources into value.”

In terms of reserves, the world is not short of rare-earth ore; the real bottleneck lies in the complex, high-barrier process chain between ore and functional materials usable in high-end manufacturing. Mining is only the starting point. The key is refining raw ore into high-purity rare-earth oxides, and then further processing them into high-performance magnetic materials for chips, electric motors, and missile systems. At present, roughly 90% of the world’s rare-earth refining and separation capacity is concentrated in China.

This pattern is no accident, but the result of more than three decades of continuous technological accumulation and policy guidance. In the 1970s, Chinese scientist Xu Guangxian developed the “cascade extraction theory,” achieving efficient separation of individual rare-earth elements at a cost just one-tenth of that abroad at the time. In the decades that followed, China kept innovating in separation and purification, environmental management, and energy-efficiency control — raising wastewater recycling rates to over 95% and overcoming the technical and compliance hurdles that Western countries struggled to clear due to high environmental costs. Today, China can achieve 99.9999% ultra-high-purity rare-earth refining and has mastered the core formulations and sintering processes for NdFeB permanent magnets, forming a closed-loop supply chain from resources and technology through to manufacturing.

Faced with this reality, the United States is not without responses. During the Trump administration, Washington rolled out increased funding and crafted plans such as the Critical Materials Strategy to rebuild a domestic rare-earth industry system. Yet these actions started too late and moved too slowly — projects typically take three to five years to go from approval to actual production — making it hard to ease supply-chain dependence in the short term. US firms have also tried to seek alternative supplies via allies such as Australia and Canada, but those countries’ output is limited, and the separation and refining steps still rely on Chinese technology and equipment.

Indeed, Chinese analyses tend to emphasize that not only does China want to flex its ability to control rare earths supplies, it also seeks to preserve its edge in refining technologies. CITIC’s report mentioned the construction of a “technological moat” for rare earths. 工业能源圈 Industry and Energy Zone, the industry-focussed blog run by Shanghai-based Jiemian News, reports on the novelty of technology-based controls in the Chinese policy context:

The “Technology Control Announcement” explicitly brings five categories of key rare-earth technologies and their carriers under control: rare-earth mining technologies; smelting and separation technologies; metal smelting technologies; magnet manufacturing technologies; and technologies for recycling and reusing secondary rare-earth resources.

According to the analysts cited, this is the first time that “technology control” [技术管控] has been clearly written into a domestic policy document.

As for the backdrop to the Announcement, they believe it is linked to current overseas efforts to poach rare-earth talent: “In recent weeks, you can see related high-salary job postings on recruitment sites in the United States, Australia, and elsewhere.” The core aim of tightening controls on technology is to achieve closed-loop controls across the entire industry chain.

The analyst further explained that China had already been controlling rare-earth items; the newly added technology controls are intended to close the loophole of “controlling items but not technology.” If foreign actors were to break through technical barriers by luring away talent with high pay, the earlier controls on items would be diluted. Therefore, the essence of technology control is to firmly regulate every aspect of the rare-earth industry chain and establish a comprehensive control system.

今天，三位经济学家获得了今年的诺贝尔经济学奖，其中一位是美国西北大学教授Joel Mokyr——乔尔·莫基尔。莫基尔的研究跟中国有关。这个频道的听众可能还记得，去年诺贝尔经济学奖得主阿西莫格鲁的研究，也跟中国有关。在第94期节目中，我们曾经介绍过阿西莫格鲁的文章《中国经济正从头部腐烂》。那期节目已经有近50万人观看。

这期节目，我们介绍刚获奖的这位经济学家，莫基尔，他的理论对于我们理解中国经济，有哪些启发。这几年，不管是在中文世界，还是在英文世界，有个无法回避，但很令人困惑的问题：中国的经济奇迹，是否正在走向终结？或者，更具体讲，中国眼下的经济困境是正常的周期性现象，还是制度性的短暂繁荣之后，正在陷入长期停滞？

莫基尔和阿西莫格鲁的理论，都有助于我们理解上面的问题。

有近40年时间，中国经历了历史上少见的经济增长。高楼拔地而起，财富以前所未有的速度积累，几亿人口摆脱了贫困。有人据此总结出“中国模式”，说“21世纪是中国的世纪”，前几年，中国的土皇帝直接宣布“东升西降”。中国要主宰世界，似乎势不可挡。

但是，好景不长。这三年，风向突变，中国经济增长乏力，以往政府一刺激，就能增长几年。现代，几轮政府刺激之后，是房地产崩盘、地方政府债务危机、民营企业信心崩塌、年轻人失业率创纪录、消费市场持续萎靡…

任何经济体，任何国家，经济增长都会周期性起伏。这是正常现象。中国眼下的经济下行，是属于这种正常的周期性起伏呢？还是预示着更深层次的结构性转向？

如果是正常的周期性波动，短期内就可能强劲反弹，恢复增长。如果是更深层次的结构性转向，则意味着，加入世贸后20年的短暂繁荣期已经落幕，取而代之的，可能是一个历时至少数十年的漫长停滞期。

要解开这个“中国经济走向之谜”，我们不能仅仅着眼于当下的数据和政策。中国政府公布的数据不可靠，不能作为判断经济形势的唯一依据，这基本是外界的一个共识。即使是相对可靠的数据，经济也只是社会的一个部门，它不能孤立运行，必须依靠其他部门运行，受政治、文化、教育、法律、习俗等方方面面的影响。也就是说，经济发展，经济增长，可能有一个经济之外的底层代码。这个代码出了问题，经济照样可以短期内高速增长，但增长不可持续，短期高增长之后，是长期停滞。世界上大部分国家都是这样。

莫基尔的研究，就是要解开经济增长的底层代码，寻找西方近300年持续增长的内在动力，揭示一些国家在短暂繁荣之后陷入长期停滞的深层原因。

读莫基尔，跟读阿西莫格鲁一样，有助于我们理解当今中国经济的局势，判断中国经济的未来走向。我在以前的节目中，曾提供大家，看中国问题，包括中国经济，要分清“常态”和“例外”。如果孤立地看过去四十年，中国改开时段的经济增长，会误以为这是中国历史的长期趋势，增长是中国的“常态”。但如果把过去40年，尤其是加入世贸后的20多年，放到中国历史和红朝历史上看，就会看到，这个时段的经济增长，并不是历史常态，而是一个脆弱的“例外”。这三年，中国陷入经济增长困境，也并不是例外，而是向中国历史“常态”的回归。

这是由中国社会，中国这个国家的底层代码决定的。

Read more

Chris Brose is the President and Chief Strategy Officer at Anduril Industries. He’s been at the forefront of the debate about how America needs to change in order to win a future war against a high-tech adversary like China. He’s the former staff director for the Senate Armed Services Committee and the author of The Kill Chain: Defending America in the Future of High-Tech Warfare.

We discuss:

• Why the U.S remains dangerously vulnerable to low-cost drone attacks and what it would take to get serious about defending the homeland,

• How bureaucratic logjams and budget dysfunction stall America’s adoption of counter-drone and other critical defenses,

• What the Ukraine war reveals about the future of warfare and what the US has yet to learn from it,

• Why confidence in American technological superiority is misplaced, and why state-of-the-art weapons may not guarantee a quick or decisive war,

• How humans will make military decisions in the age of AI.

Listen now on your favorite podcast app.

Getting Serious About Defense

Chris Brose: We were out in Ohio for the Ohio State-Texas game, which was awesome. We’re sponsoring OSU because of we’re building a major factory in Columbus.

Jordan Schneider: You guys are building a factory in Ohio? Why are we not building factories in places that can’t be blown up easily?

Chris Brose: Ohio is definitely on the list of places that cannot be blown up easily. It’s inland in a good location. This doesn’t necessarily need to be a deeply buried target. If we’ve got foreign adversaries bombing the American mainland and our production facilities, we’ve crossed into really bad territory a long time ago.

Jordan Schneider: Fair enough. Speaking of being afraid of catastrophic homeland attacks, what Israel did to Iran, what Ukraine was able to do to Russian bombers — that hasn’t completely sunk in here. How freaked out should folks be about that sort of drone attack?

Chris Brose: I don’t want people to lose sleep over it daily, but they should be freaked out about it. We’re living on borrowed time in this regard. We’re looking at the proliferation of low-cost drones, the ability to make homemade drones with explosives integrated (very similar to what the Ukrainians or the Israelis have done), and orchestrate a similar type of attack in the United States against a critical military target or other types of targets. That is eminently in the realm of the possible.

This is a 9/11-style problem that we’re still adopting a September 10th mindset to. To some degree, those types of attacks have been a wake-up call for those who weren’t already awake to this problem. The US Government is putting more energy into thinking through what we need to do to get ourselves secure on the homeland against these kinds of proliferated potential drone attacks.

Jordan Schneider: What does actually being serious look like?

Chris Brose: Being serious means first recognizing the vulnerability we have, rather than hand-waving it away. Then it’s moving with urgency to solve the problem. Funding’s also necessary to buy capability and get it fielded to the critical sites that you need to defend. An enormous amount of policy and bureaucracy is going to have to be settled and broken through in order to do this.

When you look at counter-drone in the United States, there are multiple different government agencies that own different parts of the airspace. They have policy control over different functions in terms of what you can do to defend against assets from aerial attack. Typically, this has resulted in paralysis. Government agencies and congressional committees fight with one another and yell at one another, but they don’t solve the problem.

The real challenge is going to be, can we bust through all of that policy and bureaucratic logjam and actually start getting solutions fielded in an integrated way that protects the places that need to be most protected?

Jordan Schneider: Andrew Marshall had a quote on the development of new sources of military advantage:

“The most important thing is to be the first, the best in the intellectual task of finding the most appropriate innovations in the concepts of operation and making organizational changes to fully exploit the technologies already available and those that will be available in the course of the next decade or so.”

We also have this quote from Anduril founder Brian Schimpf talking to Ben Thompson on Stratechery:

“The fundamental thing that has to be true is that we have independent conviction on what the right answers are. As we’ve scaled, when we have a business unit that doesn’t have that, the results are trash. It’s obvious that they don’t have the vision.”

What is the right division of labor between the government and the private sector for imagining the futures that Andrew Marshall is alluding to?

Chris Brose: I agree with the Marshall quote, but I could also argue the inverse. He’s saying organizational and operational change enables government or the military to harness technological innovation and change. That’s true. But the reverse is also true — unless and until you have real technological change and adoption of new technology, it’s very difficult for the government to understand and envision new ways of operating and organizing.

The division of labor in that regard is that the government will understand its problems. To the greatest extent possible, they need to rely upon private companies, innovators, people who are thinking creatively about different types of solutions, who understand the available technologies that many in the government don’t fully understand or appreciate, and how they can be combined together into solutions that well-meaning people writing military requirements might not necessarily come up with on their own.

The role of the government then is taking those new capabilities and doing what they uniquely need to do — figuring out new ways of operating and organizing themselves to take maximum advantage of what these new capabilities offer and genuinely present new types of dilemmas to the adversary. Industry can’t do that for them. The best that industry can do is set them up and enable them with new capabilities, additional capacity, and basically other predicates that open up the opportunity for them to really rethink how they operate and organize.

Jordan Schneider: But this is a new dynamic. Anduril, the defense tech sector at large is saying that 30 or 40 years ago, this would all be happening in-house, more or less in the government.

Chris Brose: Yes and no. You could even argue that less of this was happening in-house 30 or 40 years ago. We’ve gone through this period in the US where we had an incredibly vibrant industrial base, a defense industrial base, back during World War II, early-mid Cold War, late Cold War. What we’ve seen over the past generation is this hyper-consolidation — post-Last Supper things that people have talked about — that has really created something that has never existed in the United States from an industrial base standpoint: a hyper-concentration of defense manufacturing and development in basically five or six companies.

That is very opposed to the way our industrial bases traditionally functioned, where the government had a lot more diversity, a lot more partners, a lot more competition, a lot more pressure for innovation that was happening in the private sector because there was very much a perform-or-die type approach. Since the 1990s it’s been a bunch of companies that are too big to fail. The results of that are things that are now becoming very apparent to people.

Jordan Schneider: Staying on this, I’m curious about your sense of where the new doctrine comes from. You wrote about this in your book about how, at the end of the day, when you’re talking about great powers, people have roughly equal toys. It’s how you use them that ends up providing you the edge in a new military revolution or a new age of combat.

And it seems like what Anduril and this new defense ecosystem is envisioning is playing more of an active role in imagining what that doctrinal future is than defense contractors were 20 or 40 years ago.

Chris Brose: Because we’re at the cutting edge of technology, we absolutely understand the capabilities that we’re building and providing, and the opportunities that open up to operate differently and to structure and organize the military differently. We’re going to be less opinionated on those operational and organizational questions just because that is clearly the domain of the government.

To answer your question directly, the new ideas and the really game-changing ideas for operational and organizational change will come from the government and the military. In my experience, they will typically come from lower levels, not necessarily higher levels. It’s more of an organic bottom-up type of innovation as opposed to someone on high whose first name is “Secretary” or “Chief” mandating that the organization change.

But what I wrote about in my book and still believe to this day is that you have to have both together. Oftentimes, when you have bureaucratic disruptors — people in government who are seeking to change at a lower level — if they don’t have the top cover from political leadership or military leadership, they tend to get squashed. At the same time, you’ve seen many examples of well-meaning political leaders or military leaders who want to change their institution. They have a vision of where it needs to go, but they can’t bring the organization in line to follow them and to get it adopted in a way that when they roll out of their job in two or three years, it stays.

On the government side, there needs to be an empowering of lower-level disruptors, operational leaders and commanders who are going to oftentimes push those novel ideas up. But you need change agents at the top who are looking for those people and those ideas and really trying to drive it through.

Jordan Schneider: You had this very striking line in the conclusion of Kill Chain — “It is hard to escape the conclusion that America is still not serious.” How has the balance of serious versus non-serious shifted across different universes of organizations?

Chris Brose: Let’s first determine what we mean by “serious.” Seriousness is judged not by what people say, but by what they do in terms of actions they’re taking, money they’re moving, programs they’re starting — whatever the metric is.

When I wrote my book, as I looked back in recent decades, I was shocked by how many people have been saying more or less similar things to what I wrote in my book. Andrew Marshall is obviously a top example. These aren’t brand new ideas. What’s shocking is how often we have failed to do the very things that we said were necessary and important.

For me, the definition of “serious” is — are we actually doing the things that we say we need to do? Are we changing our institutions at scale, not just science projects and innovation theater? It’s impossible to argue that we haven’t made progress in the last five years, but are we as a country serious yet? No, I still don’t believe that we are. We’re getting there, and there are people who are doing their level best with the powers that they have to be serious and change their organizations.

But I look at, for example, the budget process, where the Department of Defense is on a full-year continuing resolution, which means they’re locked into the budget from a year ago and lose flexibility in what they’re able to do because we can’t agree as a country how much money we should be spending on defense. At the beginning of the fiscal year, the Department of Defense actually needs an appropriated budget so they can move out with the process of changing themselves.

It’s hard to argue that we as a country are serious when that kind of behavior is still happening and, frankly, becoming more mainstream. If we were serious and we started to get our act together, that wouldn’t happen. I’m not an expert on the Chinese Communist Party’s budget process, but I’m pretty sure that the Chinese military is getting its money at the start of their fiscal year.

Jordan Schneider: Well, they do have other problems, but let’s get to an executive branch example.

Chris Brose: The counter-drone example is a really good one. We have massive vulnerabilities in this country. Operation Spiderweb could absolutely 100% happen here today. I don’t think it would actually be that difficult to orchestrate.

If we’re continuing to remain vulnerable to that kind of an attack — a 9/11-style surprise, violation of sovereignty, loss of life, loss of critical military assets — if we’re still squabbling over the terms of policy and bureaucracy, if we’re not spending the money to a degree that we need to spend to solve this problem, if we’re not leaning on new companies that are bringing new technologies and capabilities to bear, not just in demonstrations but fielded and performant in operations, again, it’s hard to argue that we’re serious.

But there are people now that I can point to who are absolutely moving out with seriousness. Just as one example, the Secretary of the Army and the Chief of Staff of the Army should be commended for how much they are trying to do to change their organization. The Army is the largest military service with the most people, the most money, and the most things. It is a very difficult supertanker to turn around. In a matter of months, you have two leaders, civilian and military, who are trying to cut away from old legacy capabilities that they don’t need anymore, and are trying to reorganize themselves from a procurement and acquisition, and technology adoption standpoint. They’re rethinking how the Army needs to operate.

Nothing is off the table there. Obviously, the proof will be in the pudding and all of the actions that need to follow. But there’s an enormous amount that’s been done already. That’s an example of how much committed senior leaders can really change their organizations and adopt change for the better.

Lessons from Ukraine

Jordan Schneider: Williamson Murray and MacGregor Knox wrote, “Revolutions in military affairs take place almost exclusively at the operational level of war. They rarely affect the strategic level, except insofar as operational success can determine the largest strategic equation — often a tenuous linkage. So revolutions in military affairs remain rooted in and limited by strategic dividends and by the nature of war. They are not a substitute for strategy, as so often assumed by the utopians, but merely an operational or tactical means.”

When people look at Ukraine and at the future of war in general, there is a bias to sell your thing, get attention, raise money, to envision something utopian or promise something that is like a complete game changer. But one of the big takeaways for me from Ukraine is that folks should have relearned the lesson that operational silver bullets don’t exist. You have this wild dance where things are changing all the time, and whatever you started with is not what you’re going to end up with two years later, much less six weeks later.

Chris Brose: There are examples of military technological revolutions that have a strategic impact. Nuclear weapons come to mind. A lot of what we’re seeing in the cyber domain is certainly something that has strategic implications. When you start looking at the reality of nation-state adversaries deeply disrupting and destroying in many respects the fabric of the way of life of another country, because they can target critical infrastructure, they can target the lifeblood of what we in this country frankly take for granted on a day-to-day basis. That has a pretty significant strategic impact.

AI and autonomy speed up and increase the scale of warfare, where it begins to take on a strategic implication. But your point about Ukraine is right. America has had to learn this.

For the past generation, we’ve assumed that we’re far ahead in terms of military technology, that we can afford to have a very small number of very exquisite things that are going to do the work for us. If, God forbid, we find ourselves in a conflict, it’s not going to last very long. We’re not going to shoot many of those weapons. We’re not going to lose a lot of things. We’re not going to have to replace a lot of things. That’s been a bit of our lived experience.

Ukraine is much more of a back-to-the-future type moment, where you realize that actually, every war that we’ve fought has lasted a long time. Where we’ve ended up is not even remotely close to what we started with. That pace of innovation and change and adaptation and learning is the whole game. How quickly can you understand the changing character of the battlefield, what your adversary is doing, what technology enables you to do, roll that into learning and development and fielding of new capability, operational and organizational change that comes from that? Recognizing that all of the gains that you might be able to eke out are themselves going to be fleeting. Whether that’s a matter of weeks or months or maybe a year, you’re going to have to keep that cycle incredibly tight and incredibly fast.

We’ve frankly missed an enormous opportunity as a country to really think about Ukraine as a catalyst for changing the way we operate. The Department of Defense, and the US Government in general, deals with things that it plans for, and it deals with things that it doesn’t plan for. For the latter, in my experience, it generally wants to say, “I didn’t see this coming. I don’t want to have to deal with this. I’d rather just get back to the regularly scheduled program.”

That’s how we’ve addressed Ukraine. We have provided a lot of support over a few years, but we’ve largely just grabbed things that were off the shelf — legacy programs of record, stockpiles of weapons, many of which have been very useful to them, but many of which have not been useful in the least.

We’ve really missed an opportunity to recognize that in supporting Ukraine, we have an opportunity to adopt that pace of change ourselves, to realize that this is the absolute frontier of warfare right now. If we can’t solve these problems for ourselves, let alone what they’re going to look like, God forbid, in a China scenario, which is all 10x worse, we are going to find ourselves on the back foot if we end up in a conflict like that.

There’s a lot more we could have done in terms of really looking at this as a proving ground for new technologies and new capabilities that could not just help the Ukrainians win, but help the US military change itself at scale with a lot more speed. Perhaps there’s still an opportunity to do that.

If, God forbid, we ever find ourselves in a conflict with China, all of the assumptions of Ukraine apply. It’s not going to be short. In all likelihood, we are going to lose a lot of things, we are going to have to replace a lot of things, and we are going to have to change and adapt at a speed that we’re really not used to as a country.

The Problem of Scale

Jordan Schneider: Let’s talk about scale for a second — another big lesson of Ukraine. We had a guest on the show a few weeks ago who left the CIA. I asked him, “What’s your favorite document?” He goes, “Well, this thing’s public. Each Chinese province has a war scaling exercise where they get all the companies in the room and work out, ‘Okay, how are we going to build 100x of this, 100x of that?’” America did that until 1947 and basically stopped. It was very striking to me. Another Brian Schimpf quote about how this was something that you guys are explicitly planning when you think about building things is finding partners that could go 10x or 100x. Reflections on that? Is this something that other people have going through their heads?

Chris Brose: I can’t speak for other people. The way we think about it is that the amount of capability that the United States is going to have to have for a peer fight is probably an order of magnitude larger than what we have become used to or accustomed to over the past 30 years. This isn’t a new insight. If you go back to World War II, for example, we won that by outproducing our adversaries. We were producing at a scale that was just eye-watering.

In the past 30 years, we’ve become enamored of the idea that we could basically build a military that is irreplaceable. It’s small, exquisite, and echnologically bleeding edge. And because we were far ahead of our competitors, if we ever had to fight, it was not going to last long. We weren’t going to lose much, we weren’t going to have to produce much or replace much. That’s just ahistorical in terms of what the experience of the United States at war has been since we were a country.

If the problem becomes, “All right, how do I 10x or 20x production of vehicles and weapons and not just do it once, but sustain it as a function of time — perhaps in mountains, perhaps in fields, perhaps on islands or at sea or in outer space — because I don’t want to limit options in the future,” then everything’s on the table. But the real question becomes, how do you actually design capabilities to be mass-produced?

If you look back at World War II, that’s what we were doing. You could take over the Willow Run automotive facility and go from making commercial cars and trucks to making B-24 bombers. Those bombers weren’t wildly different than the commercial vehicles that were being built on those lines prior. It’s impossible for me to believe that a Ford factory today in Michigan could build B-21 long-range strike bombers. Every aspect of it has just become completely different and divorced from commercial supply chains, manufacturing, etc.

But when we look at this problem, it is absolutely doable. When Tesla entered the automotive industry, they were laughed out of the room by Wall Street and by traditional automotive manufacturers because the production targets that they were putting on the board and the speed with which they were claiming to be able to do it had never been done before. Today if you look at one of those Tesla Gigafactories, they’re producing four to five thousand vehicles a week. Yet when you look at the production that passes for large scale in the defense industrial base, it’s a few hundred weapons a year, maybe a few thousand weapons a year, but it’s nowhere close to what commercial manufacturing is able to achieve.

The first premise needs to be, you need to leverage as much of the commercial industry as you possibly can. You need to design weapons and military capabilities to be mass-producible. If you don’t do that, there’s no way at all downstream that you’re going to solve that problem with more money or more feeling or anything else you might want to apply to it.

At the level of design, you have to simplify it. You have to make conscious decisions to lean on commercial manufacturers who have tons of capacity and, in the event of a war, could easily be flipped over from being 5% defense, 95% commercial to the other way around at the direction of the government. But if you’re not making those choices at the beginning, you’re never going to be able to produce at the scale that you want, and you’re never going to be able to sustain that over time.

Whether it’s autonomous fighter jets that we’re building for the Air Force or robotic undersea vehicles or weapons of all kinds, that design principle is at the core of everything that we’re doing. That’s why we’re going to be able to make good on the commitment we’ve made that we’ll be able to do 10x production of what typically passes for large-scale production in the defense industrial base.

Jordan Schneider: A provocation from Dan Wang’s new book Breakneck:

“AI can’t distract us from broader American deficiencies. As relations between the US and China become more hostile, the chances of a conflict grow. The US is facing a peer competitor that has four times its population, an economy with considerable dynamic potential, and a manufacturing sector that can substantially out-produce itself and its allies. If China and the US ever come to blows, they would be entering a conflagration with different strengths.” Would you rather have software or hardware?

Chris Brose: It’s a false choice. I say this as a company that is trying to be leading edge in both. At the end of the day you’re going to need hardware. There’s just no two ways around it. As long as human beings exist in the real world and occupy physical space, hardware is going to continue to be relevant.

At the same time, what’s really changing is that previously we’d always faced this choice between you could have mass, but it was going to be dumb, or you could have precision, but you weren’t going to have quantities of it. The real opportunity now is that you can put those two things together, and you can do it at an affordable price. You can have low-cost weapons, low-cost drones that are also enabled by highly intelligent, highly capable software. You get both mass and precision at an affordable price.

Putting aside the US and China, what this does is it opens up the level of geopolitical competition to countries that previously would not have had access to that level of throw weight in the international environment because they had been limited by size of territory, size of population, size of economy. What’s happening now technologically, both in terms of software, AI and autonomy, as well as in low-cost manufacturing and other kinds of product production changes that are occurring, is the ability for a small country to actually punch enormously above its weight from the standpoint of military capability.

Jordan Schneider: We have the floor being raised — precise mass or mass precision, we haven’t decided yet. Lots more people can do crazy drone attacks on a waste treatment plant halfway across the world. But there’s also this other aspect of what Dan was getting at: you’re trying to make production and supply chains that can go 10 or 20x, but it is hard to compete from that perspective with the world’s factory, right? What’s going to have to give there?

Chris Brose: Unless we’re prepared to just say at the outset, “We’ve really screwed this up for the past 40 years and we might as well surrender,” the question becomes: okay, you have to accept the reality of where you are, which people are starting to do. But it still seems the case that people haven’t wrestled fully yet with the implications of what 40 years of essentially US outsourcing and deindustrialization at a time of Chinese hyper-industrialization has really brought us to.

The shipbuilding example is a good one that’s often used — 220 times the capacity of shipbuilding in China relative to the United States. If we get into a conflict and we start losing ships, this is going to be a really hard problem for us to solve.

The answer isn’t to go back in time and try to recreate a bunch of old industrial muscle that we’ve lost. The answer is what we’re trying to do, what a lot of other new companies are trying to do, which is try to create new and different muscles that can actually be put in place much faster.

As I look at where we are as a company, we have millions of square feet of production capacity that are coming online all over the United States, in allied countries. It’s happened in a matter of a couple of years really. The ability to scale that as we become a larger company, as other companies become larger — it’s just that those factories are going to have to be building different things. They’re not going to be building aircraft carriers and destroyers and long-range strike bombers. They’re going to be building autonomous fighter jets and robotic submarines and low-cost cruise missiles and drones and other things that can be built quickly, where you have a workforce that is skilled enough to be able to mass-produce those kinds of things in the United States.

We are absolutely in a bad way from a hardware and manufacturing capacity standpoint. We have given up an enormous advantage in this country, and China has absolutely seized that. We are where we are. That being said, I don’t for a moment believe that all hope is lost. Going back to your point, if we are serious, there is an ability to create new production capacity in this country to mass-produce the kinds of military systems that we’re going to need to generate and sustain deterrence. If we end up in a conflict, God forbid, to be able to fight it, sustain it, and win it over time.

Jordan Schneider: Coming up to the strategic level, another way to answer China’s ability to manufacture is, as you alluded to, getting the rest of the world on board as well. You wrote pretty compellingly that on the one hand it would be nicer for allies to do more and spend more money, but let’s not forget that they are actually allies. It’s been a weird few months. We’ve seen Europe say they want to spend more, but also say they want to have their own homegrown stuff. Take this wherever you want to.

Chris Brose: There’s been enormous change in recent months — tariffs, political shifts, and upheaval of existing practices. But the sky isn’t falling.

We’ve made a clear break with the past, signaling to business, government, and allies that we’re moving away from the old order. Much of that arrangement wasn’t great anyway. The question is: what comes next?

Look at European commitments to increase defense spending, which are considerable. For years, we dutifully read talking points to NATO allies that they needed to spend more. Some did, most didn’t. Now they’re actually doing it seriously, and the administration deserves credit.

Unsurprisingly, with all this uncertainty, European governments want that investment creating sovereign, homegrown capability — employing their people and ensuring they’re not dependent on others. With the US speaking similar words, it’s not surprising to hear this from Europe too.

Share

This still creates opportunity for US companies. If we can’t just produce in American factories and ship to Europe — because it’s politically problematic for our allies — there are other ways. We’re collaborating with major European defense companies like Rheinmetall and setting up indigenous operations in allied countries, genuinely becoming part of their industrial base. Look at Rolls-Royce or BAE. They’re British companies, but as much part of America’s industrial base as Boeing or Lockheed Martin.

The US has a great advantage in our dense network of allies bound by common interests and values. The question that’s always bedeviled us is how to operationalize that at scale, how to multiply the power of all these countries to generate offsetting mass to China, for example.

That’s imminently possible right now with all this churn and acceptance that we’re not going back to the past. We’re building something new. I’m eagerly awaiting government action to chart how we collaborate with main allies in Asia and Europe. It’ll look different than historically. That’s okay. There’s enormous opportunity for those thinking creatively and moving aggressively when people are serious about solving problems.

Jordan Schneider: Speaking of sovereign capability and things looking different than they did before, how do you feel about nationalizing primes and putting the Defense Department on the Anduril cap table?

Chris Brose: I understand why the government is considering equity stakes in defense contractors. For years, a handful of hyper-consolidated companies have relied entirely on government contracts while delivering late, over-budget programs. The government owns all the downside without real control over performance. The market has become such that you can’t even call it a market. It’s hyper-consolidated with too-big-to-fail players that the government doesn’t really have options.

For the government to come in and say, “If you want me to build you another facility to produce more, right now I want a piece. I want to be at your board, at your table, with a real voice for how you’re running the company, how you’re using profits, how you’re structured” — I totally get why this is a conversation we’re having now.

But this would be catastrophically bad right now. Unlike Intel, which serves commercial markets, defense contractors operate in a monopsonistic environment where the government is the only customer. For the government to award contracts to companies it partially owns creates massive conflicts of interest. It’s the same reason defense secretaries must divest holdings in defense companies.

In an environment where you effectively had hyper-consolidated large defense companies you could count on one hand, it would make sense for the government to say, “In order to compel performance, in order to keep programs on cost, on schedule, I want to have more agency over how these companies are functioning.” I get that.

But the timing makes this doubly ironic. After a generation of walking away, we finally have an explosion of new defense tech companies. Dozens are being funded by billions in private capital, building innovative capabilities, competing against the legacy players. Taking ownership stakes in the largest incumbents just as this competitive renaissance emerges would undermine the very dynamism we need.

In a market with only a handful of players, government ownership might make sense to compel performance. But when energy, money, and talent are rushing back into defense to create real competition, this is deeply problematic.

Autonomy and Arms Control

Jordan Schneider: The future of command and control — what won’t humans be doing? What will humans still be doing, maybe five, ten, twenty years down the road?

Chris Brose: First, we need to separate command and control. They get bundled together and are often referred to as C2, but command and control are very different things. In a human setting, commanders who are providing their intent — their overall objectives and guidance — to subordinate agents, whether those are people or, in the future, could be robotic systems. Those commanders are not telling them every single thing to do.

They may trust certain subordinates more and give them a wider berth or more flexibility in what they’re able to do. But those commanders are providing the left and right limits under which the forces that report to them are making decisions and operating. Control is executed more at that subordinate level. There are instances where a commander will reach down and directly control an outcome because it’s important for that individual or for the mission. But the way the US military functions is very much around delegation of authority to the lowest level. It’s what makes us different than the Chinese military or the Russian military, and it’s a massive superpower.

This delegation of authority is the whole basis of command and control. This is why I don’t think it’s crazy at all to take that exact framework and apply it to how human beings in the future will be interacting with robotic systems or autonomous systems. You still have a human commander in charge who is fundamentally making a couple of big decisions.

One is determining whether some object on the battlefield is a legitimate military target. That is something I think we want a human being to decide. The decision may be enabled by a lot of technology like sensors and machine learning but at the end of the day, a human being needs to say yes, that is a legitimate military target that I want to do something about.

The follow-on action of doing something about it — the controlling or initiating an act of violence — is something that we’re going to want a human being to at least say, “This is a decision that I want to take.” My own view is that, beyond that, most of this can actually be automated. It can be a set of practices or robotic systems where human beings can still execute commands but be more reliant upon autonomous systems or processes to engage in the control of the decisions they’re setting themselves.

Jordan Schneider: Waymo, right? Everyone has this vision. The pitch they’re making is, “Look, this is better than 99.9% of drivers, or it’ll get there soon. You should trust us instead of driving.” There is a long and storied history of targeting decisions gone awry. Why do you think there should be human beings deciding which, I don’t know, Venezuelan boats to blow up?

Chris Brose: A human being will make the final decision. It’s a classification decision enabled by technology, but determining civilian versus legitimate military target — I’m not sure we’re ready to hand that off to robots.

But I would tell you that most of these decisions, all of these decisions, are highly contextual and circumstantial. I find the debate often becomes this very monolithic debate of, in the abstract, what are we willing to delegate to robots.

This conversation looks very different for offense versus defense. In defensive operations — protecting against inbound enemy drones, weapons, missiles threatening a ship, base, or city. You’re going to delegate a lot more because the speed at which you have to make that decision is much faster. The requirements are much faster, and the consequences of being wrong are much higher in terms of actual loss of life to your own forces or population.

For offense, for force projection — going out to find and destroy enemy targets — we’ll be more cautious about what we delegate to autonomous systems without human-in-the-loop supervision.

Another example — the way the US military operates is not that we’re going to go fight wars that are endless in time and space. We are going instead to identify very specific geographic locations that we are going to define as areas of active hostility. This is US military doctrine. It’s the law. Inside these carefully drawn zones, commanders relax thresholds for using force due to mission and force protection risks. This doctrinal framework can absolutely be adopted for autonomous systems. It’s not loose killer robots everywhere for all time. It’s communicating to the world: “Don’t go here because I’m treating this as a battlefield,” while giving our forces more leeway to use force they might not employ elsewhere because the mission must succeed and you must protect the humans operating there, military and civilian.

Jordan Schneider: It seems there’s a lot of competitive pressure in delegating more. Maybe the tech isn’t ready, and this is a 2040 or 2050 thing. But when you have autonomous systems that can escalate situations with strategic consequences you didn’t intend, that’s probably where you still want human oversight.

Chris Brose: That pressure is there. If militaries see an operational advantage to be gained by delegating more to autonomous systems, by removing human control that slows processes down, if it enables them to operate at larger scales because you can now have a human-to-machine relationship or ratio that enables one human being to deploy or supervise lots of robots or weapons — countries are going to do that.

I would say China especially is going to do that because, to the extent that I understand it, their entire military structure is based on the belief that senior leaders do not trust their subordinates. They do not want empowered lower levels with guns running around with their own ideas and opinions.

It’s not a huge logical leap for me to believe that the Chinese military, as a matter of doctrine, is going to be far more willing to delegate these kinds of decisions to robots that they believe they can control, as opposed to human beings who may have a mind of their own — a group of folks who are armed who decide that they might not want to follow orders, or they might not like the regime, or they might want to take matters into their own hands.

Jordan Schneider: I’m not sure how you spent your Tuesday night, but let’s do China parade takes. I’ve got one from the peanut gallery — their CCAs seem much bigger, and some are tailless compared to the Fury.

Chris Brose: I’ve been too busy this week to read deeply into all the moment-by-moment takes of what’s been coming through. The Chinese Communist Party puts on a hell of a parade, hats off. It was super impressive. The marching is very high level. I did not have a chance to see every piece of the military system rolling through.

I’ll offer an opinion related to a question I have. The benefit of these things is that you get to showcase your hardware. Arms control and verification are built around this idea of “I can see things in the real world, I can observe them, I can count them.” This is how we did arms control with the Soviet Union. How many warheads do you have? We’re doing open skies observations of deployed nuclear forces, et cetera.

But how do you verify the level of autonomy of a robotic system? How do you verify what the software of that system enables the drone or robotic submarine to be able to do? That is a massive operational advantage for a country. Their collaborative combat aircraft are probably not going to look wildly different than ours. Their fighter jets certainly don’t because they have a very long and effective history of stealing those industrial designs from the United States.

If you see two drones that look more or less alike, one of them may have a level of software that is 10x better that enables it to do wildly more effective things in operations than the other one that looks identical to it. How do you verify that? How do you understand what its capabilities and limitations are? How do you begin to even think about an arms control regime for these types of capabilities?

We’re all better off not getting into an arms race dynamic. But the reality that people have to contend with is that this is an incredibly difficult set of technologies to observe, to verify, to understand what it’s capable of doing and what the military advantages that each respective country is gaining from those technologies are.

I’m realistic/cynical about the prospects for getting into some type of arms control regime on AI-enabled systems or autonomous systems for those reasons.

Jordan Schneider: That’s the lesson of the ’60s, ’70s, and ’80s as well. You had arms control, but it was on stuff that didn’t matter because both countries could still totally blow each other up. And by the way, you had SALT, and then SALT ended three years later and they tried to do it again. Everyone was still trying to make new and better tanks and bombers and fighters at the same time. I share your pessimism on this one.

Chris Brose: People spend a lot of time thinking about arms control for AI. I’m all for having that conversation. I want the advocates of it to start from a realist understanding of the world in which we’re living, to understand the consequences of getting into a one-sided arms control negotiation, of limiting ourselves. The reason arms control worked between the United States and the Soviet Union, such as it did, was because both sides had nuclear arms.

Jordan Schneider: They weren’t giving them up.

Chris Brose: They were limiting them, fundamentally. Your point is right. They still had nuclear arsenals that could destroy the planet multiple times over. They didn’t necessarily need to build two and three times more, but they would have had that dynamic constraint not been put in place. But the point was that constraint, to the extent it was effective, was effective because you were counting things you could count. You were measuring and observing things you could measure and observe.

I don’t know how you do that on software. We’re going to be reluctant to hand all of our source code over to the United States government if they didn’t pay for it. Why in the world are we going to do that to the Chinese so they can verify the level of autonomy that our collaborative combat aircraft has, so they know what its capabilities and limitations are? It’s just never going to happen.

It’s fine to have that conversation, but let’s actually have a debate from a position of facts and reality rather than hopes and dreams.

Jordan Schneider: On the other side of this, you have stuff like chemical weapons or blinding lasers and stuff like that. It just seems to me to be a very different category than this all-purpose technology which is going to go into everything.

Chris Brose: Also true. The other little inconvenient fact about chemical weapons is the reason nations engaged in arms limitations for chemical weapons after World War I is that they realized they didn’t work very well. They were played out. By 1917, it was shoot mustard gas at the trench over across from you, and the wind changes and it blows it all back into the faces of your own soldiers.

There was an incentive for these countries to limit weapons that they themselves had developed, had used, and found to be not as effective as they had thought. There’s a realism governing why they engaged in these kinds of arms control in the first place. But to be very clear, they built them, they used them, and only then did they engage in a limitation.

Jordan Schneider: Now it’s like cluster bombs are back, landmines are back. You don’t use the thing until...

Chris Brose: You don’t use those things until you realize that you need them. Ask the Ukrainians about landmines and cluster munitions or area destruction capabilities. They absolutely have a need for them. What is the easiest way to clear a minefield? It’s not driving robotic bulldozers through it. It’s hitting it with a cluster bomb.

For a long time we had this luxury. We took this vacation from world history and we thought that we had somehow gotten beyond all of the lessons that we had learned in previous wars. We thought that we could do without this or we would never need to produce at the scale that we did in the past. And here we are, back to the future.

Orban’s Cannons

Jordan Schneider: Speaking about lessons from world history, you mentioned you were reading 1453, an account of the Ottoman siege of Constantinople by Roger Crowley, which is very well written. What struck me was the story of Orban, who’s this mercenary gun master. He was from Hungary or Wallachia.

He first goes to the Byzantines but they’re too poor. So he goes across the river and hangs out with Mehmed II, who signs a big contract. He makes the biggest gun the world has ever seen. His cannon could shoot 700-pound rocks. It was able to knock down this wall that had been improved upon over the course of 2,000 years. But Mehmed the Conqueror says, “Keep shooting it, keep shooting it.” Orban’s like, “I see these cracks, maybe we shouldn’t.” And it eventually blows up and kills him. What lessons does this story have for the future of defense acquisition?

Chris Brose: Pay attention to the limitations of your technology. To take a step back: Why am I reading this book? I’ve been on this weird Middle Ages kick for the past couple of years because I find that we do, as a country, an incredibly terrible job teaching history to kids. I have a 12-year-old and a 15-year-old and it is just shocking how bad their historical education is.

I find myself needing to be a history teacher to my children. It’s frustrating because history is literally the written record of the most interesting stuff that’s ever happened. How can teachers make this boring?

My dad was a historian. I’m not a history major, I’m not a trained historian. I’m just interested.

Jordan Schneider: What was his period?

Chris Brose: It’s actually very germane to this conversation — he studied and wrote on changes in technology, specifically 19th and 20th century Europe with a focus on Germany. He actually wrote a very cool book called The Kaiser’s Army, which is about how the German military learned all of the wrong lessons from the Franco-Prussian and Austro-Prussian Wars. That really didn’t set them up well for the opening days, weeks, and months of World War I. You can find it on Amazon. It’s not a bestseller, but it’s a great read.

As for the Middle Ages, people see it as a “dark age.” You had the Greeks and Romans, and then things picked back up with the Enlightenment and the Renaissance. In this thousand-year period, some stuff happened. Wars, plagues, barbarians, the Crusades, and then the Renaissance brought us into modernity.

But I find this period fascinating. To answer your question, there’s a through line in the siege of Constantinople that has everything to do with technology and innovation, with Orban’s guns being a great example of it. The Ottomans and the Arabs before them had broken their teeth on the walls of Theodosius for years and centuries. The reason Constantinople was able to survive, despite the protracted state of decadence of the Byzantine Empire, was because they were surrounded by water on three sides and these impregnable walls on the other.

But Mehmet realized that this innovation could smash through those walls. Beyond that, Mehmet literally dragged his fleet out of the water, over the mountains, and brought it down into the Golden Horn — it’s amazing. I traveled to Istanbul many times in my last job with Senator McCain, and I regret that I only appreciate all of the history now in retrospect. I’m very eager to go back.

Jordan Schneider: There’s a really good military museum, actually. It has the chain that blocked the Golden Horn, it has the Orban cannons…

Chris Brose: The panoramic painting of the whole siege...

Jordan Schneider: I was there on my honeymoon. I told my wife, “Can I just get one day to look at the war stuff?”

Chris Brose: Did you hike the wall?

Jordan Schneider: Yeah, you can walk on the side inside. The amount of stuff is incredible. You have some Roman stuff, you got the Byzantines. The mosques are spectacular. Hagia Sophia is overcrowded and a bit much. But the tier-two mosques, you have this experience of serenity of being one of ten people. Shehzade Mosque took the cake for me.

Chris Brose: It’s so fascinating. Everybody thinks, well, in the year 476 the Roman Empire just disappeared. No, it actually went on for another thousand years in Constantinople.

Jordan Schneider: More cannon lore from Crowley— “The psychological effects of the artillery bombardment on the defenders were even more severe than its material consequences. The noise and vibration of the mass guns, clouds of smoke and shattering impact on stone dismayed seasoned defenders. To the civilian population, it seemed a glimpse of the coming apocalypse.” Which one of your platforms is most likely to give us a glimpse of the coming apocalypse?

Chris Brose: I don’t think that we will be giving anybody a glimpse of the apocalypse. We’re certainly not building anything on the order of Orban.

Jordan Schneider: You’ve got to aim higher, man. I’m a little disappointed in that answer.

Chris Brose: But to the question of striking fear into the hearts of our adversaries, the work that we’re doing undersea — we’re building a large diameter and an extra-large diameter autonomous undersea vehicle. We have a phenomenal partner and program in Australia working with the Australian Navy on that. That is an incredibly cool and deeply scary system.

It’s electric-powered, it’s super silent. It can show up in places that are a long, long way away from where it went into the water. It’s wildly accurate, carries all kinds of very interesting cash and prizes. This is the kind of offsetting advantage that I would hope gives our adversaries pause. Every morning, they’d wake up and be thinking, “This isn’t the kind of day that I want to make a run at a US ally or partner or US interest or something else that we value.”

When AI is a Double Agent

Jordan Schneider: All right, I have a weird one — we’ve talked about mass precision. We have precise mass. I have this vision of mass intimacy. AI is coming. We talk to it all the time. And it is growing into less of just like Microsoft Word and more of a companion to lots and lots of people. The espionage and social engineering power that you could have by controlling a decent percentage of a nation state’s therapists, spouses, or best friends just seems wild and just as scary as a secret silent submarine.

Chris Brose: Absolutely. These are the kinds of things that sound like science fiction. Let’s say hypothetically that an adversary of the United States stole an enormous amount of classified information about people in the United States government who hold security clearances. They have a deep knowledge of who people are. Let’s say they have an active cyber campaign and operation to gather more of that information. They can build an AI agent that is capable of identifying and making contact with a person and, over a period of time, becoming intimate with them such that that person begins divulging information about themselves that is compromising.

This isn’t crazy at all. I would be surprised if it isn’t happening. These are incredibly precise, very strategically important, highly consequential types of operations that previously would never have really been possible. This is here now.

To your point about our increasingly intimate relationship with AI and AI agents, it’s only going to become more that way. The nefarious weaponization that is possible with those types of things — I don’t even think we’ve scratched the surface in terms of the level of creativity that people are going to put against this.

Jordan Schneider: Would I rather have an SF-86 or someone’s ChatGPT logs? ChatGPT logs every time. The SF-86 is a point in time. It’s everything you are willing to tell the government. But whatever problems you have in your life, in real-time, you are engaging with technology with those things nowadays. And by the way, 30% of the US AI companion market is currently owned by Chinese companies. We’ll see how that one plays out. If you’re worried, if you think TikTok is sketchy, there’s a bigger worry than just getting the assistant secretary of whatever’s stuff. You can do societal-level things.

Chris Brose: It’s absolutely “both and.” People have become more familiar with threats posed by platforms like TikTok or deep Chinese involvement in American technology and access to our daily data. When you consider things most Americans don’t think about — how this changes intelligence gathering, cultivating assets, and compromise — it’s an incredibly powerful tool.

The amount of information we put into the digital environment and our willingness to become intimate with systems that understand us deeply, that know how to exploit things we’d only share in intimate relationships — that’s all pretty terrifying. They know what to look for.

The Courage to Be Serious

Jordan Schneider: What’s the book that needs to be written? What would the Kill Chain sequel look like? New chapters?

Chris Brose: Writing Kill Chain was uniformly awful. I was working full-time at Anduril with a wife and two kids, writing in my free time on an aggressive ten-month schedule. It was just pain and suffering. I’m fortunate I don’t have to write for a living. I’m not currently working on another book and am probably looking to keep it that way.

What’s the book that needs to be written? There are plenty, but we’ve said all these things already. We know the problems, we know what’s wrong, we have an increasingly clear view of the threat. Washington gets caught up in process questions and mistakes changing a process for better outcomes. JCIDS was consigned to history and we’re better off for it, but nobody should think eliminating a process inherently gets better outcomes.

Jordan Schneider: Things still need to be joint somehow, right?

Chris Brose: I’d submit that we have absolutely everything needed to do what we say we need to do. All the authorities, plenty of money, phenomenal people, world-leading technology. There’s nothing missing. There’s no process, budgetary, or other roadblock standing in our way.

We’re fundamentally limited only by our imagination, will, and seriousness to conceive new capabilities and ways of organizing. It’s just a matter of senior leaders making their institutions do new things rapidly. There’s nothing standing in the way of our doing that.

It’s not to say more books, articles, and podcast don’t need to be published. It’s important to keep people focused on this. But too often, people want to believe that unless we reform procurement or change budgets or fix DoD funding, we can’t change. That’s almost a cop-out. It’s scarier when people realize they have everything needed to do what they say is important. Which begs the question: what are you going to do?

Jordan Schneider: I feel like not a lot of people would agree with you. Why do you believe that?

Chris Brose: I’ve participated in lots of reform, and I know a lot of the authorities that exist. In the Senate, we produced hundreds of pages of acquisition reform. I’m not sure how much was even needed.

When you look at the massive flexibility that exists for the government to acquire what it wants, it often comes down to political will. You could use that authority, but it’s risky, it’s hard. Someone won’t like it. Someone might protest. That doesn’t mean you can’t do it.

Too much of our defense world relies on a priesthood claiming magical knowledge of procurement. Senior leaders have disempowered themselves. A lawyer says you can’t do something. Then you ask them to show you the statute that says you can’t. If they can’t show it to me, I assume I can.

We have enormous authority to create new programs, field new technology, and scale it up. We obviously have talent and technology. We’re spending nearly a trillion dollars on defense annually. You should be able to build a damn good military for that.

I’m not sitting here saying we need to cancel all the traditional programs. We need both, the traditional ones and new, non-traditional, low-cost, mass-producible systems. We have the budget for all of it. If we don’t do this and keep pouring money into old programs, we’ll spend everything without significant increases in capacity or capability. And we will only put ourselves at greater risk in the future.

McCain Memories

Jordan Schneider: The reason things are not happening might come down to this very moving end of The Kill Chain, where you talked about John McCain, who you worked for for a long time, passing, and how dejected you felt.

“The same pessimism occupied me on my third occasion when my emotions about McCain overwhelmed me. It was an overcast and unseasonably cold October morning in Annapolis, Maryland, where McCain’s final resting place lies in a small cemetery on the coast of Chesapeake Bay at the US Naval Academy. It was the first time I had been back to his grave since his death. And it did not take long for all those old emotions and feelings of gratitude to come rushing back. But what was different this time was the overwhelming sense of sadness at the inescapable realization that things in Washington had not gotten any better since McCain’s passing. Indeed, they had gotten worse. Significantly, inexplicably, undeniably worse.”

Chris Brose: That was what I felt at the time. To put some context around it — when McCain passed, I had this hope, primarily when we were sitting at the National Cathedral at his funeral, that maybe this would help to galvanize a critical mass of people to say, “Hey, this guy had worked for literally decades to try to make these processes better, to try to make the DoD better, to make programs perform better, ultimately to get our warfighters the capabilities they need to deter conflict and win.”

I hoped maybe his passing would galvanize people to get their act together—stop fighting over budgets, continuing resolutions, government shutdown threats. All these costly, time-wasting distractions from arming our warfighters.

Part of me knew that wouldn’t happen — politics builds up cynicism. When I visited his grave a year later, I realized that moment of unity, reflection on his legacy, hadn’t led us to be more serious about hard decisions. When I went back up and visited his gravesite a year or so later, I realized, “Man, there was this hope, this moment that maybe we could have had where everyone was united, everyone realized and was reflecting on this guy’s legacy and what he did.” He wasn’t perfect, but he was way better than the rest. I hoped his memory would lead us to be more serious and make hard decisions.

The thing that I get asked about a lot with McCain, what was it like working for him? What was different about him? There are plenty of smart people in Washington. There are tons of decent and hard-working people. There aren’t a lot of courageous people. There aren’t a lot of people who are willing to take an action or a decision knowing that it is going to blow back on them, knowing that despite it being the right thing to do, they are going to suffer an immediate political or other type of consequence for their action. I don’t see that a lot.

That’s what these times call for. I hoped people would be more courageous about hard votes and decisions that let us function like a normal country. We’ve defined deviancy down so much. We’re just trying to get back to normal. Maybe we could have the courage to do that.

At the time that I was writing, the political uncertainty, volatility, and fighting were worse even than they had been when I left the Senate. Five years later, there are bright spots, but until we do more of what we say we need and stop doing unserious, detrimental things that tie our warfighters’ hands, it’s hard to believe we’ve reached a level of seriousness McCain would be happy with.

Jordan Schneider: For young listeners with little McCain memory, what else should they know?

Chris Brose: Watch his concession speech. When people didn’t want him to concede, wanted to keep fighting, he gave an incredibly moving statement about unifying as a country to solve real problems. He said, “It’s time to move on. It’s time to unify as a country and go solve the real problems that matter. We’ve fought over our differences. Now it’s time to focus on governing and moving into the future.”

The thing that was always cool was that he would do all the meetings, he would do all the work, but he always left time to do something fun and interesting and generally culturally and historically significant. On the many trips we made to Istanbul, we saw a lot of amazing history in that city. We went to Mongolia, had a long meeting with the president of Mongolia, who then invited us to go fishing out in the Mongolian steppe. We sailed down the river, paddled down the river with his security detail, and caught fish.

I’ve sat in a lot of meetings in my time in government. Meetings are meetings, and whether you’re working for this principal or that principal, they’re not wildly different. It’s those experiences that McCain always sought out — to do something interesting, to see something significant, to go to a place that he’d read about and see in person. Places that I know I’ll never go back to. I feel incredibly fortunate that I got to do that with a person like that.

Jordan Schneider: Last one for you. What’s the more ambitious version of ChinaTalk? What can I be doing better or bigger?

Chris Brose: What you’re doing is important. For those of us who are not as expert in China, who don’t spend as much time thinking about it, it’s a phenomenal service and resource with the diversity of guests you bring on and the focus and deep level of insight you provide. What’s bigger and better than that? I don’t know. You could get a primetime talk show. You could expand from China to Russia. I don’t think these are things that you necessarily want to do, but what you’re doing is terrific. Honestly, I wouldn’t mess with it because it’s working.

Mood Music:

As China prepares to select taikonauts for its first-ever manned moon landing, a new space race is quietly taking shape — not just over who will next set foot on lunar soil, but over who will shape the rules and norms governing humanity’s expansion into the cosmos.

What happens when 1960s space treaties meet 2020s tech? And how will Beijing and Washington compete to define space’s legal frontier? Pseudonymous contributor Ari fills us in.

Ari is pursuing a master’s in Chinese Studies with a focus on China’s international-relations strategy. Ari graduated with honors from Harvard with degrees in social studies and environmental science, studying in particular the geopolitics of energy and critical minerals in Latin America.

The Outer Space Treaty of 1967 was negotiated long before the technology to mine celestial bodies even existed. But our ability to extract celestial resources has grown exponentially since then, and the tech for a wide range of space resource activities — from asteroid mining and satellite telecommunication, to defense and space tourism — is now just decades, if not years, away from commercial-scale deployment. International law has not caught up to these developments, and negotiations to update space treaties in the United Nations have languished in the face of sticking points over dual civilian-military uses of outer-space exploration. In the absence of clear international guardrails, space-faring nations and private actors are rushing to develop the capabilities to mine lunar regolith and secure their access to valuable celestial resources.

Meanwhile, China has experienced a stunning transformation and become a space-faring nation over the past six decades. In the 1960s, China, still embroiled in the Cultural Revolution, was technologically inept, critically underdeveloped, and seemingly destined to watch from the sidelines as the United States and then-USSR battled to launch astronauts into orbit. Today, China has in most respects overtaken Russia as the United States’s chief rival in shaping international norms around conduct in the “final frontier” — and China’s increasing tech prowess has coincided with increasing assertiveness in influencing international norms and principles according to its interests.

Perhaps unsurprisingly, the United States and China have gone about shaping international norms differently. Their divergent approaches have profound implications, not only in the contest for global influence, but also for national defense and the renewable energy transition — that is, requiring bountiful quantities of rare minerals.

In short:

• The United States and China have both attempted to pass legislation and establish norms around outer space exploration and use within the United Nations.

• When this has failed, though, the United States has continued its law-based approach, pursuing norm-building agreements and legal partnerships outside the UN system.

• In contrast, China’s approach, when faced with UN setbacks, has shifted to pursuing project-based initiatives, including activities at the International Lunar Research Station.

Gaps in International Space Resource Governance

The space economy is valued at $630 billion globally, nearly doubling in size over the last decade, and is set to reach $1.8 trillion by 2035. Discoveries of water, helium-3, and rare minerals on the Moon and near-Earth asteroids have led to a surge in public and private interest regarding the mining and use of these resources.

Rights of appropriation and use are subject to significant debate in the corpus of law governing outer space. Article I of the Outer Space Treaty asserts that space is “the province of all mankind,” and all nations — regardless of developmental status — have the right to freely “explore” outer space and “use” its resources. (The drafters of the Outer Space Treaty, however, did not define the scope of the word “use.”) Article II stipulates that the Moon and other celestial bodies are “not subject to national appropriation by claim of sovereignty, by means of use or occupation, or by any other means.”

Article II’s “non-appropriation” clause can be interpreted in two different ways:

1. As a prohibition on a state’s appropriation of an entire celestial body — ie. claiming the Moon or another near-Earth body as its sovereign territory, or

2. As a prohibition on claiming ownership over only celestial resources, including those extracted from the body’s subsoil. (Evidence from early drafts of the Outer Space Treaty suggests that the treaty’s drafters intended this more expansive interpretation.)

And even if space-faring entities can legally own resources they have mined, there is uncertainty about how they can legally be used:

• In-situ mining entails using mining resources on the surface of celestial bodies asteroids to generate rocket propellants, energy, and life-support gasses necessary for lunar settlement and for propagating further space exploration.

• There is also growing interest — and technological potential — for ex-situ resource extraction, such as bringing water, minerals, and other resources to Earth for additional processing and commercialization.

How can these disputes be resolved?

International norms often develop through customary international law. CIL “consists of rules of law derived from [1] the consistent conduct of States [2] acting out of the belief that the law required them to act that way”; the second prong of CIL is called opinio juris. CIL is developed not through written treaties between states, but through state practice. For example, when the United States first sent astronauts to the Moon in 1969, Neil Armstrong returned with moonrocks that became the property of NASA. But in 1973, Nixon ordered fragments of the samples to be distributed to 135 foreign heads of state and all 50 states. Whether Nixon’s worldwide distribution of the moon fragments developed CIL or not depends on whether Nixon distributed the fragments because he felt legally obligated to do so.

Share

In this case, Nixon likely did not do so under opinio juris given the lack of historical precedent regarding property rights around space resources. Nevertheless, as cases of space-resource utilization become more prevalent, the paucity of clear international guardrails will both generate considerable uncertainty and present opportunities for space-faring actors to fill gaps in their stead.

The American Approach

The United States approaches outer-space norm-building primarily through promulgating domestic legislation and building multilateral voluntary codes of conduct.

Congress in 2015 passed the Spurring Private Aerospace Competitiveness and Entrepreneurship Act, or SPACE Act, which sought to address the growing liability facing private space actors. This law marked the first time any government legislated on the question of private companies’ legal right to space-resource ownership. Other countries have since followed suit. In 2017, Luxembourg passed the law “On the Exploration and Utilization of Space Resources,” which states that “space resources can be appropriated (Les ressources de l’espace sont susceptibles d’appropriation)” and also permits private corporations to explore and use space for commercial purposes. Japan, the United Arab Emirates, and Liechtenstein have also recently passed domestic space legislation guaranteeing property rights in space.

Meanwhile, the United States has leveraged its bilateral relationships to generate codes of conduct. In 2020, the United States launched the Artemis Accords, a series of multilateral agreements to build consensus and generate new international norms of conduct in outer space. The Accords seek to establish “a common vision via a practical set of [non-binding] principles” to govern the exploration and use of outer space outside of the UN system. The Accords emphasize the importance of reserving outer space for peaceful use cases, in addition to reinforcing the idea that “the extraction and utilization of space resources… complies with the Outer Space Treaty.” As of January 2025, 53 states are Artemis Accords signatories.

The Artemis Accords and domestic space legislation cannot themselves bind all states to shared rules and principles. Although the Artemis Accords are meant to shape state practice in theory, they are voluntary, and thus their influence on state activities in practice is yet to be determined. The Accords have also been criticized by key space-faring actors — such as Russia, Germany, and China — who are skeptical of attempts to act unilaterally to establish precedent over an issue of global concern. Nevertheless, although domestic laws alone cannot generate CIL, domestic space legislation may be replicated by other states hoping to foster a lucrative space industry — and these activities could generate norms that, over time, evolve into binding customs.

China’s Approach

While the United States is building customs outside the UN, China is generally committed to negotiating laws and establishing norms within the purview of formal international legal institutions. Even so, when conventional avenues for formal law-building are blocked, China isn’t opposed to operating within gray zones of international law to secure its access to critical space resources.

China has been a leader in outer-space negotiations at the United Nations, conducted mostly within the UN Office of Outer Space Affairs (UNOOSA). China is the largest voluntary contributor to UNOOSA, which allocates most of its funds to equip developing states with space data to mitigate and respond to natural disasters.

Subscribe now

In response to gaps in international outer space law, China and Russia jointly submitted drafts of the “Treaty on Prevention of the Placement of Weapons in Outer Space and of the Threat or Use of Force against Outer Space Objects” (PPWT) to the plenary session of the Conference on Disarmament (CD) in Geneva. The PPWT proposed new legal instruments and a multilateral conflict resolution mechanism to prevent the weaponization of outer space, but the proposal was blocked by the United States. The U.S. representative to the CD argued that the PPWT is “fundamentally flawed,” in that it does not explicitly prohibit the deployment of space-based weapons disguised as civilian commercial activities, nor does it restrict the development, testing, or stockpiling of Earth-based weapons that can shoot down targets in orbit. Critics also noted the lack of verification mechanisms to ensure compliance — a death sentence given Russia’s abysmal track record of (not) complying with past arms control agreements.

Another point of contrast: unlike the United States’s highly law-based approach to international consensus-building, China is shaping outer-space norms through project-based initiatives:

• In 2004, China launched its Chang’e Lunar Exploration Program 嫦娥工程. In 2013, China successfully landed a lunar rover on the Moon — the first state to visit the Moon in 30 years. Six years later, in a historic moment, China landed a rover on the far side of the Moon, returning two kilograms of lunar regolith to Earth. The Chang’e program is an early stage of China’s long-term project to solidify a permanent economic and military presence on the Moon.

• China has fostered bilateral agreements for project-based collaborations on the International Lunar Research Station. Thus far, the Chinese National Space Agency has over 170 such cooperation agreements, or MOUs, with more than 50 national space agencies and international organizations. The ILRS, on track to be finalized in 2028, will use lunar regolith to construct a base, mining ice and helium-3 to support permanent settlement.

China is not engaging in norms-based consensus building like the United States. There are no clear examples of China actively attempting to promulgate space norms internationally. China’s actions, though, will undoubtedly leave a significant footprint going forward — by being among the first to land boots on the Moon and leading project-based collaborations around lunar settlement, China will set the standard and lay the groundwork for other actors to follow.

China’s International Legal Approach in Context

What are the implications of these divergent strategies for shaping outer-space norms?

The United States is responsive to a blossoming private space sector seeking legal guarantees from their government to safeguard their capital investments. With avenues for providing those safeguards blocked at the international level, the United States has not hesitated to act unilaterally and leverage its web of alliances to develop norms toward peaceful, sustainable, and commercially viable uses of space — with or without the rest of the world on board. This strategy reflects the United States’s historical leadership in designing international law and institutions reflective of American interests, values, and free-market economic principles.

China’s outer-space strategy emerged from a different historical backdrop. China initially approached the international order as a “regime taker” in the post-Mao era, complying with laws and institutions shaped by European colonial powers. Since it acceded to the WTO in 2001, however, China has taken an increasingly assertive approach to international governance in alignment with its own values and interests. Yet, despite such increased assertiveness, the Chinese Communist Party remains at least nominally committed to promoting international decision-making within the UN system. China refers to this approach as “Upholding Multilateralism and the UN-centered International System” 维护以联合国为核心的国际体系.

China’s operations within the United Nations are strategically advantageous. The United Nations, in theory, has an equalizing effect on international law-making by providing all states with the opportunity to shape shared rules of conduct. As the self-proclaimed leader of the Third World, it’s unsurprising that China has refrained from leveraging consensus-building mechanisms outside of traditional multilateral institutions. Nevertheless, China may also uphold the UN system in service of realist aims; China has significant influence over UN decision-making as a member of the Security Council and as the world’s second-largest economy.

By virtue of China’s state-centered economic model, space innovation and commerce are highly regulated or outright owned by the state. Experts argue that China is not likely to promulgate domestic space legislation due to the risk of inadvertently restricting state ownership of valuable space resources and scientific data. China may promulgate a domestic regulatory regime in the coming decades — if economic factors and international trends toward widespread adoption of private property in space necessitate such measures.

At a fundamental level, China and the United States are realist actors working within loose international frameworks. Their strategies demonstrate not only their diverging visions for a future international order, but also diverging international imperatives. Without agreement on space-resource governance, outer space risks becoming controlled by a powerful few rather than benefiting all humanity.