Bohan Liu: The markets there are still much more capital-friendly and less founder-friendly; in that, in a sense, it's a bit more of a harsher and more brutal capitalist environment, which a lot of people go there and realize in general, it's like people think of China as this, like, socialist country. No, it's, it's really hyper-capitalism on steroids. I mean, it's so competitive there. A lot of people don't even have margins. I actually happened to go to the world's biggest drone show, a drone fair, when I was in Shenzhen, and if you go there, you realize every niche vertical has, like, 200 competitors doing the exact same thing.

Steve Hsu: Yeah.

Bohan Liu: And you're like, "Does anyone make money?" Probably not, which is why they gotta export and go global. Yeah. And then when they go global to LATAM, to Middle East, to Europe, to America, then they destroy the local competition because they have been through the crucible. They're so lean, they're so efficient, they move so fast, because it's like, you know, they, they, they grew up in absolute absolute hellhole in some ways of capitalism.

Steve Hsu: Well, you know, maybe you and I had this conversation or, or you were around when we had this conversation at Manifest. I was trying to explain to people, like, just what you were saying, that it's hyper-capitalist. So at some level, of course, it's there's some state-led aspect to their economy.

There's some top-down priority setting, but when you look at the microeconomics of any subsystem in China, it's insanely competitive. There's like always many players trying to sell you similar products. Yeah. Like, just, if you just look at the bidding process that assigns you a DiDi, which is like Uber car, it's, like, incredibly complicated market process that actually assigns you the car.

Bohan Liu: Welcome to Manifold. My guest today is Bohan Liu, a venture investor. His firm is called Chemistry. Bohan, welcome to the show.

Thank you so much for having me, Steve.

Steve Hsu: Now, I was on your show a couple years ago, and that's still actually one of my favorite interviews, I think partially because, if I'm not mistaken, when you were at Yale, you also studied religion. Is that right?

Bohan Liu: First of all, you were my favorite guest, but yeah, I studied engineering, but also a lot of religious studies and Chinese history.

Steve Hsu: Yeah. And you, you actually asked me questions that I think no, almost no, maybe no other interviewer has asked me about religion and family and values and things like that.

If I'm diligent, I'll put a link in the show notes to that episode because I do really enjoy what we talked about. Yeah, so today I'm interviewing you. The impetus for this was the recent article you wrote, X article, about a trip to China that you had recently taken, and in particular about the venture ecosystem in China, which I think is poorly understood.

Well, it's poorly understood by everybody because, first of all, if you're in China, you might not know anything about venture if you're not actually an entrepreneur. Secondly, if you're outside of China, almost by definition apologies if I offend anyone, you know nothing about China, so of course you know nothing about China VC.

So anyway, for many reasons, this, I wanna drill down on that, and also just make some more general observations about what you saw in China.

Bohan Liu: Yeah. Happy to chat. It was a great trip, and, you know, there's so much information asymmetry between China and the US, and oftentimes there's, there's a bit more asymmetry on one side. The Chinese obviously know much more what's happening here

Steve Hsu: Yes.

Bohan Liu: And I talked a little bit about it in my article. But I thought this article and how it went viral was quite interesting, showing both the hunger and thirst for understanding China, maybe especially, you know, these couple weeks is very much another like, the second DeepSeek moment, the Kimi K3 moment, so people are, like, extra aware of this stuff. You know, I think for me it was, like, purely an empirical description of what's going on in China, and people were extremely fascinated in the West. And there was actually pretty interesting commentary from the Chinese internet, which as I mentioned, I knew this would get translated back there, 'cause they're so aware what's happening here.

So there was interesting commentary from the Chinese side. And also a lot of them didn't know, as maybe ethnically Chinese, and they were like, "Wow, look at these white people. They, they really know what's going on now," you know? They're like really on it. So

Steve Hsu: That's crazy. Yeah. I think we are hitting a tipping point, just based on my own personal experience in the last roughly six months. There have been sort of really active efforts, at least by some sub-communities in the US, to learn more about China. I think by and large it's still not a good situation, but gradually stuff is happening, and I would hope that, you know, even if you have a negative view that the US and China are gonna be locked in some kind of geopolitical rivalry, we should understand our, our main competitor.

And it's just, it's just laziness or cope self-soothing that I think prevents a lot of the people who think they are interested in this competition from really learning

Bohan Liu: Yeah.

Steve Hsu: The basic facts they should know, even if, if it's all just about competition.

Bohan Liu: I highlight, highlight with you about, like, I think US interest in China, I mean, these, they wax and wanes over years, you know, from, like, even a couple hundred years ago to then, like, Nixon's visit and then WTO. There's often times where it come up and down.

Steve Hsu: Yeah.

Bohan Liu: And I think, like, since Trump won in 2016, almost exactly 10 years ago, I think that's really when it came to the forefront in a lot of ways. But in different templates. You know, more that's when 2016 was the start of view China as a peer, as a threat, as a geopolitical adversary when they didn't before.

And I think now you're right, where COVID was very much a low point in kind of the understanding, 'cause there was just no human flows between the two countries. But now it's coming back, and I think the last year and a half sitting here in Silicon Valley, I've seen probably unprecedented levels of visits and interest from American founders and VCs who are going on the ground now.

That's the difference. So they were willing to go in person to visit, talk to people, and also Chinese government and society has become more open with visas and, and access.

Steve Hsu: Yeah.

Bohan Liu: And so it's an interesting moment where I think understanding is improving and these foreigners, even some of them who are quite hawkish, go there and come out nearly unanimously with more respect and even more positive view on China after being there.

Steve Hsu: Yeah. We're getting, like, ahead of ourselves, but, 'cause I wanted to start by introducing your background. But just one remark on this which is that you know, if in Silicon Valley it's become almost a cliché or at least quite common that people feel like, "Shit, you know, I should go to Shenzhen and just see what's happening there because that is where, you know, all the advanced manufacturing and the whole supply chain is there, and I should at least go and check it out."

I've been aware of many different independently organized groups of people from the Valley who went there just to look around, and a lot of them have written about what they observed.

So let's start with you though.

So where did you grow up?

Bohan Liu: Yeah, I was born and raised in Shanghai, came to the US for undergrad. So yeah.

Steve Hsu: Did you go to a, an international school? Why is your English unaccented?

Bohan Liu: Yeah, I get that question a lot. I think one, you know, I think I was generally good at learning languages, but my dad also installed a satellite at my house where I watched a lot of American TV growing up, like Disney, Nickelodeon, Cartoon Network. So both in terms of accent and in terms of probably the culture propaganda, I was very Americanized that way. But my school was kind of interesting, and maybe I mentioned to you, it's actually like a chip fab school. So I went to the SMIC private school, and SMIC is Mainland China's version of TSMC.

It stands for Semiconductor Manufacturing International Corporation. And so I grew up with all the children of these Taiwanese chip execs, who basically were all TSMC execs who went to Mainland China to build up this enterprise wherein, which when I was a kid, everyone thought it was boring because their parents worked on it, no one thought it was serious.

Now it's like, you know, China's national champion. So it's took on a whole different geopolitical valence. But yeah, I grew up at a Taiwanese school my whole life in Shanghai.

Steve Hsu: I think I told you that when I was there about a year ago, there were some readers of my blog or followers of this podcast who are either Chinese living in Shanghai or Taiwanese living in Shanghai, and through them, a bunch of them contacted me when I was visiting Shanghai, and through them I got introduced to a bunch of senior people at SMIC, and then was invited to some hot pot dinner, and all the families there, their kids went to your high school.

Bohan Liu: I probably knew a bunch of them. Yeah. Yeah, it was a very small community.

Steve Hsu: That's how those families all knew each other is 'cause they're all their kids had gone through the SMIC school, and some of them were not even SMIC employees. They just sort of figured out a way to get into that school because it was such a good school.

Bohan Liu: Well, my, my family were not SMIC employees. Yeah. Actually, my dad is a professor of oncology, and so it was just, I was one of the few kids who was neither Taiwanese nor involved with chips.

Steve Hsu: Yeah. I think the revealing thing for Americans is there's a huge community of Taiwanese living in Shanghai. They love it. They have a great lifestyle, quality of life. SMIC, big chunk of their top people are of Taiwan origin, and that's the organization, the, the chip fab organization that's really leading the Chinese push and is the counterpart of TSMC, for viewers that are not familiar with that. It's really right there at the center of the whole US-China tech rivalry.

Bohan Liu: Yeah, 100%. Yeah, it's interesting going, growing up next to the chip fab, you know? That's my childhood.

Steve Hsu: Yeah. Crazy. Okay, so you go to Yale. Did you like it?

Bohan Liu: Loved it. Best time of my life. Very, it's a very happy school.

Steve Hsu: What residential college were you in?

Bohan Liu: I was in Ezra Stiles.

Steve Hsu: Okay.

Bohan Liu: And, you know, like the story about Yale itself, it was, Yale was a, like, trad fork of Harvard, right? You know, they were all seminaries, religious schools. And at some point these ministers thought Harvard was getting too woke because they stopped making Hebrew mandatory. So this group of conservative ministers, like including Ezra Stiles, spun out to start a university.

Steve Hsu: I didn't know that. So when I was there as a professor, I was a fellow of Silliman, just because it was the closest college to the physics building, so we used to go there and eat lunch.

Bohan Liu: Yeah. Benjamin Silliman himself I think was a notable chemist, I believe.

Steve Hsu: I didn't know that actually. Yeah.

Bohan Liu: Scientist himself.

Steve Hsu: Yeah. So I It's funny 'cause I was one of the few professors who liked to talk to undergrads when I would go to Silliman. Everybody else just wanted to cluster and, like, talk physics or talk math or something like that, so yeah. But I enjoyed it. Yeah, and think sorry, I interrupted you a little bit. Uniformly, the kids I know who went to Yale really liked it, whereas the people who went to Harvard often bitch about their time at Harvard. So

Bohan Liu: You're 100% right, and obviously I'm biased, right? But I think you actually see it in, like, national rankings that Yale on average is, like, one of the happiest Ivy Leagues and happiest schools in the nation.

Steve Hsu: Yeah.

Bohan Liu: It was always very much community driven, and everyone I know, including me and my girlfriend and my roommates, you know, would go back and do it again. But most Harvard and Stanford kids probably wouldn't do that again. Like, they're clearly top schools Yeah. But it was, it was different. It was it was a different vibe, but I think more

Steve Hsu: It's so ironic 'cause it's a little bit different for the professors because, see, part of

I think part of the reason why Yalies are happy is 'cause New there's not that much to New Haven, and so then you sort of focus inward and develop this healthy community on campus, whereas Harvard, you know, there's so much more around in Cambridge and Boston and MIT and all the other schools that somehow maybe that didn't, doesn't allow Harvard to create the internal environment that they should have. But the professors like living in Boston more than they like living in New Haven. Yeah. So.

Bohan Liu: It definitely makes a lot of sense, and, like, you see that very extremely with Columbia University, right?

Steve Hsu: Yes.

Bohan Liu: Because it's in manhattan, it's like the most exciting city. The FBC have no campus culture. Very diffuse.

Steve Hsu: Yeah.

Bohan Liu: Like, who wants to hang out in Upper West Side Columbia campus when you can go to

Steve Hsu: Exactly.

Bohan Liu: Middle of New York city?

Steve Hsu: Yeah, and plenty of, like, small my daughter goes to a small liberal arts college, and they, those kids are really happy, and it's, there's nothing around but farms, you know? And so, but that means

Bohan Liu: Yeah

Steve Hsu: you find your friends on campus and you have these, yeah, I think pretty intense bonding experiences.

Now you're a venture investor. Yeah. And so tell me a little bit about your firm. Your firm is called Chemistry.

Bohan Liu: Just Chemistry. Yeah, I mean, so yeah, I'm here. I'm a partner at Chemistry, which is, you know, a relatively new fund started a bit less than two years ago. My three partners were previous GPs at Andreessen Horowitz, Index at Bessemer and decided to start a new fund really focusing on the early stage and, you know, on the chemistry between people, which is, hence our name. And we're very much generalists, investing across all sectors. You know, from biotech to robotics to agentic software to everything like that.

And so and some of them had some experience in China. Actually, Ethan used to be one of the managing partners at Bessemer, and I think he was the one who helped set up the China arm for Bessemer many years ago. And so yeah, we are very much a Silicon Valley based VC and we, you know, don't directly invest in China generally, but you know, now China's so relevant as maybe the only other place on earth where there's real innovation in AI, robotics, and frontier technologies that so many other people want to go there and stake a little ground, and I very much do that too.

Because also a lot of their talent comes to the US. When they come here, you know, they often want to be plugged into the ecosystem. Yeah.

Steve Hsu: So remind me, what is the current legal situation for US funds to invest in startups in China? Can you do it? What do you have to is there anything stopping you from doing it?

Bohan Liu: Yeah, I think there's probably quite a few forces stopping you from doing it. Like, so maybe let's say a couple years ago, I think, you know, as geopolitical tensions were already rising, there were probably you know, a lot of it was firm dependent on just the firm's own risk tolerance or stances. Let's say like Marc Andreessen, right?

You know, he's pretty openly hawkish in certain ways, or Founders Fund, they incubated Anduril. So these funds, I think they would have internal policies where they probably would not invest in companies that were majority owned by Chinese founders or Chinese citizenship, even if they moved here. And so every firm had different, you know, internal tolerances.

But I think over time it's become more of like regulatory top-down from government side pressure to not really invest where you know, the US has CFIUS, right? Not wanting, you know, foreign money to invest in US strategic assets. And then there's reverse CFIUS not wanting American money to go invest in, you know, Chinese strategic assets.

And China has basically the same counterparts, right? You know, they don't really they don't really want foreign money invest in like top Chinese AI semi-companies. You really aren't really allowed to. And actually now with the Manus deal I think it's probably even worse now in terms of the, the FDI climate.

Basically, you know, Benchmark Chetan Dsouza there, he led the Series B, which was a very contrarian bet in Manus, right? As like a top AI app. But it was really it was a Chinese team in China, then they moved to Singapore to kind of de-Chinafy, but I guess it wasn't de-Chinafied enough. And so they took the investment and then sold to Facebook, but then it the whole investment kind of got rolled back really.

You know, the money was distributed to LPs, but I think the, the government wasn't quite happy with setting a precedent that a top Chinese AI asset could be sold, you know, to an American company like that.

Steve Hsu: But just to drill down on the actual regulatory stuff. So like, say you're a US VC and technically you're a Cayman Islands fund or something. What aspect of it allows the US Fed Gov to say like, "Yeah, you can't participate in this A round in Shanghai"?

Is there, is there anything there or is it just your LPs maybe are not allowed to do it?

Bohan Liu: So, yeah, and that's another level of control. If it's not coming from government, sometimes LPs have certain mandates. I think they're probably a bit less common now for the LP to say strongly can. I mean, they probably have a preference not to because I think, I don't know the detailed regulatory status if you were like a Cayman Fund, whether the US could regulate you on that.

I think most large American VC firms probably aren't established that way. But I think, you know, now the Chinese government's very sensitive that they don't want you to come in either. And from the American side, let's say even without any regulatory hurdles, from just a risk perspective on are we gonna get returns, I think there's questions of like

Steve Hsu: Yeah

Bohan Liu: let's say you invest in a Chinese company, and where they're, they're probably taking RMB. Okay, so you invest RMB. Can you get that out of there? Right? There's capital export controls, right? Now let's say they list on the Hong Kong stock market. Okay, that's more liquid, that's more global. Maybe you can get your money out, but these are just additional levels of risk

Steve Hsu: Yeah.

Bohan Liu: You have to deal with that like maybe the investment works out like in like a ByteDance, right? What if you turn into a ByteDance, where it's probably one of the world's most valuable and most strategically well-placed companies, but it's going through so much geopolitical turmoil where people cannot exit and they're going through a huge multiple suppression.

Steve Hsu: Yeah.

Bohan Liu: Like if that thing was an American company, it'd be like a trillion-dollar company for sure right now you know.

Steve Hsu: Well, I think first of all it's a completely legitimate topic to discuss geopolitically why you know, it's difficult to succeed there, it's difficult to get your money out. Those are all challenges. I was just asking the narrow question 'cause I do work with some venture funds and I've never quite gotten a clear answer like you can't invest in China or you prefer not to invest in China? There's a, you know, difference between those two positions. Yeah. Some of them

Bohan Liu: I think to your point, it's kind of a gray area in some ways. These things are moving a lot, and a lot of these things are kinda case by case, right? Because if the regulators come in, they might review on a case by case where

Steve Hsu: Yeah.

Bohan Liu: If you really wanna push ahead, I actually think you could.

Steve Hsu: Yeah.

Bohan Liu: I think it's not a black and white thing that you cannot.

Steve Hsu: Yeah,

Bohan Liu: But there's so many barriers here that

Steve Hsu: Yeah.

Bohan Liu: If you wanted, it's up to you if you wanna go through them.

Steve Hsu: Got it. Yeah. I mean, it might not be advisable for a US fund to look for deal flow there, but that's different from literally they cannot invest there, right? Which I'm not just not sure where the actual line is right now.

Bohan Liu: Yeah. And I think, like starting with Mana, is the government Chinese side has put out more restrictions.

Steve Hsu: Yeah.

Bohan Liu: If I'm not mistaken, I think any foreign capital investing in like a Chinese AI company now has to go through a Chinese regulatory approval

Steve Hsu: Wow.

Bohan Liu: which I think before the Mana sale did not happen.

Steve Hsu: That's crazy. Needless to say, I guess you're looking for your deal flow mostly in the US?

Bohan Liu: Yeah, we're very much in it. Well, we have a couple companies in London, very much like an, an American fund. Yeah. So we're valued.

Steve Hsu: And another question I'm sure people are curious about, so you mentioned your partners are all pretty experienced and spent time at pretty well-known funds. So how does a young guy like you get to be a partner in a fund, you know, with, with guys with so much experience?

Bohan Liu: Well, yeah. First, first of all, when I joined, I wasn't a partner. I made partner I guess a few months ago.

Steve Hsu: Congratulations.

Bohan Liu: I think it was a lot of right place, right time, where I used to work at a few AI startups before. You know, I studied EECS in college as internships and like kind of robotics and AI, but then worked at a few AI startups, and I never thought about being a VC.

I always thought I'd be either more founding or I was an entrepreneur-in-residence at a different VC firm trying to incubate a gov tech startup. But my friends were like, "You know, you're, you like reading, you like talking to people, you're curious. And being an investor inherently is a more interdisciplinary and breadth-based role. You, you think more, you do less versus a founder." Yeah. Right?

Steve Hsu: Yeah.

Bohan Liu: And so I thought I'd try it out, and I happened to meet my three partners as they were spinning out and launching their first fund, and kind of like joining a startup, I decided to take a bet on them, they took a bet on me to join as the first hire to help launch a new fund, and that's probably why I had an accelerated path because I took the risk and it paid off.

Yeah. That's great. Great for you. Our fund has done well and, you know, it's, it's been a lot of momentum.

Steve Hsu: Yeah. I'm sure many, many people your age would kill to be in your seat right now.

Bohan Liu: Yeah. So very much I appreciate the opportunity. I was introduced through a friend, and it really worked out this time.

Steve Hsu: Yeah. That's great. It's a good lesson for people to just like, you know, try stuff. Like, you can just do stuff.

Bohan Liu: Well, I firmly believe most people are under-risked in life. You know? Like just, you know, on average a lot of people we know, they have more safety nets than they think about and, you know, obviously some people can overcorrect and really blow up, but on average it's not happening. And so I think I always had relatively a kind of risk-on appetite and

Steve Hsu: Yeah.

Bohan Liu: that's why I joined multiple startups. It didn't really work out, but I was willing to keep rolling the dice. I'd rather have the adventure. Kinda like, I think Joe Tsai, right? He was, I think 36 when he joined Alibaba, right?

He was married. His daugh his wife was pregnant with their first child, with their daughter, and at the time he was like a, you know, big shot Wall Street lawyer, you know, making like really good money and, you know, he was also, also had a lot of social cachet in like the Hong Kong upper echelons. But you know, he met this guy called Jack Ma and he was like, "Let me just take this adventure for two years. If it doesn't work out, I can go back to my cushy Wall Street job. It's gonna be fine."

Steve Hsu: Yeah.

Bohan Liu: But he decided to move to, at that time, a third world country to a third-tier city to, with this founder who couldn't even get a job at McDonald's to just take a risk on him. Yep. And now, you know, I'm sure his law firm partner buddies who at the time made fun of him, I mean they looked up to him and he's God

now, right?

Steve Hsu: I do wanna say though that, you know, you become aware of the guys who took that risk and then, and it paid off. Yes. You, you don't become aware so much of the ones who

Bohan Liu: There's survivorship bias. Yeah, yeah. And you know, there's that article like your life is not a Monte Carlo simulation, right?

Steve Hsu: Exactly.

Bohan Liu: The variance doesn't matter, right? You get this one life. But you know, like just like YOLO goes both ways, I also think because I have this one life I'd rather take that risk.

You know? I guess I'm that kind of guy. Yeah.

Steve Hsu: Well, I definitely did exactly what I wanted to do, like studying theoretical physics, and that definitely was not advised. Like my dad said, "Oh man, nobody, nobody can get a job in theoretical physics," you know, "You're, that's, that's not the right thing for you to do." But you know, it worked out okay for me.

Let's talk about your trip to China. So when was that?

Bohan Liu: I was there, I guess, almost three months ago now. So I went for 10 days and, and I've always gone regularly because my family is still there. But that was just personal, but this is the first time in my life kind of going for, for business in a sense, where this time was meeting a lot of investors and founders in robotics, biotech, you know, AI, LLMs. And I was in Beijing, Shanghai, Shenzhen, Guangzhou, and Hong Kong, so for 10 days.

Steve Hsu: Great. When you went, you were specifically there to check out the tech scene or specifically the venture scene? Any, any goals that you had when you set out?

Bohan Liu: I think there was a sense in my mind that, you know, Silicon Valley is so obsessed with China right now. I knew all these VCs were going, even including, I maybe you'll see soon an article that even Founders Fund is going this month, I believe

Steve Hsu: Wow

Bohan Liu: or next month. You know, and they're, they're probably not gonna invest there, but, you know, people are there to learn, right? Like Sequoia, America is going to China, like Thrive goes there, Benchmark's Partnership, the founders of Stride founded Paradigm. You know, everyone went, went there last year. And I was thinking, you know what?

All these like, you know, foreigners going there, and I'm sure like They're learning things, but a lot of them are also getting kind of like the foreigner tour, right? They're getting like the, the gringo tour. And I was like, you know, as, as like someone who like speaks the language natively and has a lot of connections, I thought I could get a more authentic view on the ground in some way.

So that's part of why I decided to go, and I think it very much panned out, even surprised me to the upside, where, you know, I just, through a lot of my network, I wanted to meet all the Tier 1 funds, and I met a lot of the top robotics companies, like, you know, some of the top kind of like CRO, CDMO players who are very much part of global biotech supply chain.

And then some, some researchers and even like some just general intellectuals, you know, I'm sure. Like I happened to be at this dinner introduced to a friend in America to these, these professors who are, They all came from this like kind of one group of Chinese scientists who were educated in the US in the '90s and 2000s, all became top like kind of endowed chair tenure professors in the US like Shi Yigong, Rao Yi, Lu Bai.

And then they all went back to China in like the 2000s, 2010s to kind of like serve their nation, and they ended up becoming like the head of the life science department at Peking, Tsinghua, and helped start Westlake University, if you've heard of that.

Sure. you know, like

Steve Hsu: Yeah.

Bohan Liu: China's kind of, I don't know, IAS, MIT thing. I was at this dinner with them, and they all happened to be neuroscientists, and it was interesting hearing how they were thinking about kind of the scientific reforms in China, not in any specific field, but in reforming the incentive, trying to get started with like, you know, kind of the Howard Hughes of China, better incentives for high-risk, high-reward research. So, so yeah, there was a lot of interesting conversations around there. Yeah.

Steve Hsu: When you go there, you, at least I find, you know, they're at a in a situation now where they are thinking about how to fix their system, that they have a lot of problems internally in their system, and, you know, it takes some out-of-the-box thinking to map out how they're gonna go from like catch-up mode to actually leading horse mode, and it that transition is, is happening right now.

It's somebody with a little more experience there or speaks the language would get a much, much deeper dive than just some, you know, for lack of a better word, white guy who doesn't speak Chinese and has, you know, never been there before and is just trying to like get a feel for it.

Bohan Liu: Yeah.

Steve Hsu: Even for the people in that second category, I think it's super valuable because I routinely meet people that are like high information. They could be like pretty well-known tech founder or VC, or they could be some supposedly brilliant rationalist guy or something.

The stuff they say about China just, just, I can't believe it. It just makes me laugh. It means there's something kind of like wrong with their epistemic framework. Because, you know, when they go there and just the fact that like people are not like in fear of the secret police constantly or something is a shock to them.

Like they get there and they're like are people really this free, or are they just pretending to be this

Bohan Liu: Yeah, they seem happy.

Steve Hsu: Yeah, they seem happy. Like, what and, and the police

Bohan Liu: And no one knows about the social credit score. No one's heard about this thing.

Steve Hsu: And the police, nobody seems to be ever afraid of the police. The police are pretty nice people. They're like the guy who, you know, like taught you chemistry in college or something. It's like some pretty, like, you know, Chinese people are usually not, like, aggressive, you know, angry people. Yeah. And so it's like this guy looks like, oh, he could be the, like, the guy who runs the Chinese restaurant down the street from me, and he's a cop. But, and nobody's afraid

Bohan Liu: there's a lot of more public accountability than people think. They're not like a North Korean or like the Iranian Shah's ... sort of police. You know, it's like the police officers are a lot of scrutiny and pressure. Like, they often, like, you know, there's a scandal. They, they get fired.

Steve Hsu: Oh, yeah. They're under a lot of pressure to behave properly actually. Yeah. And there are cameras everywhere. That works on them as well as on everybody else. Yeah. Right? The short sentence I always say to people who have, are at the minimum level, which is most of Americans, I always just say it's, it, the system there is more like Singapore than it is like North Korea. You may have heard it's North Korea, but it's actually more like Singapore. So among the conversations you had though, I guess, so you talked to founders, but then you also talked to venture investors, and there was a article that got a lot of attention in the FT. It was maybe more than six months ago or around six months ago. I think Eleanor Alcott, who covers, one of the people who covers China. I'm name-checking her 'cause I kinda know who she is. so she covers venture and tech for the FT, and she wrote this article saying like, yeah, there's basically no venture, there's no money being invested in China in venture, and she had some statistic that she quoted, and she took a lot of heat for what she said.

Bohan Liu: And maybe you can paint a picture about how healthy is the ecosystem there. How does it differ from the ecosystem in Silicon Valley, et cetera, et cetera. And I haven't read her article, and when she says no money, I'm curious if she means, like, kind of foreign LPs going in, and I don't wanna say no. That might be true. That over the last few years, it's true that foreign LPs investing in China has decreased. Like, you know, I've talked to LPs here who have generally kind of been, you know, cutting down a bit more to China allocation. Although, I do wonder whether now is kind of like the contrarian time to, to resurrect that potentially, and a wait to be seen. But domestically, there is still money. But I will say this, the system works very differently, and maybe as you alluded, I recently wrote an article that went, you know, somewhat viral on Twitter explaining kind of the main differences with the Chinese venture scene. And so, like, I think high level bullet point is just The markets there are still much more capital friendly and less founder friendly in that, in a sense it's a bit more of a harsher and more brutal capitalist environment, which a lot of people go there and realize in general, it's like people think of China as this, like, socialist country. No, it's, it's really hyper capitalism on steroids. I mean, it's so competitive there. A lot of people don't even have margins. I actually happened to go to the world's biggest drone show, a drone fair, when I was in Shenzhen, and if you go there, you realize every niche vertical has, like, 200 competitors doing the exact same thing.

Steve Hsu: Yeah.

Bohan Liu: And you're like, "Does anyone make money?" Probably not, which is why they gotta export and go global. Yeah. And then when they go global to LATAM, to Middle East, to Europe, to America, then they destroy the local competition because they have been through the crucible. They're so lean, they're so efficient, they move so fast because it's like, you know, they, they, they grew up in absolute absolute hellhole in some ways of capitalism.

Steve Hsu: Well, you know I you maybe you and I had this conversation or, or you were around when we had this conversation at Manifest. I was trying to explain to people, like, just what you were saying, that it's hyper capitalist. So at some level, of course, it's there's some state-led aspect to their economy.

There's some top-down priority setting, but when you look at the microeconomics of any subsystem in China, it's insanely competitive. There's like, always many players trying to sell you similar products. Yeah. Like, just, if you just look at the bidding process that assigns you a DiDi, which is like Uber car, it's, like, incredibly complicated market process that actually assigns you the car. or the delivery guy or whatever, and

Bohan Liu: Well, in the US you have, like, Uber and Lyft basically, and now didi Global. But in China you literally have 30 vendors auctioning in a real-time auction system for your bid, you know?

Steve Hsu: Right. So I don't think people realize that, is that, is that when you ask for the car to take you to place X, there is already, like, a market process just to assign that car to you, right? That driver and that car, and that, that isn't the case here so much. Anyway, once you drill down, you realize it's an incredibly competitive market system. It's just at maybe at the higher levels it is somewhat less,

Bohan Liu: Yes. There definitely is

Steve Hsu: Yeah.

Bohan Liu: More of state direction, and I can talk a bit more about that.

Steve Hsu: Yeah.

Bohan Liu: But, you know, so on the venture side, I think the Chinese kind of venture scene, at least from a founder's perspective in terms of, like, the relative leverage between the investor and the founder, is much more reminiscent of the US in the '80s and '90s, or even like, you know, Europe now, where the cap venture capitalist is much more, like, kinda sitting on their throne holding forth, right?

You know, like, the founder comes to them, and they have more kinds of diligence. They hold more of the cards, they hold more of leverage. And so relatively, there's less capital chasing companies in China relative to the US scene, where here it's like every firm beating each other up, begging to give money to founders.

In China, of course, the best projects have that dynamic too, but it's much more pronounced that it's more in favor of the capitalists, which is why I think something that shocks a lot of, you know, Western founders is that because of these dynamics, Chinese founders sometimes sign term sheets where they're just much more predatory in a sense, right?

Like, you're signing a term sheet where you say, "I must return money with a potential hurdle rate in, like, six to eight years," which by the way, in the US on average, we expect, you know, venture exits to, like, be 10 to 15 years now. And LPs have that patience, GPs have that patience, but China, they don't. They're like, "Well, I want results." And if you don't deliver results, guess what? You could be personally liable. So imagine that, right? Like, in, you know, the West, we have these LLCs, limited liability corporations. The whole point is, like, you know, your, your personal wealth is not on the line. But in China, like, the stakes are so high where I think I was so surprised that people are willing to take that level of risk to go all in.

It's, it's not the opposite of the, the meme right now. Like, every Stanford kid, it's easier to get a YC summer founder internship than getting a real internship. Everyone tries. There's you can only fail up. There's no stakes here, right? So everyone wants to be a founder. But in China, the stakes are tremendous, and so you really gotta know you want it because you're, you're potentially signing your life away if this doesn't work.

Steve Hsu: So let me, let me explain that to the audience, 'cause not all of my audience, you know, some people are just mathematicians proving theorems or something. They, they don't really have experience with a term sheet or dealing with venture investors. So, you know, in the US what happens is you go out, you, you create an LLC, limited liability corporation, and it is, that thing is the legal entity.

The investors are putting money into that thing. If that thing declares bankruptcy, they cannot come after the founder, me, and try to take my house, right? I don't put up my personal assets. Some people are very shocked by this. And, and actually, you know, in general, what we're used to in the US specifically in Silicon Valley is not what they're used to in Germany or in most other countries.

Like, like we, we have adopted this idea that like, oh, if we treat the founder, the magical special founder person in the right way, in this way, we all win. But that has worked in the US and Silicon Valley. Other places in the world are not, have not been convinced yet that that model is actually okay for the LPs, right?

And so there, there's like a learning curve where they have to see success before they decide, oh, you can be that nice to the founder and still everything's okay. So in the US system, there's no real downside for the founder except the time that you risk. So, so you're gonna invest years working with this company, and you may come away with zero, but at the end, your company just declares bankruptcy.

The investors lose all their money or most of their money, and, but other than that time that you spent on the company, you didn't lose anything. And in fact, you gained a lot of stuff because you got a lot of experience, you learned some industry very well, you made a lot of contacts. So, so it's actually quite positive the, the, the sort of cost benefit for the entrepreneur.

Whereas in China, as Bohan just explained, like they might make you sign some contract that says, "I will return this rate of, you know, return," or, "I will, I will generate this rate of return over the next six years or eight years for the investors, and if not, my personal assets are attach are up for liquidation or something." And now my understanding is that's a relatively recent thing.

So has that ever played out to the point where you have LPs like in a very nasty, noisy way like trying to take the assets from the founder? Well, oh,

Bohan Liu: 100%.

Steve Hsu: Yeah. So it does happen.

Bohan Liu: Totally happens. It, it can get really ugly, you know? Especially last few years there was, you know, market downturns. Even people go to litigation and may, you know, some people may even go, go to prison, you know? Yeah. Because it, it can get very ugly. And so Yeah, I mean, that's a big shock. And, and, and one other point that makes it even a bit worse is that in the US, you know, maybe for viewers who are really in startup scenes, like, you know, how can startups exit generally?

You can IPO, so you go to public markets, or you can sell to another company, right? There's M&A. And the huge thing that's different about China, there is no M&A market. There are no acquisitions. Startups do not get bought, right? And so in America, maybe there's all these like intermediate interim off-ramps in some ways, right?

You can exit somewhere and get some money back or even make some money from big exits like, you know, Wizz can sell to Google for $30 billion, things like that. But in China, it doesn't exist, and I think it's partially because, you know, China's tech sector is almost this like oligopoly of like the BAT as they say, right?

Baidu, Alibaba, Tencent. They're all extremely aggressive. They're, you know, very fast at like iterating, copying things. If you have an idea, they have strong engineering teams, they can build everything themselves, and labor is on average cheaper. They just don't see as much value in hiring your team.

There's no acqui-hires, and so they're like, "We'll just compete ourselves or destroy you." And so yeah, people so you also can't even you have way less exit opportunities in addition to shorter exit timelines. And so that's why everyone is like under so much pressure to really IPO in the Hong Kong stock market basically next year or else they're absolutely screwed.

Steve Hsu: Yeah. Yeah. There's a lot to unpack there.

I mean so again, it's the case that at least in that narrow sense, China is more capitalist and market-driven than America. Capitalist meaning capital, the guys who invest the money have a more favorable position than in the US, right? US investors are accepting far fewer guarantees from the

Bohan Liu: Yes.

Steve Hsu: People that they're giving

Bohan Liu: Yes.

Steve Hsu: Their money to, right? I would point out, though, before Silicon Valley became a success, so I actually came into this right around the original dotcom bubble, and, and at that time it was very common for, like, some very experienced semiconductor engineer to try to start some chip company, and they would give up, like, half the equity of the company for

Bohan Liu: That's right.

Steve Hsu: A couple million bucks, right? So, so that was an intermediate thing very, very different from today. Like, if you got in a time machine and got out in 2000 and you looked at the term sheets that people were dealing with, you'd be like, "Wow, these people are really screwing the founders by modern standards," right?

But if you go even a step back, 20 years before that, you could, you know, say you're an inventor and you, you have a patent on something, and you're trying to raise money to make the product. You would have to put your house up. There, there's nobody there was no one in America that was just gonna give you the money unless you put up collateral even though they liked your invention.

Bohan Liu: Yeah.

Steve Hsu: Right? So that, that's, like, the more it's just, like, it's just, like, they're earlier in the development stages of this stuff, right? So

Bohan Liu: Silicon Valley has become much more founder-friendly over the, you know, past few decades, right?

Steve Hsu: Yeah.

Bohan Liu: With the invention of Founders Fund and Greylock Horowitz, and they, they have made a splash and there was always when they got started, all the other VCs are salty.

They're like, "You're destroying the game. You're destroying the old boys' club cartel," right? Like, why are you you know, then, you know, all the old school boomer VCs are gonna lose all the deals to these new founder-friendly VCs, right? But then, you know, then the equilibrium just shifts. And then the new standard is just founder-friendly, and it just keeps going there.

And on a longer time horizon, trying to make it there, and as I mentioned, we do see signs, for example, the typical tier ones you think about China are, like, kinda the US dollar-denominated Chinese funds like Sequoia China, now called Hong Shan, or Jing Wei, which is formerly Matrix China. Like, you know, Zhenfund or Hillhouse and Sequoia. Sorry, Hillhouse, which is also US-funded. These firms often are kinda

Steve Hsu: Hillhouse is a Yale thing, too, right?

Bohan Liu: Well, funnily, Yale Business School is not a top-tier business school in America but has produced the most legendary Chinese capitalists because the founder of Hillhouse and Sequoia are both went to Yale.

Steve Hsu: Yeah. But isn't Hillhouse a Yale address or a Yale what, what is it?

Bohan Liu: Yes. It, it is named after Hillhouse Avenue, which Mark Twain called the most beautiful street in America.

Steve Hsu: Yeah, yeah. I just thought I'd mention that . So yeah. So Hillhouse is a big deal in China, but it's, it's really a, it's a Yale

Bohan Liu: Well, they're actually, I think, the single biggest fund in all of Asia, I believe. and so I'm saying these funds are, you know, .A bit more founder-friendly than, I say, the Chinese RMB-denominated funds yeah whose LPs are local provinces and municipalities. So those have, like, the most strings attached. Yeah. So I think China, there is a spectrum, and I think people, for, like, hot deals, I've talked to founders where, you know, if you have if everyone's bid on you, you do have the leverage to

Steve Hsu: Yeah.

Bohan Liu: Negotiate better terms for yourself.

Steve Hsu: Absolutely. From the viewpoint of the founder there what, what does it look like? So I met people, you know, who are like seniors at Tsinghua, and because their advisor's kind of famous and they did some great work and their advisor really loves them, they're like, "Oh, I went out and raised money for my first I raised a round for my company," and this kid is like, hasn't even graduated from college.

So from that perspective, it's feels a lot more like Stanford, right? But then like I'm sure there are guys who just all blood, sweat, and tears, and their apartment is on the line if they don't deliver, you know, a success. So, so I don't know, I don't know if there's any kind of common experience that people have.

It's just like pockets of like the superstar Tsinghua kid versus like people who are grinding for 20 years. Like how do you describe it?

Bohan Liu: To be fair, there are similarities in that if you like in America, I don't know, what's like the most sought or sought-after founder background?

Maybe it's like you went to high school at Harker, and you went to Stanford undergrad, and then you dropped out, and you interned at Anthropic. Okay, you know, there is a similar kinda like prestige hierarchy and signaling in China, right?

Steve Hsu: Yeah.

Bohan Liu: It's like, you know, you went to like Beijing Number 4 High School, then you went to Tsinghua for computer science, and you were in the Yau class, so you were the best of the best, and then you interned at ByteDance over your summers, and in like the Huawei like genius program, things like that, right?

And so if you come from that background, of course, you are also gonna have a easier time fundraising, where you're the kind of founder that venture capitalists just wanna give you money no matter what, right? And so yes, that, that still exists for like the best of the best hottest founders.

Steve Hsu: I know a lot of Yau, like there are two Yaus, Y-A-U and Y-A-O. When I go to Tsinghua, I was just there quite a lot this spring, and I'm gonna be there in the fall. I meet a lot of these Yau B- Yau 1 and Yau 2 class kids, and yeah, they're like really you know, they've got a golden path ahead of them.

Bohan Liu: Yeah, 100%.

Steve Hsu: Overall you would say the venture infrastructure there is sufficient to support creation and innovation of new companies, or is it not? Like, are, are there lots of people who

Bohan Liu: Well

Steve Hsu: who could do great things in China, but just the, the venture risk capital is just not available to them?

Bohan Liu: So I guess to your question, okay, one, empirically, yes. Empirically, of course, China has created basically some of the most innovation enterprise value just second to the US really.

Despite the pitfalls of the system and the brutality for the founder side, it has worked, right? And maybe founders just put up with more, but the ones who go big, they still make Meituan, and NetEase, and JD, and Alibaba, and Tencent. You know, they still, they still happen. But to our question earlier about, like under-risking, I think a lot of individuals are under-risked, and a lot of, you know, society and capital's markets are also under-risked.

So to your point, in America it's become more founder-friendly and yes, there's more grift, there's more bullshit, but on average it's good. It's good for society 'cause it does encourage the marginal innovation. It lowers the barrier and friction for people to just try high-variance things. And so on that spectrum, I think America, by the way, could still take on more risk, to be fair.

Like, venture capitalists could be even looser with the money, and I don't think we're at the point of a correction yet necessarily from a silo perspective. From a, like asset manager perspective, yes, Bill Smead's firm is underperforming the S&P 500, but like, you know, from what's good for society, I think you can still go further, and I think China is still behind on there in that they should have way more people giving money to young people to try crazy new things, high-risk things, right?

Maybe more like Tyler Cowen EV grants, just things to encourage people to take on variance and do interesting things. Yes, they're definitely behind, but despite those difficulties, they still come out with juggernauts like Moonshot, right? Or Unitree and things like that.

Steve Hsu: Yeah. Let's return to that in a second, but you, you said something interesting about the US scene, which is something I've thought about a lot and discussed with a lot of people, which is that you know, one could make the argument, like a lot of hedge fund guys don't like the whole venture scene. It feels weird to them 'cause they're hedge fund guys, right? So they're like, their perspective might be something like, "This is all bullshit grift, and it's all feels and vibes, and you know, and you're not marked to market for 10 years at a time or whatever. And hey, just look at the rate of return, man.

If you, if you throw a dart and hit a random venture fund, their rate of return is not very impressive. It's sub S&P 500, as you just said, right? So maybe for the investors, it's, there should be less dollars allocated to venture in the United States. But you said something else, which is that for the long-term growth rate, improvement of quality of life, development of new technologies, spread of great consumer products, we're not at a point where diminish or may, maybe, maybe we, we should still allocate more money toward high variance entrepreneurial activities

Bohan Liu: Yeah.

Steve Hsu: In the country. So, so can you you wanna reconcile those two perspectives?

Bohan Liu: Yeah, 100%. Yeah. And, and by the way, so yes, from like an asset return perspective, you know if I maybe were a pension fund, and I were very intellectually honest, maybe I would just put my money in the S&P 500, right? I wouldn't give it to a single private equity manager or hedge fund guy.

And I believe the state of Nevada chief investment officer is one of those public heroes where he basically takes like a minimum wage and, and just puts the whole state pension fund in the S&P 500, and he outperforms all these sophisticated money managers 'cause it just works. But on the other hand, I would say our asset does have these kinda like second-order effects that are positive for society.

And also because of that, there's kinda like this like sexy narrative premium pay for it, right? Where even if the fund manager under-underperforms, a lot of LPs want, you know, want exposure to cutting-edge technology, right? Like private equity has, has a bad name. You're just like rolling up companies and firing people.

But like this is like cool and sexy in the future, right? And so that's why there's kinda always flows into this asset. And, and, and by the way, like a lot of asset managers in VC have underperformed. And guess what? A lot of them can still raise funds because there's so much capital in the world that has is down bad. They're like, don't have access to Sequoia and Andreessen, and they never will, and they just wanna get any exposure they can to Silicon Valley AI, and so they're still willing to fund people.

Steve Hsu: Yeah. Well, right now is a bubble time, so this, this time is like 2000, but 10X 2000. I do wanna comment that in a lot of the social circles that I move in, I meet really rich people, and they seem more excited to just say, "I'm an investor in SpaceX."

Yeah. "I have share, I have a stake in." You know, and it's, it's through the venture fund that they invested in that got an allocation. And to them, the pleasure they get of talking about that Yes pays for the, they're not even looking carefully at the IRR or whatever, you know? It's just like, "Oh, but I get to say

Bohan Liu: 100%.

Steve Hsu: "I get to say that I have a, you know, I was in Anthropic back in their A round," you know, or whatever.

Bohan Liu: I think that's part of the exorbitant privilege, asymmetric privilege of our asset class some of the other ones.

Steve Hsu: Yes.

Bohan Liu: The sexiness and the narrative premium.

Steve Hsu: Yeah. Whereas, you know, again, 20 years ago there was so little venture pre-internet bubble, there was so little venture in this country. You know, there it was clearly imbalanced in the wrong direction, and now I just don't know whether it's overshot or not. I would guess maybe it hasn't, 'cause there's still you know, once you step away from the Bay Area, even if you go to Austin or Boston or Seattle, it's like, it's orders of magnitude less, I would say.

Bohan Liu: Yeah.

Steve Hsu: But in China, are there a lot of $20 bills on the ground where there's clearly some guy who, like, man, if that guy just spoke English and had a Stanford degree and was living in, in Mountain View, that guy would be a dynamite founder, and he can't get anything done, and he's just slaving away for 10 cent.

Bohan Liu: I'm sure. it's 100% true.

Steve Hsu: Yeah.

Bohan Liu: Right? And like, you know, as the macro has gone harder a bit, right? It looks like last few years, reality is post-COVID, the eco economy's definitely not been doing as well as before, and like technology is kinda like the only bright spot in some ways, but that's, that's a pretty small part

Steve Hsu: Yeah.

Bohan Liu: Of the country, right?

Steve Hsu: Yeah.

Bohan Liu: You know, most people are not beneficiaries of that, and that's why Shenzhen's booming. Shenzhen's like kinda like the one of the few cities that's seen like rapid growth, and there's still optimism in the air. Before it was like import, export, e-commerce stuff, like making goods, and then, you know, now a lot more robotics and things like that.

But I think, you know, because of those macro fears and Chinese people are generally a bit more risk-averse, you know, they gotta raise kids, there's more traditional values gotta buy a house. I think people are one, the, the, the capital market's already kinda like less encouraging risk, and then, you know, for a lot of founders, they're just probably not as willing to just throw, you know, throw the dice as, as some Americans are.

So yeah, there's definitely like a lot of probably very smart, talented people who are not doing startups, or even if they try, they couldn't probably raise very well.

Yeah. which the only silver lining is if you happen to be a Chinese venture capitalist, then, you know, you have a beautiful game. You know, you got the best time in the world, right? Making great returns. Yeah.

Steve Hsu: I think one of the things that's gonna happen on roughly a 10-year timescale is that international investors are very used to putting their money to work, you know, dollar-denominated assets in the United States, and very, very leery and ignorant about what would happen if they moved their investment into China.

And whereas they, they are starting to see, like, wait, all the EV innovation came from there. Well other, ex-Tesla. All the EV innovation came from there. All the solar innovation came from there. All the, you know, drone innovation came from so like after a while, you can't ignore it. Like, if you just think about it rationally like an asset allocator, it's like, wait, bunch of innovation is coming from this sector, and we have this part of the world, and we have, like no exposure to it?

Like, I think that's gotta correct. My guest, Louis Gave, one of the Gava Kao founders, I don't know if you know him. He's a, he's more of a hedge fund guy. But, but he's been saying this for many years. His a little bit talking his book 'cause he's, his firm is based in Hong Kong.

But yeah, he's always been like, "Look, this is growing and growing chunk of the global economy, growing and growing chunk of global innovation, and yet most of these Western asset managers have no exposure to it, right? I think it's gotta happen gradually.

Bohan Liu: And I think that's an, that's an arbitrage, and there is a contrarian thesis there.

Steve Hsu: Yeah.

Bohan Liu: But of course, you know, you gotta wait for it kinda the narrative be to fully be priced in and the multiples to return because

Steve Hsu: Yeah.

Bohan Liu: People then have been stuck in China for a very long time and not able to, like, make good money or even despite the companies growing well, the multiples keep getting slashed, you know?

So long enough time horizon it's probably true, but a lot of it depends on geopolitics, right? And like, you know, Mr. Donald Trump and Mr. Xi Jinping are best friends

Steve Hsu: Yeah.

Bohan Liu: and they're having a great time.

Steve Hsu: Yeah.

Bohan Liu: But with the next election, the government could turn more hawkish it could get really hard again.

Steve Hsu: Yeah, there's always that discrete geopolitical risk, obviously.

Now, let's, let's talk about AI since that's, that's the main thing everybody's looking at right now. So just as we started recording, you know, you were mentioning to me, or maybe you did mention it in the interview, that, you know, your article in part took off because of the Kimi K3 release and, like, people, like, noticing again, like, "Oh, wait, there are many companies in China that I've never heard of that seem to be able to build frontier models."

And you know, if you really wanna put your finger on it you could say, "Wait a minute, like, other than OpenAI and Anthropic, hmm, what are these other US companies? Oh, Meta, Google, SpaceX. They can't seem to make models that are competitive with the top maybe three or four Chinese models that are out there right now."

Bohan Liu: Yeah.

Steve Hsu: And it's like, "Wow, how did that happen? Like, should I, should I recalibrate my thinking a little bit about this?" So it, it's definitely a moment where I definitely see a lot of, like, cogitating going on in people's minds.

Bohan Liu: Yeah, and I think there I think people are also running out of copes in a little bit of ways, 'cause even with Kimi, I think, you know, everyone's naive answer is they just distill.

First of all, like, that itself is not, like, that trivial to just distill, you know? It's, like, not like you can just say that one word and can figure everything out. But I think with Kimi K3, I think one, it's probably not really distilled on Fable, right? It came out like too quickly afterwards. And, and two people assume there was actually a lot more pre-training work done there than people would give credit for.

Steve Hsu: Yeah.

Bohan Liu: And pre-training and post-training. So it wasn't just distilling like a drop model. So like they really got stuff going on there, and they're also like now really ramping up primary data purchasing. Like that's why Sequoia China, you know, seeded Yunotpad and Humanlion, which is kind of like the Mercor and Surge of China.

And so I think as they push towards the frontier, they are doing more getting their primary data, training themselves. And yeah, to your point, basically outside of OpenAI and Anthropic, probably all the best models are Chinese.

Steve Hsu: Yeah.

Bohan Liu: Which is a very ironic position right now that Silicon Valley and the global AI innovation stack, especially the open stacks, all run on China, right?

They're fully dependent on the generosity and, and I wouldn't say altruism, I guess they want moochai wants to make money too, but you know, generosity of giving these things out for free.

Steve Hsu: Yeah.

Bohan Liu: And so right now you're in this really interesting regulatory position where there's big lobbying forces on both sides trying to because everyone has their bags, you know, trying to push the US government one way or another, right?

Like all the people who have bags in Anthropic and are afraid of Pario, they got a lobby for blocking it. And everyone who cares about consumer surplus and innovation, you know, yeah, both from Nvidia to Microsoft, you know, to like nearly every startup wants open models because

Steve Hsu: Even OpenAI, even OpenAI came out on the they signed the open petition finally.

Bohan Liu: Right. it'd be so ironic if their name. They've already been last lambasted so many times for being called OpenAI.

Steve Hsu: So it's, it's actually just Dario and some, and some very hawkish people on the other side.

Bohan Liu: You know, the US clearly really needs to have their own domestic models that are frontier and open source.

Yeah. And so I think they're gonna put a lot of money in that, figure it out. Like Nvidia has NeMo Tron, like Thinking Machines has Inkling, and you know, they might get there. And then, you know, then there could be a whole different calculus because if you have parity, of course you would choose an American over a Chinese one. I, but we're all else healthy for it

Steve Hsu: I interviewed one of the Reflection founders on the podcast and yeah, I mean, their bet is basically that one way or another, America has to have a strong open model ecosystem, and so there's gotta be at least one like Reflection-like company that wins out. So

Bohan Liu: Yeah.

Steve Hsu: I was gonna mention something about Kimi, which, Kimi, which is or K3, and also the DeepSeek models that are underappreciated, which is that it's even if distillation were a big component of their ability to catch up, the architecture of the models

Even if it were distillation that was allowing them to catch up, the architecture of their models is quite different, and that leads to the cost advantage, and it also leads to the advantage in train in cost of training. And one of the things I think if you're not a developer you're not aware of is something called cache hit, which is like if, if you're working on a code base and you need to put the entire code base in the context, in the prompt every time, the US labs are actually billing you for that.

They're billing you for those tokens. Whereas the architecture of the Chinese models allow that information to be stored in a, in a cache, and it's pulled up through memory, so it doesn't actually require compute to do the attention mechanism again each time on all of that data. So that's a big architectural difference, which nobody seems to, who isn't, like, really in the weeds technically doesn't seem to understand.

So that's a huge difference for agentic encoding applications, and you can't say like, "Oh, they distilled that from Anthropic." No, they built that themselves and wrote

Bohan Liu: Yeah.

Steve Hsu: The core papers on those algorithms that, that make that possible. But I think very few people are following it at that level

Bohan Liu: Or like DeepSeek had GRPO, I think. You know, these are, like, their own advances, yeah.

Steve Hsu: Yes, exactly. Yeah. So I actually think they have a lot of momentum. I actually do think their main limitation is just access to compute. I think if they had as much compute also financial resources as the US labs, they would actually probably surpass the US labs. I think they're, they're at that point now. I think it's very possible,

Bohan Liu: right? Given how with a handicap they're basically at frontier or near frontier, imagine if they actually had US compute. And by the way, they could run that cheaper 'cause electricity's cheaper there.

Steve Hsu: Yeah.

Bohan Liu: If Huawei could make these things themselves, like, they would just be able to actually have the same amount of compute at much lower cost, and everything would just tilt in their favor in some ways.

Steve Hsu: It's the same thing with semiconductors because what Huawei and SMIC are able to do with only DUV, like multi-patterning, all that stuff is extremely hard, so they, they and the, the logic folding and the 3D stuff, all that stuff they figured out themselves. If someone snapped their finger and they could suddenly buy an EUV machine from ASML, they could potentially leapfrog TSMC.

You know, I was just in Taiwan, and the people in Taiwan are so, I would say, overconfident actually. I think they, they, like, have not updated on what's actually happening over there, and it's, it's actually, you know, as we were saying earlier, like, a lot of Taiwanese doing it actually in, in China. But they haven't updated on the idea that, like, the Chinese side, those fabs are getting really, really good, and the main gap is just no access to EUV.

But if that changed, I actually think they would give TSMC a very good run for their money. They might even beat TSMC. Now, we maybe will never see that situation exactly 'cause there'll be a long, hard slog before a domestic EUV alternative is deployed in China, and, you know, so we won't really see the exact thought experiment that I'm describing now played out.

But I do think, like, what the Chinese side has done is super impressive and, and people are just not aware of it.

Bohan Liu: So I think it really depends on timelines because I think the Chinese are gonna get there, right? In like, you know, five, 10 years I think they will have domestic you know, like high density chip production and, and maybe even sooner.

I don't know when they'll hit their D- EUV. And then at that point, is it just takeoff? Well, of course, maybe in the US you kinda if you're a supervillain, you kinda think AGI takeoff is actually before that anyway. So it's like, it's all a question of timelines, of which curve is gonna hit which point faster.

Steve Hsu: Yeah, 100%. So my debate with the, the guys in the Biden administration who wrote all the sanctions and stuff is my point to them was like, these sanctions are, in the long run, giving the semiconductor industry, or at least allowing the Chinese to build their own domestic semiconductor supply chain.

And in the long run it's gonna be a real problem for us. And you're giving away that long run future because your timelines on AGI are short, and you just think that it's important to keep them down for whatever the five years that it takes for us to get to AGI and then they'll never catch us. I think that scenario, which, you know, the Biden people, the smarter ones, will openly acknowledge that that's their bet.

That is their rationale for why they put the chip sanctions on. But I think things are not looking good for that. I don't, I don't think the gap is enough between the best AI models here and in China that there will be a persistent or really meaningful difference in real world competitiveness terms due to the difference in the model quality.

And so I think then we gave away the semiconductor industry for, for nothing basically.

Bohan Liu: These are hard questions to forecast. I think even the most AGI-oriented forecasters have moved their timeline projections back, a little bit back kinda continuously, right? The AI 2027 people or

Steve Hsu: Yeah

Bohan Liu: you know, some of the forecasters because, yeah, it's not turning out exactly how people thought.

Steve Hsu: There, there's a great interview by this I think he's a Russian guy who's a professor at CMU, and he was the PhD advisor for the Kimi founder when he was in the US at Carnegie Mellon, and he was also the advisor Yeah, Yang Jialin.

What's that?

Bohan Liu: the Kimi founder, Yang Jialin. Yeah.

Steve Hsu: Yeah. Yeah. So his PhD advisor at CMU, there was a great interview on YouTube with this guy, and he was he's also obviously being at CMU and having had many, many students, he's got many students at the big US labs. And he just said, "Look, anybody who's serious working in this area is not telling you that AGI is imminent."

Other than Dario, who I don't know if you should call him serious. All these other people that this guy refers to, this guy's an old guy. He's like my age. He's a professor. He's not gonna bullshit, right? So he's like, "You talk to the people," I guess I talk to the same people, "who are working at these places," and nobody says, unless you're an alarmist guy at Light Haven, you know, or Dario, no one says AGI is right around the corner.

Literally nobody. So, so you know, maybe if they're trying to raise money or something from, you know, right, right before their IPO or something, but I just don't think it's right around the corner. Now, it definitely could happen in the next five years or something, and of course, we have to define what we mean carefully by AGI. But it's not as close as many people believe.

Bohan Liu: I think my timelines have shifted a bit, and I think the valley here is interesting where I think there was a peak moment of kind of permanent underclass fear maybe end of last year as maybe Anthropic was raising their 380 and 960 billion rounds.

But I think people have calmed down a bit. Well, maybe because they're just desensitized and numb, and you can't just be depressed all day. But I think people have also calmed down a bit that like, "Hey, maybe the world is not gonna end that soon," you know? And like, life is okay for a bit, you know?

Steve Hsu: Well, I also want to differentiate between permanent underclass and x-risk. So, so permanent underclass could mean, like, AI creates so much value, and 90% of that value is captured by a handful of companies. And if you're not in those handful of companies, yeah, you're in the underclass. But it actually could be from an in an absolute sense a, an abundance utopia where like, yeah, you're not one of these super, super trillionaire people at the, at the labs, but these AIs are making all kinds of shit for you and curing your cancer and stuff.

So that's not the s- x-risk scenario. The x- x-risk s- scenario is that RSI runs out of control, and the models become so powerful that they can do anything, and they basically, you know

Bohan Liu: Yeah ...

Steve Hsu: get rid of us or do something bad to us. That, that's, that's more the, what I would call the x-risk scenario.

Bohan Liu: Yeah. Well, I think the abundance scenario is still unappealing to a lot of people in tech. One, because they're ambitious, and they always want to be at the top. But two, it's like, you know, so much of human happiness is memetic and relative. And so as we know empirically since the, you know, Enlightenment, Industrial Revolution, a lot of like material measures of prosperity from longevity and food and shelter are growing.

People aren't necessarily subjectively happier because they see all the billionaires on TikTok, you know? So I think people are very worried that, yeah, maybe my cancer gets solved, but I still won't be happy because that person has their beachfront property

Steve Hsu: Yeah.

Bohan Liu: And that is physically scarce, and I don't, you know?

Steve Hsu: I totally agree with you. I think the permanent underclass mentality which pervades San Francisco is a status competition

Bohan Liu: Yeah.

Steve Hsu: perspective where it's like, "Oh, I don't care if my cancer got cured. You know, I don't care if I have a flying car. I don't care if, you know, I have a super intelligence, you know, in my earbud talking to me whenever I want, and it's nice to me.

I just don't like the fact that Sam is so rich and I didn't get that rich," right? It's like, really? Well, if that's your perspective, then you're just really a shitty human being I think personally. Now separately you could just say like, "I don't believe in that level of social inequality, and the government should take control and tamp down the Gini coefficient or whatever." That's a that's kind of a separate thing. But viscerally I think it's mainly just status jealousy, from a lot of these people. Yeah.

Bohan Liu: Yeah, and you know, SF is probably one of the most kind of status prestige hierarchy driven places in the world to be honest.

Steve Hsu: It's insane. Yeah. Crazy. All right.

So I guess we've talked for about an hour. maybe I think we covered all the topics. Is there anything I didn't ask you that I should have asked you?

Bohan Liu: Going into the China trip I was very much infused with a lot of the Silicon Valley, I think narratives that, wow, the Chinese and I'm, I'm sure you, you talk about, a lot about too, is the Chinese are just destroying us in so many ways, right? They're, like, leading in biotech and all the, you know, like, most American clinical trial drugs are, like, licensed from there. The majority of Chinese have gained drugs are, like, novel therapies.

And then on robotics, they literally, they build hardware as fast as Americans build software, you know? And it's like, and they have data advantages in terms of producing tele-op data and human data, and so they're structurally advantaged to finally on AI. Look at that, like, with less chip they can, they can make Kimi K3 or GL 5.2.

So I think there was a lot of narrative that China's just gonna dominate everything. But I feel like coming out of that trip, a visceral reaction was that it didn't feel that way. I mean, first of all, Chinese definitely didn't feel that way. They very much still look up to America, and it's definitely true that the average American is more hawkish against China than the average Chinese is against America, right?

Like, on average, they're very friendly. They love America still. Like, day-to-day if you visit, they really want to learn from the Valley and still look up to the Valley, despite actually, you know, beating in certain ways. But, you know, maybe they're just a more paranoid or humble people, but I felt like the game really wasn't over.

Like, you know, America still had many advantages, and I think a lot of people there were clear-eyed America's advantages, which includes China's sorry, America's capital markets, which are just so much deeper and truly is advantage in terms of, like, high CapEx technology like AI robotics. But also, you know, a lot of the frontier algorithmic breakthroughs, I mean, you still come out of here.

And so I think, you know, if Americans are really worried about just losing China every way, I don't think the game is over. You know, of course, China's also not gonna collapse. I think there will be some new equilibrium of these two global powers and, you know, both sides will have to learn to accept this new norm, which may be harder for Americans to stomach in some ways.

But I think there was a lot of fear-mongering here too, and I don't, I don't think that's the end of the world. And, you know, I still want to see a lot more, you know, exchange of both ways because, you know, people really do wanna learn from each side, you know?

Steve Hsu: I agree with that. I think actually it's, there's still plenty of room for win-win cooperation between the US and China and and certainly avoiding the worst case scenarios is, is super important for the next decade or two.

You know, I would say you're right. The, the very few people in China are super triumphalist. The, they, they might be triumphalist in the following sense, like we knew when we started closing in on America, they were gonna try to crush us because they weren't, they weren't gonna tolerate us, tolerate us getting to parity with them.

But now the triumph is we're confident we will survive that. Trump took a shot at us. Whoever comes next is gonna take a shot at us, but we're probably, they're probably not gonna be able to keep us from rising. That's different from saying like, "Oh, we're so far ahead of America. Americans should lose hope.

America's not viable." That, that's a different statement. Nobody in China actually says stuff like that. People just say like, "They're not gonna be able to stop this from happening. Yeah. I think that's the level of and, and not everybody, not even not everybody there is that optimistic, right? Some of them are probably like, "Yeah, we're kinda, we're gonna kind of get, end up like this," right?

So a triumphalist in China is someone who thinks, like, this is gonna happen, right?

Bohan Liu: Yeah.

Steve Hsu: I would say that even people there, and even people who are there leading, like, K3 or Kimi or leading DeepSeek, and maybe who even study in the US, everybody tends to use lagging, backwards-looking indicators.

So it's still true that the top-level scientific research is stronger in the US in some sense, that science culture is still stronger than in China. It's still very nascent, building up in China. But the amount of talent they have is just insane. And so people might be under-emphasizing, underrating, like, what is about to happen next, and they're kind of a little bit backward-looking and saying like, "Oh, still Silicon Valley's the place, and still, you know, MIT and Harvard are the place, and, and not Tsinghua."

But I'm not sure that's actually true. I think actually if you look at the momentum,you know, those guys may surge ahead. I won't be surprised if they surge ahead in the next decade.

Bohan Liu: I mean, the raw talent, it's there. I mean, the pipelines truly are are definitely in place. And so, you know, a lot of them, of course, still go to American grad schools, but as the tide shifts where maybe a lot of them just stay there, go to local grad schools, or, you know, more people who go to American grad schools go back, I think so many things are still downstream of human agency and talent.

I mean, American AI is still built by Chinese people too anyways, so I mean. So, yeah, I really don't know how it's gonna go. But I think these two countries are just basically the only places in the world where you have real innovation going on.

Steve Hsu: Yeah.

Bohan Liu: Europe really doesn't have anything going on. So, you know, so I'm, I'm very bullish on both these countries and very interested in how they continue to interact.

Steve Hsu: Yeah. I think it's definitely the, you know, China-US competition is the most important and interesting thing going on in, certainly in geopolitics, but also it impacting everything including AI, and, you know, this energy transition.

I think most people who think about climate, you know, they think about, "Oh, it's so hot in Europe this summer, we're in trouble." But if you really look at the unit economics and what the Chinese have now done with solar panels and batteries, it's kinda clear we have solved the problem. Like, I don't think people have realized this yet, but we have the whole set of technologies and it costs where we can decarbonize now, and that just kinda happened without people noticing.

If you're a physicist and you look at the numbers, you realize that, yeah, we can decarbonize now. Before it was like, we might be able to decarbonize at great cost, but now we can decarbonize. And a lot of these developing countries are gonna power their houses and, you know, they're not gonna build the same

Bohan Liu: Yeah

Steve Hsu: kind of grids that we had to build in the past, and they're suddenly gonna have electrical power and air conditioning at home through a whole new set of technologies that came out of China that just weren't economically viable

Bohan Liu: Yeah.

Steve Hsu: You know, 10 years ago, so.

Bohan Liu: Yeah, eventually, you know, people won't be able to have a stranglehold on the Strait of Hormuz anymore, right?

Steve Hsu: Yeah.

Bohan Liu: like, oil is no longer

Steve Hsu: Yes.

Bohan Liu: like a bottleneck on the whole world, eventually.

Steve Hsu: Yeah. Yeah. I mean, and we're gonna use all that oil, but I think you can see a future where we won't actually need it.

Bohan Liu: 100%.

Yeah. All right. Well, hey, thanks for being on the show, Bohan. Pleasure's all mine, Steve. It's always great to chat with you. I definitely plan to go back to for more trips and hope to share more. I'm sure you'll be back there too, and I hope all your listeners get a chance to go see for themselves, really just eyes on the ground. Just go visit even as a tourist, and I think people come out with just a different perception of, you know, probably one the only other most relevant society on earth right now.

Steve Hsu: Great. All right. Thanks a lot.