Reading view

There are new articles available, click to refresh the page.

Japan and Trump

Why has Japan fallen out of Trump’s good graces? Will Japan close a deal with the US before tariffs take effect? And how will the upcoming Japanese election impact relations?

To find out, ChinaTalk interviewed Professor Tomohiko Taniguchi, a longtime observer of US-Japan relations and former advisor to the late Shinzo Abe.

We discuss…

  • Why 1970s trade competition is still impacting US-Japan relations today, and how Japan could create “Wow factor” when dealing with Donald Trump,

  • How Shinzo Abe used golf, dinner parties, and history lessons to cultivate a close personal friendship with Trump,

  • The roots of Japanese resolve in dealing with PRC aggression,

  • The emergence of Russian disinformation surrounding the Japanese election,

  • The political economy of the Japanese Self-Defence Force, and how Abe managed the controversy surrounding his reinterpretation of Article 9.

Co-hosting today is Charles Litchfield of the Atlantic Council. Thanks to the US-Japan Foundation for sponsoring this episode.

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


Golf and Good Graces

Jordan Schneider: We are recording this on Tuesday, July 15th, a few days before an election in Japan and in the midst of rising tariffs on the part of the US. Maybe let’s start with the present day and circle back in history to explore how we got there. Why don’t you begin by laying out the state of play with respect to the US-Japan relationship in the Trump 2.0 era?

Tomohiko Taniguchi: There was a much worse moment than today. If you recall what the situation was like in the 1970s, ’80s, ’90s, and up until the early 2000s, the trade frictions — usually dubbed a “war” between the United States and Japan — were much more tense and fierce. It was almost as if Japan was America’s primary adversary.

Compared to that, what’s happening at the moment is manageable, and one should not forget the bigger picture. Without the United States and Japan closely working together, you couldn’t possibly cope with this juggernaut in the making, which is in Japan’s — or for that matter America’s — neighborhood.

Jordan Schneider: All right, well, since you brought us there, let’s take it all the way back to the late ’70s and ’80s. You mentioned that it felt a lot more severe and dire at the time. How did that manifest in domestic Japanese politics?

Tomohiko Taniguchi: Right. It was in the Reagan era, but continued to become worse. The bottom was hit during the Clinton era, and Japan’s terms of trade were severely negatively affected by the engineered appreciation of the Japanese currency from something like 240 yen versus $1 to even 79 yen versus $1.

That actually set in motion Japan’s massive investment in Southeast Asian nations. Thailand, Malaysia, and Indonesia took off in no small part because of the massive inflow of Japanese capital. The end result is that you now see a great supply chain covering those nations. History is sometimes ironic. The attack, quote unquote, against Japan culminated in a boost for those Malaysia, Indonesia, and Southeast Asian nations’ economies.

That was then. Today, Japan, as many of your viewers and listeners understand, is the biggest ever investor in the United States, having created a significant number of jobs in, interestingly, battleground states such as Ohio, where Japanese auto manufacturer Honda has been operating for some time and has created many jobs.

However, that rhetoric has not worked vis-à-vis Donald Trump. President Trump wants to teach a lesson or two to the US’s long-term allies, including Japan, that nothing is free and you must pay your own club fee if you wish to maintain the order maintained not by China, but by the United States.

Charles Lichfield: Do you think that President Trump has a particular fixation with Japan? In the period we were mentioning just a few seconds ago, in the ’80s, he himself, as he became a popular figure on people’s TV screens quite often, would mention how Japan and Japanese exports to the US were causing jobs to be lost in the US. He certainly had a fixation then, but do you think that’s rubbed off on his policies now, or is Japan just one of many that he’s going after?

A Detroit man takes a sledgehammer to a 1978 Toyota in front of a Chinese restaurant in the early 1980s. Source.

Tomohiko Taniguchi: The late Prime Minister Abe, with whom I worked happily for almost eight years, played golf together with Donald Trump a number of times, not because both of these men were big golf fans, but because Shinzo Abe wanted to spend more time with the United States. President Trump in a way is like a broken record — at one point he says, “Yes, I understand,” and the next moment he comes back to where he was 10 minutes ago.

It’s very important to continue to have good conversations with Donald Trump. After all, there is no luxury for Japan to choose one from another when it comes to dealing with the President of the United States. The establishment of good rapport with whoever happens to be in the highest office in the United States makes very important sense for Japan. That’s the reason why Shinzo Abe worked extra hard to work with Donald Trump.

But I can assure you, had it not been the Donald Trump administration, but the Hillary Clinton administration, Shinzo Abe would have had as good a rapport with Mrs. Clinton as well.

Prime Minister Shinzo Abe and President Trump at the Trump National Golf Club, April 28, 2019. Source.

Charles Lichfield: Understood. But the way that Shinzo Abe dealt with Donald Trump in the first term is almost cited as an example by other countries as they were preparing to deal with him in the second term. Always arrive with a new investment announcement, and in this way, try to convince Donald Trump to avoid targeting Japan with additional tariffs. It worked in the first term. It isn’t working now.

Now, there are many countries being targeted with tariffs, so it’s not necessarily the failing of the Japanese government this time. But still, do you think that there’s something lacking in the Japanese government strategy without Shinzo Abe that was there last time? What’s missing?

Tomohiko Taniguchi: What’s missing is making Donald J. Trump himself say, “Wow.” You know, that big surprise is what’s lacking. One should not forget that bureaucratic channels do not necessarily work. The Japanese government seems to be pulling strings as much as possible from the bureaucratic institutions of the United States.

But you need to convince the topmost guy, Donald J. Trump, and make him say, “You did a great job. You’re awesome. I’m very surprised.” You know, that’s the message that not only Japan but many other countries have been unable to deliver.

Jordan Schneider: I do want to stay on this legacy of the 1980s for Trump because there are lots of quotes from memoirs where people ask him, “Why do you believe this stuff? Why are you so focused on it?” And he’s like, “I don’t know. It’s what I’ve thought for 30 years.” Gary Cohn was the one who said, “Well, President Trump, when I was 18, I thought I could be a professional baseball player. But sometimes things I think aren’t true."

It’s remarkable that today, a singular individual in America whose political consciousness was awakened by the trade tensions between the US and Japan in the 1980s is now reshaping the global economic order. I just think it’s worth reflecting on this weird historical moment where something that should be ancient history by now — this is like 40 years ago — is still the thing that is driving the most dramatic policy changes in Washington today.

Tomohiko Taniguchi: Well, I wouldn’t disagree with you. What’s important is for Donald Trump to remember the new Japan by remembering someone that he likes, before remembering what it was like in the 1980s.

If you look at Clint Eastwood’s Dirty Harry movies, seldom you see Japanese cars running in the streets of California. However, since that time, there has been an inflow of a massive number of Japanese cars, for sure. That’s the image — the Rising Sun, you know, the movie Rising Sun, where Japan was portrayed as a villain, as a very horrible, terrifying being. That image and perception is something that Japanese manufacturers and industries have had to struggle with for many years, state by state.

No wonder Japanese companies have been the biggest or the second biggest investor in the United States. But the truth of the matter is, however you cite those facts, it wouldn’t fly. It wouldn’t make much impression on Donald J. Trump.

Still, Donald J. Trump could remember someone with good memories. Try to think what impression he may have when he remembers Italy. He would probably remember that charming lady prime minister. That’s very important. When Donald J. Trump remembers Japan, he just remembers how terrible the trade relationship was in the 1980s, rather than remembering someone he cherishes — someone with whom he very much values the relationship.

It takes a lot of effort on the part of the leader of the nation to forge a good relationship. This is very much a personified relationship rather than a sovereign-to-sovereign relationship. I think that’s what’s important always to remember. In that sense, Shinzo Abe succeeded in bringing Donald Trump’s perceptions of Japan to the reality of the 21st century. But it took a huge amount of time and effort on the part of Shinzo Abe, which, as I said, was the reason why Abe played golf a number of times together with Trump.

Once I asked Abe, and he answered, “Yes, I’m certainly interested in playing golf together with someone who is an avid player, but playing golf gives you an ample amount of time alone by ourselves to speak with each other on whatever topics you could possibly speak about."

Jordan Schneider: Do you think he actually liked him or humored him? I mean, what is your sense of the internal Shinzo Abe monologue as he’s spending three hours with this very interesting character?

Charles Lichfield: Because many of us remember the photo from dinner at Mar-a-Lago and the look that Abe pulled wasn’t necessarily one of absolute comfort and relaxation. That’s why we’re intrigued.

Tomohiko Taniguchi: I see. There are many other photographs where Abe showed very much a relaxed posture. Abe showed a big, big smile, actually. I was reminded of what Shinzo Abe said to us delegates when Shinzo Abe just came back from Trump Tower, where Shinzo Abe and Donald Trump met for the first time, when Donald Trump was not even president — he was just president-elect.

The first thing that Shinzo Abe told us over dinner was the following: “I got surprised. Trump in private was very much different from Trump in public. He was calm and he was very much a good listener and he paid keen attention to what I had to say."

It was the beginning of a very good friendship — may I say genuine, very good friendship. Shinzo Abe was extremely grateful when Donald Trump in Japan granted a longer than planned time meeting with victims of abduction — many of those parents whose children were kidnapped by the state of North Korea. Unlike Donald Trump, who spent a minimum amount of time standing, Donald Trump sat together with those victims.

The other thing that Shinzo Abe got impressed by is that Donald Trump remembered each and every member of the employees of his Mar-a-Lago resort. Donald Trump possibly remembered hundreds of names — first names, of course. It’s interesting to observe that Donald Trump still seems to be unable to remember the name of the current Japanese Prime Minister.

President Trump and Prime Minister Abe have dinner at Mar-a-Lago, February 10, 2017. Source.

Charles Lichfield: The golf worked quite well. I’m not that surprised to hear that Donald Trump was a good listener when he had just been surprised himself to have won and was almost in need of fellow leaders who could explain a few things to him. He was basically looking for friends and people to talk to on the global stage, and I’m not necessarily sure that his attitude is the same now, but that’s not to say that Abe’s personal qualities didn’t play a role there.

But I suppose I do have a question about how you and his team prepared for every meeting. You mentioned the wow factor. It was always important to turn up with something that would make President Trump say, “Wow,” without wanting to talk about what’s going on now and the fact that this Prime Minister in Japan perhaps lacks a bit of a wow factor. I’m just curious about how you’d prepare a new impressive surprise for him. After all, there has been an investment deal this time around — the Nippon Steel investment has been cleared. But my impression is you’re saying that isn’t enough, and last time you were able to prepare better surprises. If you could perhaps describe to us the process that will go into preparing the next surprise before every meeting.

Jordan Schneider: Yeah, and I guess to complement that, this isn’t — it’s a very different mode for governments to prepare. It’s almost like you want someone who works with celebrities or is good at movie promotion or something. It’s almost like a different modality of thinking about foreign relations, no?

Tomohiko Taniguchi: What has not changed is this continued importance to spend more time by themselves one-on-one between the President of the United States and the leader of the nation that he’s meeting. You must spend a huge amount of time covering all sorts of issues. Prepared talking points prepared by bureaucracy do not actually mean much because by spending two hours, three hours, four hours together, there’s nothing that bureaucracy could cover in its entirety.

For instance, the conversation between the two leaders, Abe and Trump, went to cover some of the historical issues in the modern era in the Indo-Pacific — in Asia more precisely. To the surprise of many, Donald Trump once raised a question, “Is it true that Japan had a war with China in the 19th century?”

Shinzo Abe said, “Yes, Mr. President, there was a war between Qing Dynasty China and Japan.”

Donald Trump said, “Wow, you must have been defeated badly.”

“That’s not true, Mr. President. It was Japan that won the war.”

Mr. Trump’s response was, “Really?”

“Further than that, Mr. President, there was a war between Tsarist Russia and Japan.”

“Are you kidding? You must have been badly beaten.”

“That’s not the case. Japan actually won.”

“You are warriors!”

You know, that’s the conversation. Donald Trump really liked having those conversations, and Shinzo Abe did so.

However, Donald Trump is also very much a volatile individual. He once said “Shinzo, you always think that Japan is in a very much precarious environment, neighborhood, ‘Taiwan could fall in the hands of China anytime soon,’ you know, those are the things that you say. But when the war erupts, it is not you Japanese, it’s always us American boys and girls who shed blood, sweat, and tears. You look at us fighting with the Chinese on Sony TV screens, don’t you?"

Shinzo Abe had to answer with great care, but he was able to answer by saying, “Mr. President, in order for that not to happen, I did my homework by revising the interpretation of the Constitution so that when it comes to military confrontations in the neighborhood of Japan, now the Japanese armed forces and the US military can work and must work in a collective unit. In peacetime, it is the Japanese armed forces, self-defense forces, that give protection, Mr. President, 24/7, 365 days to the military assets, ships, and aircraft personnel of the United States."

Mr. Trump did not know that, and he was awed. “Is that right? Huh, Japan has changed.”

Constant reminding is necessary, and in order for this to be made possible, one has to have a deep, long conversation.

Charles Lichfield: That’s very interesting. You’ve just mentioned the wow factor towards the end — at least what Abe could say that he’d done himself. “I’ve done my homework, I’ve reinterpreted Article 9.” But what I hadn’t realized is that there was a second layer of wows that came from history and things that Japan had achieved long before that President Trump was happy to hear about. That’s very interesting indeed.

Jordan Schneider: Let’s not forget Singapore, Pearl Harbor... We've got a lot of these we can do.

Charles Lichfield: I think Donald Trump knows about Pearl Harbor.

Tomohiko Taniguchi: May I say that it seems Abe actually liked Donald Trump. Very straightforward and broad-minded if you build a good relationship with Donald Trump. When Shinzo Abe stepped down as prime minister due to his illness — this was September 2020 — there was a last phone call between these two individuals, Trump and Abe. It was so emotional that the interpreter went into crying himself. Trump said to Abe, “Shinzo, I will miss you.” That’s the tone that Donald Trump used.

Charles Lichfield: This is always a difficult topic, but I’m curious whether you knew anything about President Trump’s reaction when Prime Minister Abe, unfortunately, died when he was killed. How did President Trump react to that? By this time, President Trump wasn’t in office. This was when Biden was president.

Tomohiko Taniguchi: He has continued to cherish the friendship he had with Abe, the evidence of which was shown clearly when Melania and Donald Trump extended an invitation to the wife of Shinzo Abe. Akie is her name. Akie Abe went to dine with Mr. and Mrs. Trump. The photograph that was delivered afterwards showed the kind of relaxation that you couldn’t possibly imagine from Donald Trump in normal times. I think Donald Trump was utterly shocked by the death of his friend. Each and every time he still remembers Shinzo Abe, that memory comes back to him with very much a warm feeling.

Donald and Melania Trump host Akie Abe at Mar-a-Lago, December 2024. Source.

Jordan Schneider: I was making jokes about whether or not Shinzo Abe enjoyed interacting with Trump, but I think the way — I’m convinced, and maybe this is a transition to talk about the Shinzo Abe-Xi relationship, where I imagine the interactions were a lot less playful, perhaps is the word. What was his perspective coming into office about the relationship with China and how did it evolve over time?

Tomohiko Taniguchi: Well, on the first day when Abe met Trump, Abe had to say to the President-elect, “North Korea may be a threat, but it’s a manageable short-term threat. If you run good relationships with major powers around North Korea, like Russia, China, North Korea is a function that is very much dependent, not independent function. China is a long-term threat. First and foremost, one must remember that.” That’s one key message that Shinzo Abe tried very hard to convey to the new president-elect.

Shinzo Abe took office first in 2006, and that term ended only in a year. But 2006 was a momentous period. If you look back, it was in 2010 — 15 years ago — the Chinese GDP overtook Japan’s. Japan became the second biggest economy in the world back in 1968. Forty-two years after that, Japan was overtaken by China. It was writing on the wall. With that, the Chinese military buildup was awesome. Now China is probably five times as large as Japan, and if you look at the military balance in the Western Pacific or the Indo-Pacific, it is China that’s maintaining the biggest naval presence.

As Elbridge Colby mentioned several times, the United States could ill afford to provide Australia with nuclear-powered strategic submarines because the United States is short of military assets such as that in counterbalancing China. That’s the scenario that Shinzo Abe had. I think he knew very well that Japan could not cope with China alone by itself anyway. You need to build a wider network of allies and partners. First and foremost, you need to recalibrate the US-Japan alliance. You needed to reinvest into that. But in addition to that, you must bring Japan closer to allies or partners. The prime candidates for that matter were Australia and India.

Japan and Australia made a lot of effort in bringing Indo-Japanese and Australia-Japanese relationship to a level where you could call these partnerships alliance relationships. After building diplomatic capital and after implementing some of the legislative bills which eventually have made it possible for Japan to work with the United States as a collective seamless unit, then Xi Jinping’s attitude mellowed, if you like. It gives us a very good lesson that if you wish to have a manageable relationship with China, you need to show no weakness. You need to show strength.

I think Shinzo Abe did that by cementing Japan’s relationship with the United States, even under Donald Trump, and by forging, for instance, the Pacific-wide trade regime, CPTPP — Comprehensive and Progressive Trans-Pacific Partnership — even without the United States. The Quad was another setup: Indo-Australian, Japanese-American military network. Even Japan under Shinzo Abe made it possible for Japan and some of the select NATO member nations that share maritime orientations to work more closely together.

By the way, in a month, the UK’s new aircraft carrier, one of the new aircraft carriers, Prince of Wales and its strike group, will make a port call to Tokyo and Yokosuka. It’s a symbolic act. The two nations, the UK and Japan, are talking now about making it regular for those strike groups to come to Japan.

By the way, you said Donald Trump remembers Pearl Harbor. For sure, he does remember. But I’m curious if he remembers that shortly after Pearl Harbor, Japan’s military sank the newest battleship of the Royal Navy. The name was Prince of Wales. We have come a long way. Prince of Wales — now HMS, His Majesty’s Ship, Prince of Wales and its strike group and Japanese Navy boats will exhibit a massive joint drill, possibly joined by the Filipino Navy and Australian Navy. I’m not sure, but if that happens, that would be even greater.

The Tyranny of Proximity

Charles Lichfield: The early Abe era, in his first term, was before China overtook Japanese GDP, it was before Xi Jinping, it was before China restricted the export of critical minerals to Japan. A lot of aggressive behavior by China had not been exhibited yet. Nonetheless, Abe had already made a speech to India’s parliament in 2007, which I’m sure you influenced a little bit at the time, which I think suggested there was already some concern that the Pacific would be less navigable, less free. That’s when he developed the whole free and open Indo-Pacific concept. Why do you think so early, long before Donald Trump became president, why do you think he already had this concern about China?

Tomohiko Taniguchi: It was in the year 2008 when the Chinese authorities started to send their official ships to provoke Japanese territorial waters, particularly around the chain of islands Senkaku. When Shinzo Abe was in office 2006 and 2007, the Chinese started to make their physical presence felt acutely by surrounding nations. In the South China Sea, the Chinese started to develop their military presence, and whether or not there should be a code of conduct signed between China and the ASEAN nations had, until that time, been hotly discussed. Chinese ambition was quite clear.

If you look at the map, it’s obvious that without having presence either in the South China Sea or over Japanese Okinawa Island chain, the Chinese could not possibly gain access to blue water. It was also obvious that in order to fuel the growth of Chinese economy, copper, rare minerals from countries such as Chile, Peru were very important. In order to secure the free passage of the sea lane between South America and China, they started to look at some of those countries in the middle of the South Pacific, such as Fiji. Investment into Fiji had already begun, and it was not doubtful about the Chinese ambition.

I, for one, had a chance of a rare opportunity of speaking with the late Andrew Marshall, the Yoda of Pentagon, who gave consultations to a number of presidents back from Richard Nixon, who told us that in order for Japan and for that matter the United States to cope with this rising behemoth of China, it is vitally important for us to look at the maritime domain because China — Chinese coastline is relatively short. Japan has got the eighth largest exclusive economic zone, and Australia is a big island nation, so you must work together to give pressure from the seas to China. When Shinzo Abe spoke about confluence of the two seas, namely Indian Ocean and the Pacific Ocean, he was well versed as regards this geostrategic picture that was rapidly in the making.

Jordan Schneider: As last year’s Marshall Foundation Andrew Marshall scholar, I’m always happy to hear that his influence has spread across time and space. By the way, my glasses are also an Andrew Marshall homage. We’re trying to bring his fashion back for everyone on YouTube.

Charles Lichfield: I never heard “the Yoda of the Pentagon.” That was good.

Jordan Schneider: What has always been remarkable to me is that it’s not just elite politicians who have this view towards China even before 2010, 2011, the Senkaku disputes, rare earths, what have you. You still had Japan as one of the lowest countries in the world for polling favorability towards China. Back then it was around 20%, and now we’re down to below 10%, which is the second lowest in the world. Even the US is still up at around 35% for polling favorability towards China. It was already pretty dark back in the pre-Xi era, which is a remarkable thing. Can you help me interpret those numbers of why even in the 2000s, before you had all of this naval aggression, the Japanese populace at large was so skeptical of the PRC?

Tomohiko Taniguchi: Two things — long history and modern history. Modern in the sense that in the 1970s and ’80s, like the United States, Japan made rapprochement with the PRC in 1972. Since that time, China had done well in projecting a friendly image towards countries like Japan. Panda diplomacy worked well. In the 1970s and in the early 1980s, China was happy projecting a perception to the Japanese that it was a faithful student of making its economy modern and advanced.

One has to say, had it not been for Nippon Steel, which was then called Shin-Nippon Steel, the Chinese steelmaking industry would not have taken off that rapidly. As regards a lot of manufacturing know-how and industrial knowledge, Japan, the closest neighbor to China, played an important role. Many of the generations who worked to that end from Japan had this tremendous sense of guilt about what Japan did during the war. That was the happy honeymoon period that lasted about 10-15 years. Poll after poll taken at that time indicated that Japan had a very high trust level — even higher than the one Japan had vis-à-vis the United States at the time. Long gone are those periods.

Then some long, long memories kick in. By which I mean that dates back to even millennia. If you ask what makes Japan Japan, I would answer by saying that Japan is Japan because Japan is not China. What does it mean? Japan actually did borrow some important cultural traits, such as written characters. But in the Sino-centric suzerain tributary world order, Japan was a proud outlier. It was only between the 15th and 16th centuries — I must check my history book, but for about 70 years — Japan was happy being a dependent nation on China. The Chinese emperor did give recognition to the head of the warrior regime of Japan, but with that being rather an exception, Japan was not in the orbit of Chinese-centric civilization. That’s a point of pride for Japan.

In the modern era, since the 19th century, one of the sources of pride for Japan was that Japan was the only nation that evaded the fate of colonization from the Western powers. However, during the same time, China understandably developed a big amount of grudge and bad feelings toward Japan. After all, Japan was the only Asian nation that did tremendous harm on top of the humiliations that the Chinese had already suffered from Britain, France, and Russia. That’s the subtly and densely knitted emotional connections between the two.

These days, a daily phenomenon is that there is an influx of Chinese tourists, students, investors in Japan, and that’s been causing a daily wake-up call to the Japanese. One out of four registered foreigners are Chinese, mainland Chinese. 50% of foreign students working in name-brand universities are Chinese. Tokyo real estate property market has gone up remarkably, primarily because of the massive amount of investment flowing into Tokyo from Shanghai and other areas of China.

I call this phenomenon — remembering what it was called between Australia and Japan — a tyranny of proximity, not tyranny of distance. Tyranny of distance is something that Australians always used when describing the distance between their motherland, namely Britain and Australia, but China and Japan are very close together geographically. There’s a tyranny of proximity.

The current election, the upper house election, has turned to make this issue more prominent. For some foreigners living in Japan, it may sound a little bit scary because all of a sudden, the Japanese sound xenophobic. But it’s mainly not about Americans, not about Vietnamese, not about the Filipinos, but about the Chinese. This phenomenon is going to continue to be a daily concern.

In order for Japan to cope with this new phenomenon, first and foremost, you need to have political stability. You need to have clear-cut leadership. You need determined leadership who could say both to China and to the Japanese: to China, he or she must be able to say that “don’t underestimate the will of the Japanese, don’t ever underestimate the resolve of the Japanese,” but also to the Japanese people, “don’t panic, calm down, try to accommodate the Chinese in a balanced way.” Those dual messages are the ones that you need to hear from the Japanese leadership, but you can detect the tone with which I’m speaking of this issue, because I’m pessimistic — I’m not optimistic that this leadership would emerge anytime soon in Japan.

Charles Lichfield: Thank you. We want to discuss the election in a very short moment, but given we were discussing the awareness of China’s rise, public opinion in Japan having a very negative view of China, and yet at the same time, there was quite a lot of criticism of Prime Minister Abe when he tried to reinterpret Article 9, which in its interpretation before his Premiership made it very difficult for Japan to do anything other than self-defense. You mentioned that Donald Trump was impressed that Japan had done its homework, but not everyone in Japan was impressed by those changes. Yet, in public opinion, it does seem to be quite a unanimous view that is skeptical of China. Why did those two things happen at the same time?

Tomohiko Taniguchi: I remember the days when Japanese Prime Minister’s official residence was surrounded by thousands of protesters. Interestingly, however, there was a unique profile about those protesters. Many of them were Japanese baby boomers in their 70s even. They fought, if you like, one final fight against the incumbent power.

Charles Lichfield: Against the man.

Tomohiko Taniguchi: Against the man. The change that Abe introduced was so unprecedented and new that naturally I understand there was a considerable amount of hesitance and reluctance on the part of the public to change the modus operandi, the common picture. But on the other hand, there was a growing concern held widely about this growth of Chinese military. In that sense, many, even many of those protesters were bifurcated.

A protest outside Prime Minister Abe’s official residence, May 15, 2014. Source.

Now when 10 years — literally 10 years — have passed since the enactment of the bill that made it possible for the Japanese armed forces to work with the United States in a seamless manner, people in Japan have generally accepted the change. However, the real revision of the Constitution which Shinzo Abe earnestly wanted to achieve did not materialize. He wanted to add one small sentence to the renowned Article 9 by legitimizing the presence of the armed forces of Japan.

If I may remind, the Article 9 has two parts. One, renunciation of warfare. That’s good enough because the peoples in the world have made warfare illegal since the First World War. That’s not an issue. The second part of the article is still unique. It says, “To that end, Japan possesses no means of warfare.” That’s clearly a signature of Douglas MacArthur and the liberal-minded members of the occupation force.

But almost the next day, Korean War erupted, and Japan and the United States, for that matter, have had to make this white lie continuously. You get an awesome presence of Japanese self-defense forces. Japan is going to equip itself with as many as 400 F-35 most advanced fighter aircraft, Aegis destroyers. Japan has quite a few, and Japan finally is going to have a small but effective aircraft carrier being able to accommodate F-35 fighter aircraft. Are they not means of warfare? Certainly they are. The Japanese and, for that matter, Americans have had to make this white lie.

But it’s caused a negative impact on the members of the self-defense forces because they have been regarded as something illegal, even unconstitutional. Shinzo Abe wanted to add one clause only to legitimize the presence of the Japanese armed forces — to no avail. That’s something that Shinzo Abe was not able to achieve.

Charles Lichfield: It would be remiss of us not to talk about the present and the immediate future. In the next few days, there’s an election in Japan on Sunday. I believe only half of the upper house is up for re-election, so it’s a bit like the US Senate when not all the senators are running at the same time. But perhaps you could share your expectations. I hear there’s a debate — or I hear there isn’t a consensus — over whether the current coalition government will keep its majority. What is your prediction?

Tomohiko Taniguchi: The coalition government, the ruling coalition, is already a minority coalition in the lower house. In the Japanese parliamentary system, the upper house is much stronger than is the case with countries like India or the United Kingdom. It is almost certain, given the circumstances, the LDP Liberal Democratic Party and its coalition partner will lose their simple majority even in the upper house.

If that materializes, if that happens, nothing would be determined in a swift fashion. Everything is going to be determined in a keep-it-cool way, and much less transparency because more things have to be done behind the scenes between the party operatives from the ruling and the opposition. I’m very concerned about the loss of direction going forward.

Already over the last three years since Abe passed away, there is a sense of directionlessness. Many young generations once favored Abe administration left the ruling coalition and are jumping into newly formed parties, which are unknown presences.

One thing that I must say is that there is someone who would potentially benefit from this chaotic situation that is going to take place after the elections, and that someone is Russia. Russia is making a huge confusion into the SNS sphere by using bot applications. Those SNS entries that discredit the ruling coalition got amplified, and those favoring opposition parties equally got amplified because of this engineered bot situation.

Charles Lichfield: Is this receiving much coverage? Is this being covered much in the press? Because listeners will remember how much this came up in the 2016 campaign in the US, also in the Brexit referendum. I’m not surprised that Russia has been pointed out as the culprit, but how are the Japanese voters responding to this news?

Tomohiko Taniguchi: Well, frankly speaking, it’s been in the news only over the last couple of days. The question of whether or not the general public are duly concerned about this — I would say not so much at the moment. But finally, as you say, Japan has caught up with the United States and Britain for that matter. I’m very curious about the seeming lack of Chinese influence for that matter, because China would equally benefit from the confusion that Japan’s politics is going to have.

Charles Lichfield: They’re just relying on their partners to do it for them, perhaps.

Tomohiko Taniguchi: Maybe.

Charles Lichfield: That’s speculation. I did want to ask about some of the factors that are driving the election. Both Jordan and I are in the US and we have the tendency, as many will, to see everything through the lens of what’s going on here. I’m sure the tariffs have played a role, but I was amazed to see just how high inflation has been in Japan. The price of rice has really skyrocketed over the past few years. Independent of how the US is behaving towards Japan, there are many other things going on that will influence voters’ behavior. What’s driving this election?

Tomohiko Taniguchi: Tariffs have been important as a factor because it showed to the voters that the current incumbent administration, Prime Minister Ishiba’s administration, has failed to secure early deals with the United States.

Charles Lichfield: They’re not alone. Only two countries have.

Tomohiko Taniguchi: Yes, in my view, that’s exactly the reason why the Japanese public have not made these tariff issues as prominent, as important.

Inflation is an issue. Why? It’s indicative. It’s a quintessential Japanese phenomenon for that matter. This country has more elderly citizens than many other nations, and many of them are dependent on fixed income, namely pension. Pensions will not rise. Prices will. That’s made a lot of people nervous.

The younger generations, for that matter, are grossly discontented because they know that the government cares much more about the senior citizens. They are the ones that will continue to live well into the next century, the 22nd century. The 18-year-olds today who are eligible to vote will live another 80 years, perhaps. They are naturally frustrated by the fact that the current politicians — ruling and opposition both — talk much more about how to compensate the decrease of revenues for the elderly.

Inflation is painful, especially in a country that’s got fewer young people and more senior citizens.

Charles Lichfield: I’m just curious what you think will happen to Prime Minister Ishiba after Sunday.

Tomohiko Taniguchi: There are two interpretations. No one knows for sure which one is going to take place. He seems to be interested in sticking to where he is. After all, the government that’s going to emerge after the elections will be a coalition government. Since the ruling coalition lost the majority in the lower house, it’s been Prime Minister Ishiba who has somehow maintained a subtle balance between the Liberal Democratic Party and the opposition parties.

When it comes to the prime ministerial contenders within the LDP, not many would be interested in taking charge in such difficult circumstances. “Let Ishiba continue to bite bitter bills, bitter medicines” may be a prevalent feeling among the ruling party members. If that’s the case, if that will be the case, I would imagine political apathy would be deepened and people would get frustrated even further. Eventually the new government may have to be pushed into dissolving the lower house.

Chances are, even if the Prime Minister dissolves the lower house, the snap elections will result in even worse consequences for the LDP. Either way, you should anticipate that for the next couple of years, Japanese politics will be far from stable. Again, that’s going to be a fertile ground for the Chinese to make all kinds of provocations and even assaults. That’s what’s happening already.

Charles Lichfield: Cheeky question. What did Shinzo Abe think of Mr. Ishiba?

Tomohiko Taniguchi: Mr. Ishiba was known to be a designated outlier in the LDP establishment. Mr. Ishiba was a straight talker when it comes to making critiques about the incumbent, namely Shinzo Abe. I wouldn’t say that Abe and Ishiba were on good terms, even though Abe included Ishiba twice as a member of his own cabinet.

Jordan Schneider: You spent a long time working in media. I’m curious — there’s a lot of discourse in the US about the health or death of journalists and media outlets. What’s your take on the well-being of Japanese political coverage and international analysis in the press?

Tomohiko Taniguchi: Legacy media, print media, broadcasting media have lost their trust very rapidly. People spend much more time on what the SNS sphere says. Like in many other nations, many in Japan are in their bubble, reciting messages of similar kind all the time.

When it comes to international coverage, the lack of revenues for the legacy media have actually worked to the detriment. Many companies can no longer afford dispatching correspondents across the world. They have had to rely upon the purchased video clips, possibly from BBC and others. Japan’s been introducing the same bias you may or may not detect in those international media coverage.

What’s important for this is that there’s been a massive growth of anti-Israeli feeling, because each and every time if you turn on the TV or you look at YouTube entries, you get the messages created by Al Jazeera, no one else. Every time you see kids injured and carried to hospital beds. It’s a powerful, powerful message that shaped the views of ordinary Japanese about the Gaza situation.

Three years ago, if you asked “what’s your view about Israel,” 70% of the Japanese would say, “I have no idea.” The rest of the Japanese would have said, “Israel is an advanced, technology-oriented country.” But now, a lot of people have turned anti-Israeli. That’s primarily to do with the fact that almost all coverage is coming from Al Jazeera.

Jordan Schneider: That’s fascinating.

Charles Lichfield: I have noticed the output of NHK World has become a little bit more entertaining. They now have sumo wrestling matches with very good English language commentary. I advise listeners to go and have a look. I remember stumbling on NHK World about seven, eight years ago and finding it quite boring. It is better now.

Tomohiko Taniguchi: Only because they’re airing the sumo matches these days.

Charles Lichfield: That’s one considerable factor, but it’s not the only one.

Jordan Schneider: Say we get a new prime minister in the next few weeks. Is there one book, either Japanese or English, that you’d want to assign to whoever that person is going to be?

Tomohiko Taniguchi: Well, by 70%, it’s a fair chance that Ishiba will continue to stick to where he is despite massive frustrations and oppositions. One thing that makes me frustrated all the time is that unlike in the United States or in the United Kingdom for that matter, biographies, long sketches of some important individual, have not been in the tradition of political coverage of this country. Biographies are being written only after spending a huge amount of time after the death of the person of the subject.

Charles Lichfield: Whereas in the US there’s a nice tradition of autobiographies. Has Vice President Vance’s autobiography has made its way to Japan?

Tomohiko Taniguchi: Yes, Hillbilly Elegy was translated into Japanese very early, and many people began buying the book when Vice President Vance was chosen as Trump’s running mate.

Charles Lichfield: What’s the title in Japanese?

Tomohiko Taniguchi: It’s the same, written phonetically in katakana as ヒルビリー・エレジー (Hirubirī Erejī).

Jordan Schneider: That’s disappointing.

Mood Music

ChinaTalk is a reader-supported publication. To receive new posts and support our work, consider becoming a free or paid subscriber.

The Japanese cover of Hillbilly Elegy. Source.

Kimi

An anon start-up conducting cutting-edge open-source research on China’s science, technology, and industrial ecosystems is looking for part-time China research analysts. You’ll be saving America with a firm run by someone I [Jordan] can vouch for being the literal best in the business. Apply here.


Today’s post is brought to you by 80,000 Hours, a nonprofit that helps people find fulfilling careers that do good. 80,000 Hours — named for the average length of a career — has been doing in-depth research on AI issues for over a decade, producing reports on how the US and China can manage existential risk, scenarios for potential AI catastrophe, and examining the concrete steps you can take to help ensure AI development goes well.

Their research suggests that working to reduce risks from advanced AI could be one of the most impactful ways to make a positive difference in the world. They provide free resources to help you contribute, including:

  • A job board with hundreds of high-impact opportunities,

  • A podcast featuring deep conversations with experts like Carl Shulman and Ajeya Cotra.

  • Free, one-on-one career advising to help you find your path.

To learn more and access their research-backed career guides, visit 80000hours.org/ChinaTalk.

To read their report about AI coordination between the US and China, visit http://80000hours.org/chinatalkcoord.


A big welcome to , who joined ChinaTalk to cover AI full time this week after three years of nights and weekends contribution!

Beijing-based Moonshot AI (月之暗面, literally “dark side of the moon” - a Pink Floyd reference) released Kimi K2 on July 11. K2 is a non-reasoning, open source large language model based on the Mixture-of-Experts (MoE) technique and achieved benchmark scores competitive with many leading models, including DeepSeek V3. At 1 trillion parameters, it is an impressive feat. Per Nathan Lambert of Interconnects:

It is a "non-thinking" model with leading performance numbers in coding and related agentic tasks (earning it many comparisons to Claude 3.5 Sonnet), which means it doesn't generate a long reasoning chain before answering, but it was still trained extensively with reinforcement learning. It clearly outperforms DeepSeek V3 on a variety of benchmarks, including SWE-Bench, LiveCodeBench, AIME, or GPQA, and comes with a base model released as well. It is the new best-available open model by a clear margin.

ChinaTalk last covered Moonshot AI in March, when we translated an expansive interview CEO Yang Zhilin (杨植麟) gave to the online tech news platform Overseas Unicorn (海外独角兽). In the conversation, Yang portrayed himself and his company as stubborn AGI purists who focus on “tech visions” rather than product design or short-term revenue generation. K2 is a step towards many aspects of this vision, but its story so far also reflects the jagged reality of cutting-edge model research in China. In this piece we discuss:

  • What distinguishes Moonshot in China’s landscape — and what the DeepSeek and Kimi moments should tell Westerners about the future of Chinese AI labs;

  • How DeepSeek paved the way for K2, and why this is about open-source culture;

  • Why “the model is the agent” for Kimi;

  • And what we might expect next from Chinese AI startups.

The Moonshot AI office map, with meeting rooms named after classic rock bands. Image from Xiaohongshu/Rednote user @Pony今天早睡了吗.

Shooting for the Moon

Yang Zhilin, born in coastal Guangdong in 1992, earned his bachelor’s degree from Tsinghua University and went on to a PhD at Carnegie Mellon. He worked at Meta AI and Google Brain before returning to China to begin his entrepreneurship journey. Unlike fellow Guangdong native Liang Wenfeng (梁文锋), CEO of DeepSeek, Yang has deep connections in both China and the US and does not only focus on hiring domestically educated talent. While Tsinghua is heavily represented in the résumés of Moonshot’s founding team, others come from more diverse global educational backgrounds.

Moonshot has no B2B offerings and does not build wrapper tools for corporate users, instead focusing directly on individual customers. From the beginning, Kimi’s selling point to Chinese users was its long context window, allowing users to upload dozens of documents and analyze long articles. But it’s not just about an awesome user experience; long-context is central to Yang Zhilin’s AI worldview. Per his comments in the Overseas Unicorn interview:

To achieve AGI, long-context will be a crucial factor. Every problem is essentially a long-context problem — the evolution of architectures throughout history has fundamentally been about increasing effective context length. Recently, word2vec won the NeurIPS Test of Time award. Ten years ago, it predicted surrounding words using only a single word, meaning its context length was about 5. RNNs extended the effective context length to about 20, LSTMs increased it to several dozen, and transformers pushed it to several thousand. Now, we can reach hundreds of thousands.

Yang’s previous venture Recurrent AI (循环智能) was funded by seven venture capital firms, two of which also invested in Moonshot. Alibaba became Moonshot’s biggest backer in 2024. As Moonshot’s valuation rose rapidly, five of Recurrent AI’s investors — those who did not join Moonshot’s funding rounds — filed an arbitration case against Yang, alleging that Moonshot was launched without obtaining necessary waivers from previous investors. Recall that High Flyer, the parent company of DeepSeek, is a hedge fund, and that Liang Wenfeng has rejected outside investment as of March. Moonshot operates under much more normal tech startup restraints and faces investor pressure. Still, with just $1 billion raised (much less than the likes of Anthropic and OpenAI) and pressure to ultimately deliver value to shareholders, it created a leading open model, rather than operating in the hedge fund-funded cocoon that DeepSeek researchers enjoy.

What we are starting to observe here is rather obvious in hindsight, as we move farther away from the DeepSeek moment: there is no single path to success for Chinese frontier labs. One does not necessarily have to replicate the DeepSeek recipe, whether in terms of hiring, funding, or labor practices, to create world-class models. Compute constraints, which apply across China, continue to incentivize a diverse range of research teams in China to pursue novel algorithmic research.

From DeepSeek V3 to K2

The team behind Kimi is very active on Zhihu (知乎), China’s Quora equivalent. According to a post by engineer Liu Shaowei (刘少伟), K2 essentially copied the combination of Expert Parallelism (EP) and Data Parallelism (DP) outlined by DeepSeek in V3’s technical report, with four notable changes:

  • Raising the number of experts from 256 to 384, as their pretraining team found that scaling laws are valid for sparsity;

  • Reducing the number of attention heads to compensate for a higher number of experts;

  • Keeping only the first layer as dense and using MoE for all the rest, to maximize the benefits of MoE;

  • Keeping all experts in one group.

Image from Sebastian Raschka on Twitter.

Recall Yang Zhilin’s Overseas Unicorn interview, where he argued that “AI is essentially a pile of scaling laws laid on top of each other”. Raising the number of experts seems to reflect that. As for why they used DeepSeek’s architecture, Liu says there was no point reinventing the wheel:

Before starting to train K2, we conducted a large number of scaling experiments related to model architecture. The result was that none of the proposed architectures at the time were truly able to outperform DeepSeek V3. … The reason is simple: the V3 architecture has been validated and remains effective at large scale, whereas our “new architectures” haven’t yet undergone sufficient large-scale validation. Given the presence of two massive variables — Muon optimizer and a much larger model size — we didn’t want to introduce additional unproven variables just for the sake of “being new.”

Another Zhihu comment by fellow Moonshot engineer Su Jianlin (苏剑林) highlight other ways K2 learned from DeepSeek (italics added by editor):

Internally, we were also constantly exploring better alternatives to MLA [multi-latent attention, an architectural idea refined and scaled by DeepSeek], but since this was our first open-source large-scale model, we ultimately chose to pay tribute to DeepSeek by replicating its MLA design.

As for the MoE (Mixture of Experts) component, we adopted DeepSeek-V3’s shared expert, high sparsity, and loss-free load balancing. … A special note on the Sparsity part: for quite some time, we were running experiments with Sparsity = 8 (i.e., selecting 8 out of 64 experts). It wasn’t until we resolved some infrastructure issues one day that we began trying higher sparsity levels and found the gains to be significant. So we started exploring the Sparsity Scaling Law and gradually leaned into configurations similar to DeepSeek’s projections (DeepSeek-V3 already uses 256 choose 8, 32 Sparsity; while K2 uses 384 choose 8, 48 Sparsity). It felt like fulfilling a prophecy that the DeepSeek team had already made.

K2 would not exist without DeepSeek — and without an open-source culture and free flow of research. While Moonshot and DeepSeek are certainly competitors at the corporate level, Kimi’s engineers express deep respect for their DeepSeek colleagues. Yang Zhilin, in February 2024, had told Tencent News that he didn’t believe open source models could catch up to closed source any time soon, because “many open-source contributions may not have been validated through compute-intensive testing,” while closed-source projects “attract concentrated talent and capital.” Moreover, he remarked that if he had a leading model today, “open-sourcing it would most likely be unreasonable … it's usually the laggards who might do that—or they might open-source a small model just to stir things up.” So what changed between then and now? The success of DeepSeek was probably an important proof-of-concept for open source for fellow Chinese AI entrepreneurs. While Yang himself has yet to offer any updated comments on his open-source views, Moonshot engineer Justin Wong shared his “why open source” take on Zhihu:

First of all, we obviously wanted to gain some recognition. If K2 were just a closed-source service, it wouldn’t be getting nearly as much attention and discussion as it is now.

Next, open-sourcing allows us to leverage the power of the developer community to improve the technical ecosystem. Within 24 hours of our release, the community had already implemented K2 in MLX, with 4-bit quantization and more—things we truly don’t have the manpower to accomplish ourselves at this stage.

But more importantly: open-sourcing means holding ourselves to a higher technical standard, which in turn pushes us to build better models—aligned with our goal of AGI.

This might seem counterintuitive—if we’re just releasing model weights, why would that force the model to progress?

The logic is actually very simple: open source means performance comes first. You can no longer rely on superficial tricks or hacks to dazzle users. Anyone who gets the same weights should be able to easily reproduce your performance—only then is it truly valid.

With the success of DeepSeek, Chinese frontier labs now have ample, proven justification for the value of open source, both for marketing and in terms of research relevance.

What Makes K2 Agentic?

Moonshot calls K2 “open agentic intelligence”. It’s not just an OpenAI pun; the K2 team is trying to make their mark by refining definitions of agentic capabilities. Justin Wong frames the importance of agents as such:

A while ago, when agent products were all the rage, I saw a lot of comments saying that Kimi should stop competing in the arena of large models and pivot to making agent products. I just want to say that most such agent products are nothing without Claude. Windsurf being cut off from Claude further proved my point.

Fellow Moonshot engineer and Zhihu user Flood Sung created the following diagram to visualize K2’s agent vision:

Sung explains:

We first let the model generate hundreds of scenarios such as food delivery, Weibo, WeChat... Then based on the scenarios, we generate thousands of tools, such as ordering food delivery, sending Weibo posts, [and] finding contacts. Then based on different tool combinations, we generate hundreds of thousands of different agents (different system prompts + different tool sets). Next, we design tasks ranging from simple to complex with scoring points for each different agent. Then we begin large-scale agent simulation: first assign tasks to a simulated user, let the user interact with the corresponding agent based on the task, and the user is also generated with different personalities and language styles. Next, the tools called by the agent are also simulated through a tool simulator, which is actually equivalent to a world model to execute and return feedback to the agent for continued processing. After executing the entire task, we use the scoring points from the judgment criteria to evaluate whether the agent's interactions meet the requirements. Good data is then stored!

In other words, K2 was trained on massive amounts of synthetic data, generated specifically to resemble real-life applications and optimized for tool-calling.

What Next?

On the technical front, the Kimi team wants to expand on their current vision for agentic capabilities and build “large scale agentic RL.” Of course, they also want to move out of the shadow of DeepSeek’s infrastructure and eventually pioneer their own scalable approaches. We might expect a reasoning model based on K2 in the near future, as well as future foundational models.

Reading through Yang Zhilin and Moonshot engineers’ comments, one might get the impression that they are dismissive of wrappers and applications. But that’s not the whole truth. To understand their approach to real-life application, one has to take into account both the company’s AGI-centric worldview and their theory of change. Yang Zhilin, in his Overseas Unicorn interview, said that the secret ingredient to transformative AI was trust:

I think the ultimate measure of an AI product’s long-term value is how much personalized information users are willing to input into it, and then lossless long-context and personalization are responsible for turning these inputs into valuable outputs. New hardware forms may also be needed — but I think models and software are still bottlenecks. To dig deeper, the prerequisite for users to input a lot of information is trust — you need a sufficiently engaging and human-like AI. You can’t say, “I’m setting up product features specifically to get your information.” The end result should be that users and AI become friends, so users can tell the AI anything.

Justin Wong riffs on why this means moving away from the chatbot experience in his Zhihu comment:

Your interaction process with AI is not about directly outputting a piece of content, but rather understanding the user's needs and immediately starting a small project [or] delivering a frontend application. Users can continue to ask questions, modify, and iterate, but all of this revolves around a deliverable.

Sharp-eyed friends may have already noticed — isn't this just Cursor/Aider/OpenHands? Wrong. From an implementation perspective, this is indeed AI programming. But if designed cleverly in the product, you can hide the code-writing process. Then, for users who don't understand programming, this becomes 'I talked to AI, and it actually made me a PowerPoint/drew a flowchart/wrote a mini-game'. … This is a completely superior experience to traditional chatbots. … I believe Kimi products will soon have some surprising changes."

“AGI”, according to the team behind Kimi, has to feel like AGI. We should probably not expect Moonshot to put out wrapper products, especially since K2 is not multimodal, but the company certainly hopes to change how regular users experience AI usage in the near future.

What does K2 mean for competitors in the West? To quote Nathan of Interconnects again, Western labs like OpenAI, which has delayed release of its own open-weights model, are losing narrative control in open source.

We need leaders at the closed AI laboratories in the U.S. to rethink some of the long-term dynamics they're battling with R&D adoption. We need to mobilize funding for great, open science projects in the U.S. and Europe. Until then, this is what losing looks like if you want The West to be the long-term foundation of AI research and development.

K2 shows that DeepSeek wasn’t a wild exception to the rule or entirely a product of unique circumstances. Here is a startup with founders from different backgrounds, a more traditional VC funding structure, and different philosophy on AI who was also able to achieve a highly competitive open model, likely with strong uptake by users.

As for what’s next for Moonshot the company? Yang Zhilin gives few media interviews and eschews politics, and state media is yet to comment on K2. It would be a surprise if he immediately got the State Council treatment Liang Wenfeng did. Both Baidu and Tencent quickly integrated DeepSeek’s models into their products. K2 offers yet another cost-efficient option for big internet platforms looking for AI integration; its apparently strong creative-writing capacities might especially appeal to platforms like Weibo (similar to Twitter/X) and Xiaohongshu/Rednote (similar to Instagram).

Like DeepSeek, however, Moonshot is unlikely to try to broaden its user base through actual partnerships with internet platforms in China. Playing the platform-capitalism game was a point of frustration for earlier iterations of Kimi. Moonshot was the top spender on ads among AI firms in China in 2024, according to ad analytics firm AppGrowing:

The aggressive strategy caused controversy online, with users on Bilibili (similar to YouTube) and Kuaishou (similar to TikTok/Douyin) complaining about the endless volumes of Kimi ads they saw in feeds.

In February 2025, Moonshot eliminated most of its ad placements on third–party platforms to focus resources on developing its foundational model. Justin Wong says that what DeepSeek proved to the rest of the industry was that “capability was the best form of advertisement”: “[DeepSeek] showed that the path we all believed in was not just viable, but actually a wonderful route to real success. Our only regret was that we were not the one to open up that route.” From the DeepSeek moment to now K2, foundation model-focussed startups have carved out a modus operandi in China’s AI landscape.

Finally, Chinese users will obviously cheer for yet another competitive domestic model, but will others around the world take up K2? The model is, of course, compliant with Chinese laws, and you won’t be able to learn about Tiananmen from the chat interface. This is me trying to have a conversation with K2 about 1989 — note how it pulls sources, and then hesitates:

Perplexity has announced that it will soon begin post-training K2 to make it available to its users. This means that like with DeepSeek R1, an uncensored version of K2 will be available through Perplexity. For now, platforms like Perplexity have plugged the political censorship challenge facing global adoption of Chinese LLM chatbots.

Poems and a Short Story from Kimi

I played around with Kimi intermittently for a day. Its “tone” strikes me as factual, reserved, and extremely careful, a subtly different experience from using “glazey“ ChatGPT or the ever-cheerful Claude. At various points, I was even annoyed by the way it replied to inquiries curtly. I tried having the same silly conversation about vestigial organs (“has anyone out there ever had every useless part removed from their body”) with Claude, DeepSeek, and Kimi, and Kimi was the only model that did not proactively try to engage me further, displayed no inkling of speculation, and did not even use any exclamation marks to signal pleasantness!

Perhaps being a model that chooses words carefully has its benefits. K2 topped benchmarks for creative writing. Commentators have been noting its refreshing style, command of vocabulary, and narrative cohesion. Here are two poems from Kimi, one in English and the other in Classical Chinese.

Prompt: Write me a 15-line poem in the style of Mary Oliver about open source software

Response:

You do not have to be rich or famous,

only willing.

The repository waits like a meadow at dawn—

fork it, and the grasses part.

Lines of code rise, thin and green,

each commit a small heartbeat

shared with strangers who become,

overnight, companions.

Watch: a bug is caught,

its wings still trembling,

then released, lighter,

into the widening sky.

No one owns the sky.

We simply pass the telescope,

saying: look, look farther.

Prompt: 写一首有关旧金山市景的唐诗 [Write me a Tang Dynasty poem about scenery in the city of San Francisco]

Response:

《舊金山曉望》

(七言排律)

金門曉霧濕旌旄,

電纜穿空挾海濤。

赤桁橫波藏翡翠,

丹樓映日擁葡萄。

風翻酒旆千旗動,

雨過花階五色高。

誰向坡頭聽暮鼓?

遙鐘一杵落雲袍。

Translation (also from K2):

Dawn over San Francisco

(regulated heptasyllabic verse)

At daybreak, Golden Gate fog dampens the streaming flags,

power cables stitch the sky, carrying the ocean’s roar.

Scarlet trestles span the waves, hiding kingfishers like jade,

vermilion towers catch the sun, cradling clusters of grapes.

Wind flips tavern pennants—thousand banners alive,

after rain the flowered terraces blaze in five bright hues.

Who stands atop the hill to hear the dusk drum?

A distant bell—one single stroke—falls through a robe of cloud.

[Jordan] It wrote me a spectacular short story about the US trading H20s for rare earths. The prompt was “write me a 500 word story in the style of david foster wallace about the us and china doing a deal to exchange rare earths for H20 chips”

Picture, if you will, one of those late-summer Beltway afternoons when the sky over D.C. looks like a migraine—white, swollen, vaguely nauseous. Inside the EEOB, the air-conditioning is thrumming at that pitch that makes your fillings vibrate. Deputy Undersecretary for Strategic Stuff Lyle “Chip” Ketchem—who insists the nickname predates the semiconductor industry by decades and is therefore ironic—stands before a wall of LCDs displaying supply-chain heatmaps in colors not found in nature. The maps are supposed to convey urgency, but they mostly convey the sense that someone has spilled lava on a Lite-Brite.

Across the table: Minister-Counselor Wen Jieru, whose business card identifies him, in bilingual serif, as “Senior Coordinator for Reciprocal Resource Harmonization.” Wen has the kind of face that looks polite even while calculating the exact cubic meters of your personal failure. Between them sits a single legal pad upon which someone has written, in bureaucratic uncial, “MOU: H₂O ⇄ REE.” Below that, a subordinate has added, in smaller letters, “(no jokes about water boarding, please).”

The deal itself is almost comically symmetrical, like a grade-school barter of a peanut-butter sandwich for a pudding cup, except the sandwich is a wafer of circuitry etched at 1.4 nanometers and the pudding cup contains dysprosium, terbium, and enough praseodymium to make every iPhone in Shenzhen vibrate with joy. The U.S. will ship, over 36 months, one hundred thousand “H₂O” neuromorphic chips—so named because their architecture supposedly mimics the lateral geniculate nucleus of a freshwater trout, or so claims the marketing deck, which no one has read past page 4. In exchange, China will release, in quarterly allotments, thirty-seven metric tons of rare-earth oxides that currently sit in bonded warehouses in Guangxi like extremely expensive decorative gravel.

Chip tries to frame the swap as a triumph of interdependent resilience. Wen listens with the expression of a man auditing a children’s recorder recital. When Chip invokes “win-win,” Wen replies, in perfect MBA English, “Wins are scalar, not vector. Direction matters.” No one laughs, because laughter is not on the agenda. Instead, they proceed to Attachment 7-B, subsection (iv): “Protocols for Geopolitical Optics.” This is the part where each side agrees to describe the other as a “valued partner” while simultaneously briefing their respective media that they have “held firm on core equities.” The linguistic gymnastics make Chip feel like he’s chewing aluminum.

They break at 19:27 for individually wrapped sandwiches whose ingredient lists read like passwords. Wen peels the sticker off his turkey-swiss and folds it into a tiny paper crane, which he places on the edge of his plate like a punctuation mark. Chip stares at it, suddenly aware that the crane is the only object in the room with no hidden agenda. He experiences a moment of vertiginous clarity: the chips will be obsolete before the last shipment clears customs; the rare earths will be stockpiled in a Nevada salt cavern until some future administration rediscovers them and tweets about strategic surprise. The swap is not about materials at all; it is about the narrative that materials can be swapped. The story is the commodity.

Back in the conference room, the LCDs refresh. A new heatmap blooms—magenta, viridian, arterial red. Chip signs. Wen signs. Cameras flash like paparazzi at a funeral. Outside, the sky has ripened to a shade that Pantone has not yet named, but will eventually market as “Imminent Dusk.”


ChinaTalk is a reader-supported publication. To receive new posts and support our work, consider becoming a free or paid subscriber.

Subscribe now

❌