Host: Lambert, who has the best China open source model? If you put a gun to my head, I would say as of today,

Lambert: ……the biggest disadvantage that the Chinese lab still face today, yeah, it's compute. Meta or OpenAI have more GPUs than all of these like Chinese companies combined.

Host: So, how good are domestic accelerators such as Huawei chips?

Lambert: All of those companies are like, if you can give me Huawei chips, we will like maybe get more customers.

Host: Anthropic CEO Daryl Amade, he's a little too aggressive in the near term about risks posed by open weight models. How many months do you think US SOTA models are ahead of the Chinese models? And for that gap, is it going to become narrower or wider?

Host: Hugging face. Analysis. Fore. Next. Nathan Lambert. Hello, Nathan. Welcome to Valley 101.

Lambert: Hey, thanks for having me.

Host: You just came back from a trip to China during which you actually visited many of China's leading AI labs. Tell us more about that trip.

Lambert: People always ask me now that I've been back. It's like what are your big takeaways? And I feel like AI is fairly well known and who is building models and things. So, there's not that many things that are like super shocking. You're like, oh, I have this hot take. The trip for me as somebody who covers these models, it's to get to know the people who are doing it and get to pick up on the nuance of why are they doing this? I think I'm just learning a lot this year and there's a lot of people in the US that will just say why they might be doing this and just to get the first is very nice. I expect to keep covering open models in the future and trying these models is an interesting ecosystem that's set up that there's a lot of interest in the US and companies that are building around real open model intelligence stack which is serving value for all sorts of people and then a lot of the models are getting built in China and these historic tech communities like have not intermingled a lot and how they work. So there's traditional policies at places like Meta and Google where if you travel to China, your laptop automatically breaks itself. And there's like been a wall where people can't easily work together, but they're becoming integrated. And I'm just very interested in it. I think it's surprising in some ways that these models are so competitive, but then when you learn more about all the talent there and things, it's like it becomes somewhat normalized. And I personally like a lot of the reason that I'm motivated by AI is that I would think that in the next couple years it's like a lot of risks particularly like just social risks and how people are informed and then also geopolitically it's obviously a topic of interest. So just building bridges across people and humanizing the people that are actually doing this because they don't make a lot of the decisions that are listed as political or extreme in any way and just like building these connections is really useful because it is one community that is kind of moving forward towards these various goals whether you call it AGI or whatever you want to say the goal is. So I'm just like very happy to meet the humans behind this.

Host: Yeah. Great. And how many days did you go and which cities?

Lambert: Yeah. So I think it was about six or seven nights and it was primarily in Beijing and Hangzhou. So it was like three or four nights in Beijing which we were staying right near the Olympic Center. So like go for a jog in this the bird's nest stadium and all these things. It's just like where you find a hotel but then most of the companies are by Tsinghua University and there's this whole cluster of AI startups there which is really interesting. It feels a lot like the Bay Area where you're like oh I can just walk between all the leading AI companies of the region and like you don't have to do anything. It's like a very Bay Area feeling. And then we also went to Hangzhou for a few nights, which is where Alibaba is. And I think Deep Seek is. We didn't go to the Deep Seek office. We talked to some people and we did visit Alibaba and it's just great. It's my first time in China and nice to see them.

Host: So, which labs or companies did you visit this time?

Lambert: Yeah, I'll try to list many of them. We visited also a bunch of robotics companies which I didn't focus on as much as writing but for language models we visited well I went to two different Alibaba offices. I got off the plane in Beijing and immediately went to meet... Moonshot as well Zai and then Florian who writes Interconnects with me. We went to Meituan which has their language modeling effort and then we also visited Tsinghua University and we visited the Xiaomi campus in Beijing and then when we went to Hangzhou we visited the Alibaba campus there the Antling team and also Model Scope which is also affiliated with Alibaba and then this was a group trip part of mostly organized around this like "Sale" which is a bunch of Substackers that write about AI which is like my friends making a vehicle to do interesting things in AI media and events. So, it was mostly organized around this and then the rest of the group in Shanghai talked to I think definitely MiniMax and maybe Stepfun as well. So, there's really a lot of the names. I think there's obvious missions like DeepSeek is the most secretive one and ByteDance is also like you visit them and they're like all the companies talk about DeepSeek as their like kind of mysterious leader and then ByteDance is like oh the one closed AI lab that has a ton of resources that everybody's like worried about. the Doubao chatbot is the dominant thing. You pick up on these things pretty quickly when you're there because everybody talks about it.

Host: So in your Substack article, you said you returned with humility. What does it mean?

Lambert: Yeah. So I feel like as an American scientist, I never really expected to have anything to say about the Chinese tech ecosystem or US-China interactions at a global level. I described it as like I knew I didn't know a lot and I left and I was like I still have so much to see. It's like I saw these two cities in China and I was on like a busy work trip and feeling like you get teleported around in a van. We went to some great dinners and we saw some cute neighborhoods in Beijing and some of the older buildings and stuff, but it's like I feel like I don't know so much about China. And to like truly have immediate instincts and intuitions about something, you kind of need some more closer to full stack knowledge and appreciation for like what would a person's instinct here be? like what would a Chinese researcher's instinct be? And it's like it would take a lifetime for me to fully understand that as an American as things like how is the education system difference. We went to like Tsinghua and they would talk about like we want to they would acknowledge the reputation where it's like Chinese researchers haven't been as good at this like zero to one style science which is kind of an arbitrary term. It's like what does that actually mean? But they would bring it up and part of this is in like how people are trained and how the talent gets funneled through different layers from a much broader domestic network than in the US. It's like these are things that people would talk about and actually seeing them is useful. This year I read Breakneck and Apple in China is these two wonderful books on the US tech ecosystems interactions with China. But that's still like a very limited view. So that's why I'm like okay I kind of expected this when I left that I would be like I don't really understand everything I left but I still feel like I have more to learn.

Host: Were you meeting with the senior level people or the middle level AI researcher or engineer level people? Do you think those conversations actually helpful?

Lambert: Yeah, so we were targeting meeting with more researchers just to get a dialogue. We also targeted trying to get leadership especially at startups. So it seems like mostly starting to talking to researchers. We talked to some executives say like Alibaba or some of these people that we talk to are very well known within companies and interface directly with the founders. But I think...

Speaker A: When you're going to do like a formal visit, it's a lot of showing respect to the people that are there. But at the end of the day, like the leaders are more media trained and sometimes will tell you less and that kind of suppresses nuance to stay on the company line.

Speaker B: So I think it was some of both. The bigger companies I think were more like open to show some of their senior cloud people without saying like without giving too much away. I think some of the conversations were off the record. I don't know. It's not the best answer, but it's a mix.

Speaker A: Yeah, especially with the coms people around.

Speaker B: Yeah, like most of these have comms handlers in the room. Like some of the best conversations were I announce that I'm here. You're I'm on WeChat. I message people and then you like go get Bobo with a random researcher, not on the corporate office, and you're just like, "What are you working on? Like, what are you doing?" And then it's like, okay, there's not the veneer of like I'm at the office, my manager is sitting next to me, the comm's people are sitting next to me. So those are almost more fun but more rare of the trip.

Speaker A: Definitely. Um so what kind of questions did you mostly ask them?

Speaker B: There's a distribution. So I think that I was interested in to start with something that's probably pretty accessible to them which is like why are you releasing these models? Like why do you build your model? Why don't you just use Deepseek's model?

Speaker A: A lot of these companies are very practical and it's like we need AI to serve our users and we want to own the full stack and therefore we're going to build it. and they're much more daunted about building AI models being this like super expensive impossible thing where it's kind of an allure in the US of like being a frontier lab. It's like restricted to an elite few but it's like all the companies that's why it was like we wanted to go visit Muan. It's one of our favorite like random Chinese tech companies to have big models. So we're like just go see how they're doing. They're like yeah we need this to serve our agentic products. So we build the model we need it for products. It doesn't matter if it's open we'll release it as well and get some feedback. and they're very practical about this open science sort of idea.

Speaker B: And this would lead into different things like what are your current bottlenecks which can obviously quickly be a dead end on compute or but a lot of them are like Huawei chips are good for inference. We use Huawei for inference but not really for training. So like a lot of people would say things like this. Also in our group people would want to talk about like what is the coming future of AI? So like what do you view on how it's going to impact labor markets and economics and like do you feel like you have a responsibility as a scientist to understand like think about these impacts and that's where we saw in China that researchers are much less inclined to share a lot of views on this. Some of this is the corporate environment, but also I think it's just like a environment that's much more structured and what they're encouraged to talk about in in their education, which is like I built the thing and there's much less of this like podcast ecosystem in in China where there's researchers that are like rewarded for being visible and trying to sway the direction of the company. It's like a bit more practical-minded, which I think I sometimes in my writing got over overlapped with the humility, whereas like you can be a humble person, but also you could be in an environment where you're just not used to being asked certain questions.

Speaker A: Cool. Let's talk about some of the companies you visited. In your opinion, how different were they from one another? Let's talk about Moonshot first, shall we?

Speaker B: Yeah. So, I think they do have pretty distinct feelings when you could go into them. I think Kimmy was one where it's like These people seem so close and they're just in it for kind of the thrill of building a cool thing and they're building a business alongside building this cool thing, but they're also just like we're going to build these models. We use them ourselves and they seem very close in a way that's like you just pick up intuitively when you like go to the office. You're like, okay, these people are really bonded and they're tight and they're just like spend their lives together building this. I describe Kimmy as like the best vibes company of the one that we visited which is just like they seem like they're happy and they're going on their path and it's not very corporate in a way where if you go to obviously Alibaba it's like this is a huge tech company it's very corporate the ant group Antling team has some of that because they're also a huge tech company.

Speaker A: but even in some of the middle it's like Zai was very AGI forwards and they were like had a lot of pride but at least in the interactions we had it wasn't as much of this like charm that Kimmy and Moonshot have. And some of this might just be like random daytoday like I've met these people for a few hours over lunch. You can't conclude a company's trajectory based off this. But it's also interesting because like people are expressive and you can learn things from that.

Speaker B: There was a paper from Moonshot named attention residues with one of its authors being a 17-year-old high school student. You know about that one? I think that paper was reposted by Elon Musk on X saying impressive work from Kim. So this high school student got very famous in China now. So what does that tell you about the culture at Moonshot?

Speaker A: It's like less of a moonshot specific thing, but the this the talent there is very young. These companies are all on a push to have more of a western media footprint and there a lot of these researchers are on X and it's like I knew a few of the people. I think that it's like people that do a mix of product and comms at the company. One is like Crystal, I think at Kimmy, and then Lou at Zi. And it's like I met Lou, I don't know if I'm pronouncing her name exactly right, but she's like 20 and she's still a student. She's like, "Hi." I'm like, "Wow." It's like these people are so young. And then one of the Xiai researchers we met, um, Li is like also a PhD student and like one of the lead researchers at Xiaomi. There are some people like this in the US which are just extremely motivated to end up at a frontier lab. But it just seems like we're getting so many when you meet one after another in China of these super young researchers deeply embedded in this labs. It felt like a bit of a structural just like they absorb this talent much more quickly than other labs and are more open about it.

Speaker B: It's wild. I mean like I'm in my early 30s like these people are so young. It's like I feel young in my life and it's like

Speaker A: it's cool to see. But is it different here in the US?

Speaker B: I think so. Like I I think like young talents here in US labs.

Speaker A: Yeah. So something that defines AI2 a bit, it's like secret sauce is that like Udub is behind you and we have a lot of PhD students that have been like core to mo and core to other major projects who are concurrent students. But that sort of open door is at least like less fluid at some of the other labs. Like I know a handful of PhD students that work at labs, but these are like the absolute best of the best in the US. Where in China it feels like the intermingling of academia, especially at a place like Chinua and these startups nearby is much more intentional and much higher overlap for the talent.

Speaker B: Why does that matter?

Speaker A: Students are young, they're hungry, they have nothing else going they have much less going on in their life. Like I'm think I'm getting married this year, thinking about family. Like I'm 30. I have other things but students are like this is the technology of my life. I'm just going to work all the time and absorb all the context. And when you build a language model like I work in post-training, you need to know what the pre-training team is doing. You need to know the multiple stages of your recipe and then you need to figure out like what is the bottleneck and go address it. So

Speaker A: There's a really big fixed context of technical knowledge that you need. And I think it's just like if it's the biggest thing in your life and you don't have a lot of distractions, it's just easier to put all of this into your brain. One of the researchers told us like, oh yes, new students don't have priors for what used to work in deep learning. They're not looking at a different architecture or anything. They're just like, I want to know what the current state-of-the-art is and make it better. And it's kind of like a simple approach that they can do. Some of the researchers in China told us this. We're like, yeah, we that's what we're doing.

Speaker B: I see. I see. Let's talk about Jup. How was the trip there?

Speaker A: It was fun. I think you like. So part of the thing that I learned is that Chinese companies have like a showroom culture where when you have visitors you like show the history of your company and JIPU's showroom is all like every panel has some different like AGI feature so it's like AGI loading or it's like their thing is that AG we're like 42% of the way here which is from the Hitchhiker's Guide to the Galaxy reference and you just like immediately notice it you're like whoa and but at the other time there like I felt like It was I didn't get to know them quite as much as the Kimmy people and but they're very proud of their work and they're they're very successful and I think it's some things of like being public and being randomly entity listed by the US. They're just like immediately vaulted to a wa you're like a real big company status type of thing that Kimmy hasn't quite hit that level of.

Speaker B: But in my mind they're so similar. It's like when I look at the models, they have different tradeoffs and it'll be interesting to see if that kind of diverges the cultures a bit more. If Kimmy stays private longer and Jipu is staying public,

Speaker A: Being listed on the secondary market probably sometimes make a difference just normally get more attention.

Speaker B: Yeah. So, JIPU and Miniax are public now and it's good because they can raise short term, but it's like what if they need more capital to raise models? It's harder to do like you don't really do like a second public offering. It's kind of unclear and right now the markets are really rewarding any company that's public. So there's so much demand to invest in it but like that could turn.

Speaker A: So like right now it's good.

Speaker B: I don't know. I don't make those decisions.

Speaker A: But you can make arguments both ways.

Speaker B: Earlier this year there was a surge of API requests on open router for Kimmy as well as Japur uh during the open call sensation. Do you think that trend will continue?

Speaker A: Yeah, we're seeing kind of multiple markets emerge. You have the cloud code and codeex which is like the top end. I describe it as like knowledge work tools, but essentially there's going to be agents at price points for wherever they work. The thing that makes open models nice is that the price point is so low that the experimentation is much more doable. It's kind of a natural place where a lot of people will experiment on these models that are known to be strong and a step down in price while still having these established brands which is where this whole Deep Sea Japu Kimmy Quen kind of circle lands. Personally, I think like OpenClaw itself is a bit of a a fad where the name but the type of agent it represents will be sticking. So like right now like Hermes agent is far more popular but it's like even this will like come and go and there will be another but it's just like this type of agent that you can run it and it is not super expensive on these like hosted models on open router is is interesting. Open router itself is not a huge proportion of the actual inference market. So like in the US like fireworks and together are much bigger. They both started sharing some numbers on the amount of tokens they're doing every day. But open router is like more of like community oriented. So when you think about people that are following open models, that's like that's where they think and that obviously has an advantage and I think companies are very actively leaning into it because when their models surge on it, they use it's like good co-branding and it feeds into itself.

Speaker B: And let's talk about your trip to Muan and Xiaomi.

Speaker A: Yeah, it's uh interesting because as we discussed the story is very different. China is almost like Uber and Apple are developing their own open-source models but even in the US they are not. Why is this the case in China though? When you ask Muan when you like Mtoan and Xiaomi's corporate position would be like we're going to have agents that are used you could think of it in Maidtoan is like an agent could do something for a user to figure out the ride to request or thing to order like it can do this.

Speaker B: And the first instinct in the US consumer would be I'm just going to get an API key.

Speaker A: Exactly. But in China, it's very much like we can build this and this will be cheaper. Like we have compute, we have talent. Why don't we just try to build this?

Speaker B: And this isn't specific to Mwtoan, but there's a few companies that are like we build these general models.

Speaker A: And that's not where like our use of them ends, but we can release the general model because more people will use it. And then what they do is they like fine-tune this to make specialized agents and specialized models which they use for their services internally. And Mron's a good example where it's like those specialized models they don't release but they hope to use them in their apps and you can imagine this with Xiai where it's like it could power features in the car and I'm guessing like Xiaomi works like a Tesla or something where it communicates with the cloud and there's all these interfaces and you don't really know what this is be but AI is going to be intermingled with like all digital technologies. Yeah, I think Xiaomi is another good example of big corp is trying to do everything. It is doing like and robotics models, robotics hardwares. It is really like trying to do every single thing itself.

Speaker B: Yeah, it's like Xiai's models are pretty new too. They released a like 7B their original Mimo model in April to June of 2025 and then come December when they had their flash V2 model which was very popular and then I think 2.5 was recently. It takes a long time for people to realize that these models are going to keep coming and like it's definitely surprising to see a new lab come from scratch and they hired I don't remember her first name but like Fuji from Deep Seek and Lui. Yeah.

Speaker A: Yeah.

Speaker B: And they seem on track to be like one of the best model builders from China. People in the US are mostly still at like Deep Seek, um, Quen, Kimmy, and Zi. Some know Minia, some probably don't.

Speaker A: And it's like it's such a big list, but I expect having visited Xiai, like I expect them to continue like very nononsense. It makes sense for them to do this. They've done so many incredible things in the past. It's like building an AI model almost seems easier than building some of the cars and stuff that they have built.

Speaker B: I think for Muan they also invested in Moonshot's latest fundraising which indicates M20 is betting big on AI. Chinese companies just tend to have this massive ecosystem. They love the super app. They love the ecosystem. They love to build everything from scratch themselves. Do you think that's the right approach to go during the AI era?

Speaker A: It's worth to like try your hand at it. So I think AI because it's so resource intensive, it by its nature will benefit very large companies and companies that can get onto this flywheel of like we have a lot of resources. AI makes our products better. We get more returns on those. We can keep investing will win so much. I think in the US it's really obvious to see like Google is winning. Open AI and Enthropic will be fine. Like Nvidia obviously like all these companies that are in this feedback loop are going to be so good. And then it's interesting to see how this works in China. And I think like Ant Group is along this way where they're building their models and they see how their platform spreads like right now it's spreading into health and they

Speaker A: talked about this a lot and they will surely also use their models there and it's like for them I think this actually makes a lot of sense but when you look at a western observer you're like why is antling a thing but they don't realize the like broad service area that exists when you're using this alipe thing. So, it's like I think for these companies with very large footprints, it actually does make sense. I worry that it suppresses some of this competition. So, it's like how did the smaller Kimmy's, Minia Maxes, and Zis like fit into this ecosystem where such big apps exist? I think like I forget which one of the companies, but they were telling us how they like built an agent to navigate one of Bite Dance's apps and then Bite Dance banned them because they couldn't take the data. And it's like there's interfaces emerging between the small and the bigger companies where the small companies are trying to find their niche. And I feel like mtoan's in the middle where I don't have a good grasp from my perspective like how big of a tech company it is and is like is it going to get swallowed by consolidation or is the spend going to be too much on the language models where I don't really know but I'm pretty bullish on AI as a business. So I feel like most of these companies that are early are going to be okay and sell they sell a lot of tokens. They sell intelligence and there's going to be a lot of demand for it.

Speaker B: You wrote that Chinese companies have a technology ownership mentality. Can you explain that a little bit?

Speaker A: This goes to some of the like super app ideas and they just see that they can build things and make it better and if they succeed they'll put more money into it and keep making it more efficient and they'll tune it to their use cases. And this is like how the many domestic industries in China have grown over time to be some of the largest in the world. this is how it's kind of like they want to build rather than buy if at all possible so they can kind of own the long-term trajectory. It's like a very I think it's like Dan Wong of Break Neck said that like China's capitalism is more intense in the US because it's like they're thinking of even longer term of what gives them the best margins and the ability to provide products and beat competitors and like you don't beat somebody if you're relying on their service and if you're hyper competitive you have to do it all the best and do it all yourself. So that's why I think of it as like this ownership mentality of where it's not currently an impossible line item. I would say that they're probably spending 500 million to billions a year to make these models which is a substantial amount of money but not impossible for these labs. It seems likely that the market will consolidate eventually if progress on capabilities keeps going up. It's like capabilities is like log of investment where it's like you have diminishing returns but if you invest more get smarter but it's an interesting time where it's like I remember I interviewed Richard from Mantling and he was like yeah it took us six months to set up this team to build these large models that's because there's all this talent in tech already and they just like move the teams to do this.

Speaker B: We're in the phase of like I don't know how long they will keep doing it but for now it's like kind of a fun thing.

Speaker A: Let's talk about like some other of the companies you visited. So, and Quinn and Alibaba, um, it's a big corp. It's different from OneShot and other smaller labs, right?

Speaker B: Yeah. So, I think Quinn has won the hearts and minds of open source AI by releasing these small models consistently with incredible performance at all these sizes. They kind of were doing this before Llama 4 and before Llama Faded, but they like really hit their stride right around the same time as Llama 4 happened in Meta Paused. And they know this. So we asked them, it's like they think about things that developers will use. They want to keep building models that developers use and thrive on, but it's like when you look at Quen, they haven't released their biggest models for a while. So it's like Quen Plus and Quen Max have been closed source. And this is very much like the cloud side of Alibaba talking where just like the US like you see Google Cloud's crazy expenditures and crazy earnings. It's like Alibaba is the default in China for that to happen and they know this. So like they're not going to release a model like they might tell us they might not release it because no one will use it but also it's like I'm going to be like look you can make a lot of money on this and I think there's a lot of money to be made in AI but I think the most interesting thing is hearing various people talk about them being intentional about these small models and it's like not an accident maybe it was at the beginning or like oh let's try to do this thing somebody else is doing but they have a very refined recipe of these 1 to 10 to 30B models that are very popular and that's intentional and it's kind of nice to just like hear how they like they're like yes we do this we understand it's a time sync so it's hard to release models it takes a lot of work in the ecosystem and I think they felt that stress of like we could build more models but if we build a model we have to support it we have to go to BLM we have to go to all these companies in the open ecosystem and like the modeling team is still pretty small and it's like they don't have a way to like outsource that work it's somebody build somebody very close to building the model still does There's the two ends of caution is like you can release too many models and also like they haven't been releasing the biggest models because they want to monetize them in the cloud.

Speaker A: Part of going into China for this trip, one of the questions was like is AI investment kind of like a a SAS like this software as a service thing or closer to a cloud spend, right? Because one of the whether or not it's a trope or actually true is that the views in the US is that China doesn't spend on these these services where it's like the Slack equivalent is much smaller and people will maybe just like rebuild minimal software in house or download something and much less willing to pay this like per head per seat monthly cost on software. But all these companies do run on the cloud like Alibaba has a large business and like a lot of modern software you need via the cloud. I think that China will shift to be more of viewing that AI spend is like this cloud spend. It might not be as exorbitant as the spend in the US, but they'll view it as needed spend to make the technology that they want to make.

Speaker B: So therefore, it's like these clouds will exist.

Speaker A: It's so interesting because when you arrive at the Shanghai airport or you take the highspeed train to Ho, you will see the Alibaba cloud like a commercial banner like I walked into Beijing and the first thing I saw was like Alibaba cloud plan something. you know like okay here here I am.

Speaker B: I think one difference between Alibaba cloud and AWS is that actually bedrock carry every single big model and their own model and smaller models at the same time but for China cloud they only carry their own models. That's a big difference. And I think that this Neocloud market in the US where there's just like a lot more people reselling other people's models as tokens or reselling compute is like a huge part of the US market. You could name so many companies. It's like Corweave, Nebius.

Speaker A: Yeah.

Speaker B: Together, Fireworks, Lambda, like there's going to be way more. We talked about Open Router and like I don't know any of the equivalents of those in China.

Speaker A: I don't think. So that's a those companies are largely looked at as being supported by Nvidia in a lot of ways. Like Nvidia likes to prioritize the diversity of the ecosystem because it's better for them to not have most of the revenue be from opening eye like for those few companies and I it seems like that dynamic is not the same.

Speaker B: Yeah, that one is interesting because you're missing this layer in China. they don't have the infrastructure or the optimization

Speaker A: because you look at the likes. So what a trend we're seeing now is like cursor took the Kimmy model and they postrained it within the fireworks like RL sandbox training environment and serve it with fireworks and like that's a really interesting multi-party collaboration to build a like a very strong model and product which might have to be like more verticalized in China to get the same thing done. probably like Chinese companies just want to have everything in house.

Speaker B: Yeah. It goes against the like the ownership and the integration where it's like in order to support a company that operates in these middle layers, you have to be okay with like offloading a important thing whether it's GPU hosting or inference because you want to focus on something else and that just seems like a more western startup style of things than otherwise.

Speaker A: Yeah. Yeah. I do think a lot of the model labs in China are going to be more willing to try different things than the bigger older tech companies in China because like Kimmy knows cursor trains their model like they will go talk to cursor and stuff like this

Speaker B: and um 01.ai AI founded by Kafuli. They now experienced a major transition to a total like 2B kind of business model. Where do you see the future for China's 2B market during the AI era?

Speaker A: It goes back to if these relationships can kind of be formed. So I think a lot of them are wondering if they can hit the kind of use this global to business software market. I feel like I like it's so hard for me to make a bet domestically whether or not it'll win. I think that like yes, you can take AI expertise and build it within another company and it'll be very valuable for the company that had less AI expertise, but it's like whether or not this willingness to like you're losing ownership if you're relying on an external partner to do this. It's like the latest trend in the US is like forward deployed engineers is this buzz word FTEEs. It's like opening eyes, hiring a ton and stuff like this where it's like going to happen in the US, but it's like is that going to be a like viable thing where you can go through the front door of a company and actually get welcomed to them? It's very interesting. It's like they're going to try this and I think that like if it works, being the first one to retry it in the AI era is going to obviously be so useful, but I'm like I have no idea. like I could flip a coin like it depends entirely on kind of this like Chinese culture the business that I can't fully understand.

Speaker B: Yeah. But did you who whom did you meet with for

Speaker A: I we met with

Speaker B: Kyouli directly? Yeah. Yeah. Yeah. How was the conversation?

Speaker A: It was good. He's I mean it's like such a legend to meet him and it's so cool to be able to see people like that and he's like it's focused like he could do he's done so much in his life. he could do whatever he wants or not work. And it's like to see him like trying to solve a problem and figure out how to reimagine his business is like a very AI type thing where it's like it's just drawing all these people back to like try to think about what the next phase of building businesses is going to look like.

Speaker B: Why Kyu's lab failed to become the leading lab in China?

Speaker A: I don't really know. You should ask him.

Speaker B: Yeah, that's a tough question. Yeah, I think there's a lot of opportunity for doing things with AI and building models is a different type of culture. The building models is very much like turning the crank and continuing to ratchet up the resource rungs. And I don't think it's just 01 AI that has like built a model and then stops. I think like in the US like data bricks was building models and then they were like no need for us to do this. Other people have got the bill. I don't know if they're on the record, but I think a lot of leadership at data bricks would tell you this where like we're happy to not pay for building this model and we're going to do our business. And I think that that's why I'm surprised that so many companies have kept ratchet or they like 1 trillion parameter which is way more expensive like 01 AI and data bricks got out at like small or small dense models. It's like way less investment.

Speaker A: Yeah.

Speaker B: So in some ways it might seem smart if you thought that you were not going to be competitive. You better it's better to stop earlier than later. Not everyone needs to have or to to build the best model, right?

Speaker A: Yeah.

Speaker B: A cliche question. How many months do you think US SOTA models are ahead of the Chinese models? Um, and for that gap, is it going to become narrower or wider? Why?

Speaker A: So on traditional benchmarks, it's been like six to nine months for a while. In real world use, especially early in reasoning models, it was very tight because it was a new paradigm. And it's like when you're in a new paradigm, the rewards to doing it are so steep. Both US and Chinese labs are climbing this really fast rate of improvement and the Chinese labs release sooner. So you talk to somebody at Chinese lab, they're like, "Our model's still in RL. It takes about a week to do this last RL run and then we'll release it within like a day of it being done." Where the US labs have some process that normally takes weeks to get the model out. So if you think about if you're on like steep progress, you have two different lines and it's like if China is behind but you don't have this like no progress horizontal is when you're waiting to release it. So China releases it faster. So it kind of like converges the performance that users get. So at the end of the day the performance gap is like the models that are available to people and through this conversation we've talked about a few things like some is like this knowledge work domain which is all US workers and things like this which I think realistically the raw utility of the US frontier models are just like a bit more robust in a way that's been hard to measure. So it's hard to know on the benchmark numbers. I think the six to nine months on like the benchmarks that are popular will become stable but there's this like intangible unmeasurable where the US models will pull ahead. The best example is like this cloud code and codeex revolution that started in December and January where it's like it'll be very obvious if an openweight model enables this type of behavior because everybody in the world will be talking about it which is like I no longer I can pay $1 a month instead of open AI is $200 a month. a lot of people are going to switch and as of recording that's say like five to six months and I I just like don't really expect that like really clear watershed this year which will like shed some light on the gap and how like the way we're looking at the gap is like a bit narrow to how AI is used in a really broad set of use cases.

Speaker B: Now why haven't we seen a really good Chinese coding model as of yet? I think a lot of what's popular with cloud coding codecs is that they're just very easy to use and

Speaker A: I think the Chinese labs will keep solving coding. So like the actual code quality of the models will be incredibly high and that's based on like public data and things like GitHub and stuff like this and you create puzzles the model model learns to solve there. But it's in this like ease of use thing that the US labs have had a really big lead and there's a difference in how they operate where the US labs get consumer data. So like cursor trains their model on people using cursor and chatgbt and codeex and cloud code are the same while all these openweight labs they release a model and they have their CLIs that are a bit less popular but a lot of the usage of the model just doesn't go back to the developer. So some of it is a bit of like that real world data is really useful to kind of making the model more robust across general circumstances and that's where the closed labs have a huge advantage.

Speaker B: Cool. And I do have some open source model questions. Why are Chinese labs released in so many strong open source models?

Speaker A: It's a path to relevance. I think they know they're behind and they know they want to become more in the conversation. And if no one knows

Speaker A: If you have an API, nobody thinks about your model. But in a world where AI models are so valuable and pretty rare when you release them openly, there's a lot of people that are like, "Oh, I wish I had my own AI model." That are going to take a look and try it. So all the labs are really practical as like this is our way to be in the story of like what is happening at the frontier of AI and they will keep doing this and I expect this to be stable because I expect this like fast follow gap to continue. And this is why it's like a very practical mindset. It's like there's not much added risk to us doing it. It helps us accomplish our goals. It's the only way to get some customers. So like we've got to do it.

Speaker B: If Meta fails to become the best open model, does it mean that US actually is losing open model leadership to China?

Speaker A: In the last year, US has seated the open model leadership to China. I think especially in the summer of 2025 when a bunch of these new Chinese labs were starting to get really prominent. It's when um GLM 4.5 and Kimmy K2 came out as their breakthrough models. And like from that point on like the US no longer had open model leadership and that's been the case. I think since summer of 2025 anyone claiming otherwise is a little bit diluted on where the models are coming from that actually matter.

Speaker B: Does US still have the chance to catch up?

Speaker A: They do because they have Nvidia and Nvidia's financial interest is to not have AI be owned by any one company. And the best way to do this is to release open models for the world.

Speaker B: So the only hope is Nvidia?

Speaker A: realistically potentially like and open models like I think there's the likes of Meta and Microsoft where they have products whether user or enterprise that are supported by AI so they could release openweight models. It wouldn't really hurt their businesses but they've chose not to in recent years. Maybe it's a distraction. Maybe it's a like a marketing and security risk or something, but so long as that stays, I think Nvidia is best positioned to change this. As of recording, like a couple days ago or yesterday, and Jensen announced their Neatron 3 Ultra model. The way I think about this is like that's the completion of the first arc to Nvidia starting to try open models again where they had these Neatron 3 Nano Super and Ultra which is their like mixture of expert models and Ultra is the biggest one which they will then use to make synthetic data and all sorts of smaller models and distillation inhouse. So now that they have their first big model established, they have a lot more flexibility in building other sizes of models and continuing to improve it. And this is where it'll be more interesting to see their open model efforts. It's like once somebody gets to this big, can they keep kind of scaling it from there? Or Nvidia had a lot of models in the past that were like were good but kind of fell out of the um usage or coverage cycle a bit more quickly.

Speaker B: How about some smaller players like reflection AI or maybe you guys?

Speaker A: Yeah. So I think there will be many more people building models in the open and the open ecosystem can support like a diversity of models. So these smaller medium size models are very useful because they're so cheap and if you can get them to do one task repeatedly you're going to save so much money over using cler codecs and why like there's kind of a fixation in open models of chasing the frontier which is these big models such as an imatron ultra Kimmy K2 and whatever. So I think that's kind of something where over time that fixation on the largest model will fade. So like reflection wants to build a big model, but it's like people competing with Quen that will win the hearts and minds, which is this small model that every single researcher in the world is building off of. Like the extent that AI research right now is done on Quen is incredible. It was llama for a long time, but it's like everybody like all the researchers know how Quen works at that scale. They're used to using it and that's like a really remarkable achievement for them. And I think it's like not talked about enough. And I think over time that distribution from small to big will have more levels of interest. And I think like for a reflection it might be better to like pick a niche that's less served. And like AI2 builds smallo models which are great for research. And it's like I don't think AI2 should chase the frontier and try to build a one trillion parameter model. I think philanthropically it's not possible but also it's just like good to do something strategically different than like we there's time to fill different niches before people realize they can't financially keep up with the frontier.

Speaker B: Who has the best uh China open source models?

Speaker A: It goes back and forth. I think that I personally want to spend more time playing with the various models. I would say that the best models are like Kimmy and Zai. So like Kimmy K 2.6 and GLM 5.1 are probably the ones that I hear most as like actually making a difference at the frontier of usage. But it so fast like I think all of 2025 Deep Seek was the like unanimous king with the best models particularly like their V3 and R1 lines. And I wouldn't be surprised if Xiaomi was to to take a crown in China. Quen's biggest models have been closed. That has held back some of their adoption in mind share the biggest sizes. But I don't know if you put a gun to my head, I would say Kimmy and Kim Kimmy and Zai as of today are the best open models in the world.

Speaker B: But the competition is intense, right? That changes very fine margins.

Speaker A: very fine margins.

Speaker B: You mentioned that a lot of Chinese researchers in your article. Tell us about your impression on them. How are they different from the US AI labs?

Speaker A: There's so much talent in China whereas you go to a lab and it's mostly like Chinese researchers grew up in China. They're often younger. they lot like we went to a lot of places near Singwa and there's other great universities. So it's like you go to the labs for one it's way more uniform in who is there like a lot of Chinese researchers it's like here at AI2 or anywhere I'm like I feel like I work with all sorts of people and as somebody who is like not a immigrant or parents who are an immigrant in the US I'm like I feel a little bit unusual it's like I work with people from all over the world whether it's Europe Middle East other places in Asia and that's in the US labs that's also the case there's a lot of Chinese researchers in the US but that's like the population level is like It's definitely a a different feeling there. And then I think that some of these things we talked about earlier where a lot of the like top US researchers are like talking about where AI will be going. This kind of philosophical or artistic design of models and I think it's an example that the organizer of the trip um Katherine Renul pointed out is that like the likes of Amanda Ascal that are talking about like model character and really crafting the tone of the models. So, it's like that whole conversation didn't really come up in China. It's like there's some pockets there where the topics they talk about or like how far they're willing to project where AI is going is is pretty different. Otherwise, it's so similar. I think the things that are more interesting China specific is like the newest generation of researchers in China seem like even better at speaking English. Like so many of these really young people speak incredible English and they all follow the western AI ecosystem on Twitter and read blogs and they're all writing these papers where the opposite is not true where it's like the US ecosystem a lot of people just have no like they don't think about what's happening in China they don't like it's hard to get reliable information or what I feel like is of what's happening in the Chinese tech ecosystem even if somebody's like really interested in it and wants to know how these companies navigate. So there's these very different information flows that will feed into how people think, but also it's like a fairly American character trait to be like very proud of the stuff around you and like slightly boful. It's like a very

Speaker A: American thing and it's like oh the Silicon Valley kind of like rewards people for doing this and it's like a bit tunnel vision in a way where Silicon Valley is obviously developing its own narrow culture and discussion around AI with the like permanent underclass nonsense and now it's the recursive self-improvement and the singularity. There's sprinkles of this when you visit the Chinese labs but way less. They like got part of it from the US. are like, "Oh, that's a little weird, but we'll be entertained by it." But when you go to SF, it's like that's really what people talk about. It's like senior PhD students are like, "Oh, I need to make my money now. I need to join a lab." There's this very intense culture in SF around the timeliness of AI and how their career is changing that didn't feel the same in China where they're like, "We're going to do this." maybe a bit more down to earth in that regards like less anxious about things.

Speaker B: how do you view Chinese AI talent in the current wave? I mean like if you even look at the leading frontier AI labs in the US, you will see a lot of Chinese researchers, right? Chinese American researchers working there after completing their PhDs. Um they might be immigrants, they might be second generation immigrants. What kind of roles are Chinese talents actually playing in the AI research today?

Speaker A: They're in every company playing crucial roles. Like I think the US current political stances to limit high school immigration is bad for the country. Like I I want I work with a lot of Chinese or international people and I think that I want them to be able to live in the country where I like to live and I would like like I disagree with the policy that is going on. Yeah. And I think anyone that works in tech knows how crucial this talent is. And I think there's worries as for example there was recent news about like exit bans on passports for some of the top executives AI companies and people in the US that are sensitive to how important this talent is and the rising geopolitical tension will worry like what if they call back the top researchers cuz the US companies will be severely hurt. Like that's that's the reality of it. I don't know if that call gets made at all in the near term. That is not remotely near the circles of people who listen to this podcast. Like that's not who makes that decision, but like people talk about this stuff in the US as being a major potential bottleneck on the tech industry's progress.

Speaker B: It looks like you only need one person to lead the team, right? Then you have that model to be

Speaker A: vision is hard. So, an undersold thing about building language models is that so much of it is an org chart problem where you have talent and you want to make the model better. And the hard thing is organizing the talent and figuring out how to break down the big problems into little problems where a lot of talent can funnel into this one model.

Speaker B: I don't know if I'm ready to state this like really strongly, but it like could be the way that Chinese engineers are trained that they like weren't suited to making these org charts where it's just like they're just going to grind on their little thing and be happy that it contributes to the bigger model. I think there's a lot of uncertainty in this cuz like how the frontier labs actually work is not that well documented and it might just be more of chaos where it's just like people running around trying to make the model all the time is like not the best analogy but it's like it's definitely a works chart limited where you have to get all these people into this one thing.

Speaker B: Yeah. Are Chinese AI researchers more willing to drop old assumptions and just um absorb the latest paradigm? If you look at RL agents, coding models and tool use for example wasn't invented by deepseek but the company optimized the technology in an extremely effective way.

Speaker A: I think this is something that's more of just with new and like youthful motivated researchers. There's a very grim saying that's like science advances one funeral at a time which is like the most bleak way of saying it. But this is just that like new people are much more willing to accept the newest things and they have no reason to defend their old ideas cuz they have no established track record and they want to dive in and take these things and to the extent that China is potentially younger and both through engineering talent and leadership like those people just have less priors and therefore it might be true. It's something if you're trained in science you it's like the younger you are like that's where a lot of the new ideas and change comes from. for Chinese culture doesn't mean um it is harder to cultivate someone who is truly visionary for example that level of leading expert

Speaker B: that's the impression that I have and the impression is built on like like understanding that the culture is built on like standardized testing from such a young age and some of these tests rewarding memorization students know that like if they make this test they get to go to the best school at the next rung and they always are this like super competitive thinking about this kind of gate in their progress and correct me if my understanding of the Chinese system is wrong but that's my understanding where the US ecosystem has like less of these gates where it's like your fate is decided as better or worse and it's more of a like a continuum where a lot of the best people go to the best schools and the second best schools are also very good and they blend through each other and yes there's credentialism but maybe a bit less extreme so I think it's very possible that that is the case and some of leadership would off the record talk about this of like this rewards memorization and it's very hard to rework an education system despite if you want to

Speaker A: some tech leaders you talked to off the record said that that was the case and I was like okay I I have no reason to not believe you

Speaker B: you describe Chinese labs as almost perfectly set up to be fast followers can you explain that what makes them execute so fast

Speaker A: yeah this is why I was talking about the various things on the young talent and the culture of just like being very technically driven and focused on their task and potentially like they're just lock in on this thing and they have abundant talent where it's like you can just have people working across the whole stack and funnel it into a model and we've seen this many times it's like once you know that something is possible it's a bit easier to do this and we've seen Chinese tech economies come for other things it's like you go into like Xiai's technology you go into it feels like an Apple store it's very likely they're inspired by the iPad and the iPhone and their designs and they're good at it. Like they've done this not just in AI and I'm like I expect them to do it in AI again. And the the debate is whether or not like them doing this fundamentals to keep redoing it is enough for them to like people like to wonder is like oh is China going to leaprog the US AI labs? And I think that in an equal resource world they might be able to where they're just like they have more talent that's willing to work more hours over a 5 to 10 year time frame but I think human talent is only one of the resources that it takes to build these models and just like capital through mostly compute and data is actually like the biggest lever and it seems like they're disproportionately fewer resources than the US hypers scale companies. So it's like I think talent will only take them so far, but the technology diffuses where the best models make it easier to make more efficient like models. So it just kind of like is going to keep propelling through there.

Speaker B: Can a fast follower become the frontier leader?

Speaker A: They can. I think AI is weird because of the resource requirement. I think China's path to leadership is by it taking a long time to get to the final destination and that's time where Huawei builds out its capacity and then its net capacity becomes bigger. So the common scenario you'll hear in people talking about balance of power and chips. So I think that most people think that it would

Host: It could be a long haul for that to happen, but don't rule it out. As we talk about the leading AI scientists culture difference, you know, here in the US right, but I think back to China, maybe Xiaomi is an exception with Lulu Lee being quite active on social media as well as podcasts. There was also Ling from Quinn, he was the tech lead of Quinn, he left the company earlier this year. Especially for Linguin, I think there was apparent conflict between what's called the individualism and collectivism which prioritizes the company's interests per se. Do you see more of that kind of conflict coming out from the China labs going into this western ecosystem?

Guest: I think some of the researchers will find that they are very well suited to this. So I call him Justin, the American name. And it's like he was so good at this. And I kind of see it as something we've also seen in America, whereas like people in tech companies that can get a lot of influence externally as well can push against leadership and the big tech companies are like we don't need to listen to you. So then like they part ways, like we've seen it.

Guest: Like we've seen it in the US. I think a prominent example was around some of the AI ethics researchers at Google where it's like Timnit Gebru sent an ultimatum to Google leadership and they said no and it's like there's walls that you can hit within big tech companies that are hard to hit. And I think that like some of these Chinese researchers will probably get fed into this ecosystem and we might see some of it again. But for startups, the startups like Kimmy or Zi would love to have a person that is making noise and bringing attention, you just figure it all out later. Very much depends on the company size. But I think it would be good if more of the Chinese researchers get into this because it's good to have more views and like this whole part of the world if they can be seen in the US so more people understand how they think and how they approach models. It would be good for the kind of global equilibrium.

Host: Yeah, that's great. We're trying to do that. We're trying to invite more Chinese researchers coming to our show.

Guest: Yeah. So subscribe to our channel. And when you talk to the Chinese researchers right, what was the mode inside the Chinese AI labs? Did you see more confidence, anxiety, curiosity, or something more practical?

Host: I think there definitely is an anxiety of not being able to keep up in compute. So I think it's like all the Chinese labs want more Nvidia compute to train the latest models and the researchers are the people who feel this post. It's like they're the people that are hitting the resource constraints, but otherwise there's less social anxiety. I think in the US there's a big upswell of social anxiety, whether it's people I work with here that are seriously concerned about their job being replaced. And I'm like, you're one of the best AI researchers I know, but like even people like this will be like, I'm really really concerned about this. And this is in the broader discourse where American CEOs are going around talking about how people are going to lose their jobs which is not a good messaging decision but that's like very deeply embedded in how people think about AI now and the data center discussion. It doesn't seem like any of these cultural shelling points has landed in China. Some of the developers are like we ask them what is AGI and they're like when they take my job and they're not like it's not like a cloud hanging over them. They're like, I want to build this thing and figure it out later.

Host: Why is that?

Guest: There's some hesitancy around developers or like this is changing a lot. They all use Claude, which is supposedly banned in China. They all use it to do their work and they told us very forthcomingly like we love Claude, like Claude's the best model. So, there's definitely some hesitancy growing around like software jobs are changing, but it's much less of a collective like social opinion on AI is bad and disruptive. Like that doesn't feel like it's established in China at all. I think they some of them discussed how China's they've been living through so many technological changes in rapid succession as the company makes so much progress technologically and economically where they're like this is like the next one where they've had so many with these like the Chinese mega apps like they have electric cars now like they're just kind of like oh this is this is just how it is where in the US it's like these changes have come with more cadence in between them and then like this is the latest one and there's kind of like more inertia that it's going to have to move aside.

Host: Are people talking about alignment at all, safety issue at all within Chinese labs?

Guest: The word that's used more is security. The companies definitely care about this, but it was much less front of mind. It was like if we're releasing model weights, we want to make sure that it's not enabling any gigantic new risk. And we would talk, we would poke at like what about multimodal models? Like, yeah, some of these things have moved to be risks a bit faster, but it's much less of this big text LLM agent that we're releasing is a major risk and therefore we need to like not release it. They would say reasonable things about this, but it was not like in the US it can almost sound like ideological and like deep rooted concern that like AI like the next AI is going to be the one that is of a lot of concern. So, I think it's measured but not very front of mind.

Host: Yeah, probably like Chinese are just being more practical.

Guest: I felt like it was reasonable. I was like that's kind of my stance. It's like I support open models, but I'm not an absolutist. It's like I don't think every model should necessarily be open. Claude's release is a bit of a marketing stunt, but also it's like an incredible breakthrough model. It would not be good if that was just uploaded to Hugging Face the first day that they had it. Like it takes time for technology to diffuse and access diffusing over time for transformative things is reasonable and the equilibrium we're in right now where the closed models are ahead and open models follow it like seems like a good way to get technology into the world and I think that a lot of these companies would feel similarly.

Host: Which parts do you think Chinese labs are genuinely ahead of the US labs today and which parts are they not ahead but moving faster than Americans expect? So a lot of people say the Chinese labs are building more efficient models. And I think that's kind of overblown because the US labs are serving way more tokens at inference. So they're way more incentivized like the financial gain to make Claude or ChatGPT 1% more effective is like billions of dollars. So I think they have like more reasons to be efficient. But I think that the Chinese labs by doing it longer have a bit better understanding of like how people are using their open models. And we talked to people at Alibaba and they're very intentional about this. But also now the likes of Kimmy and Zai and all these others with these established that have multiple cycles. They're following how people are using their models and this expertise is very transient like I try to figure out like how are people using Qwen and it's like a massive like it's really hard to get the direct information but it is what makes their model successful or not. Obviously those benchmarks but then it's like why did cursor choose the Kimmy model? Kimmy's going to think about that. Yeah, really deeply or ask them and this is like very valuable expertise at the frontier of AI that's hard to get. So they like know what makes a bit better open model or not and you only get this from releasing models a lot. So I think like Nvidia is building momentum on releasing models and they'll start to get more of this feedback and then there's reflection in the US that's like we'll release a model when it's ready but they're going to get none of this feedback of people actually use the model and yes this benchmark is good but this very related thing doesn't work why and I think that's like the intangible knowledge of how the open ecosystem works is consolidating there right.

Host: Now let's move to the opposite case. What are the biggest disadvantage that the Chinese labs still is today?

Guest: Yeah, it's compute. There's likely incorrect estimates, but you could say something like Meta or OpenAI have more GPUs than all of these like Chinese companies combined, and they very well likely do. And I've heard rumors that say like an OpenAI researcher has about can get like 5,000 GPUs for themselves. And this is like would be a major proportion of a Chinese open model lab's research budget. So I think their aperture for risk and amount of dynamic things they can do algorithmically is much more narrow and I think there see like just optimizing the current path of the models is much more of a necessity because they know they can keep making this better but it's a less dynamic ecosystem. One of the Chinese labs told us like their latest pre-training run took like six months and this is a model that's out on hugging face and it's like if that run didn't work like that whole line of models just like wouldn't exist and that is a very precarious position to be for the company and I think that they're exceeding most people's expectations with the amount of compute that they have but that is the limiting factor.

Host: And yes the Chinese labs are desperate for more Nvidia GPU chips. How much does compute scarcity change their model strategy.

Guest: The biggest thing is that every model you essentially design it based on how you expect it to be used downstream and then what the shape of your training clusters is. So like the training cluster will configure how efficient it is to train a different size of model and then you also know you need to serve it. So based on probably having smaller clusters which is going to change how they do make these types of decisions or maybe they can change their base model architecture less frequently because pre-training runs take longer and I do think that we're going to see for some of these smaller labs I think Alibaba and bite dance with the resources they have is a bit different but like miniax moonshot etc like most expectations are clawed and GPTs are at this like three to six trillion parameter range and I think it'll be a lot harder for for them to scale up unless they're getting these like latest NVL 72 racks and Blackwell chips and things like this. So essentially there's like there's ceilings to these but it's a multi-optimization problem whether it's like what's your training cluster what's your inference stack they might want it to work on Huawei inference now.

Host: Like who are your users so they do all of these things and we see the output as a model on hugging face of like 600 billion to a trillion parameters it's hard to map but it it's one of the most interesting decisions in a lab and most people won't tell you.

Host: Yeah. So, does compute scarcity force better efficiency or does it ultimately cap the frontier?

Guest: It does, but it's more of like you're changing how you work like everybody wants compute efficiency because if you have a more efficient research, if you make your experimental unit half the price, you run twice as many experiments. There's like a lot of very basic math where every AI researcher wants this. It's hard to say that the Chinese labs would necessarily be better at it. It's like maybe because they have less inference compute, they train a model that's more efficient at inference and that has a different trade-off. I think Deep Seek's famous for these like memory saving innovations like some of which are used everywhere now, but it like might just be overkill when you compare like a US lab might look at that and be like h it would give us a small gain but it's not worth the effort or something.

Host: So how good are domestic accelerators such as Huawei chips in practice? Um I heard many people saying it's ready for inference but not for training. That's mostly the consensus. We would go to some more random companies, so not like core LLM, and they're like, "Yeah, we have to have Huawei. We don't use it." But most of the LM language model companies we talk to are established enough where it's like they all have services that people are signing up for. Zai serves trillions of tokens a day on their API. And all of those companies are like, "If you can give me Huawei chips, we will like maybe get more customers and they are figuring out how to use them for inference." But on training, it's only like very small models that could be trained on Huawei. And I personally expect that to be the status quo for a while. That's why researchers focused on training or like I would like more Nvidia trips.

Host: A while is how long?

Guest: I mean ask a semiconductor expert, but I would expect order of years whereas like still Nvidia is so far ahead because Nvidia is channeling the global supply chain into making the top end products versus necessity is the mother of invention at Huawei and it's like how far do they get? I don't know. I don't know.

Host: How do you see the recent debate between Jensen and Darkh? For people who are not very familiar with the context, Jensen argued against the strict limits on Chinese access to Nvidia GPUs, saying that China already has the energy infrastructure, chip production capabilities, and AI talent to develop advanced models. Independently, he said that restricting sales only cuts Nvidia off from a massive market while failing to stop China's technology process. while Dar Cesh was holding the China threat theory and of course Jensen is biased because he wants to sell more chips in China, right? But whom do you think is right?

Guest: I think realistically I'm in the middle and more on Dwaresh's side. It ends up being a geopolitical discussion which is like do you think AI is a technology that is going to be used for this? And it's like I would rather see what is traditional American values of like freedom of expression and all these things spread into the world. And I think that giving AI chips to China does pose some level of of risk in this capacity to like on the balance of power. As an American, that's how I've fallen on this.

Host: It's hard to know like to what extent. So when you listen to people within Nvidia, they essentially are saying there's no more power to build in the US. Therefore, no one in the US would buy these chips, so we sell them want to sell them to China. But at the same time, you have US companies asking for more allocation and US policy makers talking to constituents who are like, the US companies would have bought more. So, I kind of talked to both parties, which is these US labs that are like, I want more chips to sell my companies. And I talked to policy researchers at Nvidia where they're like, the US can't support more volume. So, it's so hard to get the ground truth information of like how much would the US buy and what is the diverted to this Chinese market. This is why like I end up with big air bars because I don't know who exact like what the equilibrium is and therefore there's like there's some chips that Nvidia definitely wouldn't sell in the US and we should tell them China and support Nvidia's AI leadership. So I don't know.

Host: The question I'm more curious is that will the ban uh encourage better AI chips in China which is Huawei chips right?

Guest: So it is definitely the case. We talked earlier about this like software as a service versus cloud and I think like AI is going in the cloud direction and this is going to support all sorts of new companies in China that need this cloud in order to build their viable products and there's going to be this growing inference demand. We talked about open cloud, we talked about cloud code like the inference demand is there. I think China is earlier on the adoption. You talk to developers in China and they're like I use claude like this is getting routed to what is western capacity and it's like as China unlocks more domestic AI inference capacity and these use cases like this demand will all funnel into how many chips can Huawei make.

Host: So this is coming in months not years like there will be new apps like cloud code that...

Guest: ...are so enabling in China that people have crazy demand for them. So I see this happening and in this capacity Jensen is somewhat right where it's like that is...

Speaker A: what's going to drive the local AI stack to be developed like the local labs have Huawei chips that they're designing inference for like in some ways it's happening and it's just a ratio of how much and to what extent is okay for people.

Speaker B: So the bottleneck is really on the production capacity for Huawei, right?

Speaker A: Yeah. and the ecosystem for how easy it is to actually run a model on these chips. So which is like the cool down.

Speaker B: Yeah.

Speaker A: So I think like I'm very bullish on AI demand continuing to increase. So therefore like demand creates this need.

Speaker B: Anthroic CEO Daryl Amade has consistently positioned himself as one of the most vocal tech leaders warning about the national security and geopolitical threats posted by Chinese AI development. How do you see his approach?

Speaker A: I think it's a little too aggressive. Like I think that some of anthropic media is pretty careful about separating the concerns from Chinese government and Chinese individual scientists, but it's a little like he's a little too aggressive in the near term about like risks posed by open weight models and so on. So I think it could be a little offputting to people to hear him kind of be the tip of the spear in that regards.

Speaker B: And you mentioned that that China's weakness in the AI data industry is more important than people realized. Can you elaborate more on that?

Speaker A: One of the most surprising things is that the data industry is much less developed in China. So we when we flew this whole environments craze so to say where companies like Enthropic and Open AI are saying that they have billions of dollars of data budget to buy environments from external vendors. We asked the companies and they're like no we don't really have this. We build some in-house. People would point to the likes of Bite Dance and Alibaba being bigger and being able to build internal data teams but there's very little talk of like you go and buy some data from an external vendor which is so normal in the US like the French lab playbook. If you take it a little like cynically in training the models is like the model is bad at something then you go to try to buy some data for it and you see if the data makes the eval go up and that's like kind of a cynical take on US Frontier Labs is like the boring stuff that they do. They do much more exciting things but that dynamic doesn't seem to exist in China. That looks like the biggest difference. It's like obviously open AI researchers have more compute than Chinese researchers and then therefore these Chinese startups the people work on different problems and maybe it looks like more efficient solutions and so on. But the data one was the only one was like okay this didn't necessarily need to be different and felt very different.

Speaker A: How people currently view US frontier models is that they're going into this like knowledge work tool. So like as a nominal knowledge worker like I use cloud code and codeex to now mediate the vast majority of my work and people see this happening for healthcare law consulting finance and the US models are all going to push into this capacity and a lot of the training data for this comes through these data company networks where they're giving money to professionals to create this training data. Yeah, there's a whole market of Scaleai, Merker, Surge, Turring. There's a whole plethora of companies here and the market is pushing them away from offshored preference data for RHF into this very high skill labor industry. And this is like huge, it's measured in billions of dollars of how much people are paying for this across each model lab. And they do this because it serves value and makes the models better at these agentic tasks and I think it's a question of like do the Chinese labs unlock this domestic market for their own agents and people do using their models for this because that demand is what will let them buy this data that's very expensive versus if the Chinese AI researchers and then knowledge workers are like I'm using claude like claude is just has all the more reason to keep doing this in the US and that's all for the US data industry so like it's I talked to my friends that do post training at labs and like these data companies are like really important to getting this like incremental performance bump on all of these domains. And if so, if it doesn't exist in China, like you're going to have to be very creative to make like an AI researcher is then going to be making up an environment in data of say like a finance problem of like making an Excel spreadsheet. And I just don't think this is a task that an AI researcher is naturally suited to do. It's like a different world than they live in.

Speaker B: I have a few more questions about Chinese demand and business model. You know like during the mobile internet era Chinese companies are so good at launching new apps applications. Um you know people would say companies like Muan, WeChat and Alibaba totally changed our daily lives and assumption is that China will move faster this time as well on AI applications once the models are ready. Do you see that coming?

Speaker A: I think both in the US and China there's going to be incredible dynamism in the applications. I think some arguments are even like the US consumers more used to adding new apps and then like going through the like getting very used to it and paying for it than in China where they have these mega apps and it's a question of like can the mega app support the same amount of diversity of new use cases where in the US you have things like Sora which is the number one app and millions of people try it it's a flop like and it's its own sandbox it can be a more defined experience so I would almost say in the software side the US is more set up for like this dynamic trial of new apps.

Speaker B: Were Chinese labs worry about monetization at all?

Speaker A: Not super vocally, but we also didn't ask. I also don't think that many of the like frontier AI labs are that obsessed with monetization. Open AI and anthropic want to IPO, but most people building a AI models are just like, I want to build the best model. And if you assume the model's going to get better, like it they're so useful, the funding won't be a problem. depends on whom you ask, right? If you ask a product manager, probably they're concerned about or the leadership that's trying to support buying more compute.

Speaker B: Should US labs visit China Labs more often?

Speaker A: I think we'll see some US especially small and medium players that are trying to build economic value around open models whether it's a fine-tuning API or specialized deployments or so on want to build relationships with Chinese labs because the Chinese labs are so open to sharing their technology and expertise after traveling to China we've seen like I've gotten inbound from a lot of companies are like how do you get access to all these labs we would go visit and this is a you could look up a known own either impra or training company in the US that works on top of open models where if this evolves it'll be kind of a new era of these new companies learning to work with each other across borders. I would expect more US companies to go to China because it's much easier in the visa situation right now. So I traveled on the visa free transit where you enter and exit through different countries and that's super easy to do. pretty much just show up. Whereas like a Chinese research or engineer traveling to the US, you have to wait for the visa games. So I expect more US medium-sized startups that are interested in open models to go to China, which is probably going to happen in the coming weeks to months. Like this stuff happens fast. The summer of 2026.

Speaker B: That's good to know. You've been a frequent guest on the podcast show with Lex Friedman. He was in China recently. I think one of his videos catching a truck ride really going viral on Chinese internet. Do you and Lex actually talk privately about your China trip or his interest in visiting China?

Speaker A: We haven't done a postmortem of like what do you think of China? I think both of us are just interested in seeing this large swath of humanity that influences tech that we live with so much and seeing it firsthand because that's the best

Nathan: I think I'm pretty neutral. I think most of my worries on AI are like very centered on like US opinion being sour on AI and how that could hurt progress. I think the global AI race is roughly what I would expected geopolitically in this biggest picture. It's like the US is ahead and the Chinese wants to catch up and there's this nuance in terms of how much chips are being sold and what that like what that pin ends up doing to the ecosystem. I think it's mostly just a benefit to be able to bring more humanity to this and like do what I can to make it less of a very extreme circumstance for people on both sides. So people building and using the technology that have very little say on what the highest level decisions are. It's like I could do a small part to make that more understood for both sides.

Host: All right. Thanks so much Nathan. I think all those are my questions. Anything else you want to add? Any question I didn't ask?

Nathan: No. Thanks for having me. I'll hopefully sometime soon go for a second trip to China. There's plenty more to see.

Host: Yeah, we'll love to have the second interview after your next trip.

Nathan: Thank you.

Host: All right. Thank you so much. Yeah, tell us about what you guys do.

Nathan: So, AI2 does all sorts of kind of open-source AI research and development. I think it's historically been fairly academic and over time it's been moving more into this artifact defined development where you're building models and code bases and data sets that the rest of the community can build on. So I think Mo and these open language models has been the most popular one but there's also like Mommo the multimodal Mo Earth and other things and then AI2 has historically been very well known for its integration with AI literature. So I think the application there has been Semantic Scholar which is kind of seen as alternative to Google Scholar for searching academic literature and this team has evolved into building AI agents for science. So it's this ATA brand where they're building tools to assist people in building and understanding the scientific literature. So I think at its core it's cult like deepest culture is so scientific and like has this community value of building a lot of just like net good for the scientific community which is really great and that comes through in these papers that are very detail-oriented and caring and I think there's just like so many different ways which way do you want to go there's so many different ways that this comes through and it's the best one of the best things about working at AI2 is you just like you get so focused on trying to build things that various researchers use when you travel as an academic and you're like all these people are building on Elbow and know of AI2 and it's just kind of like a lasting thing that is great when there's interns or student researchers and they go out into that academic community and all these other research institutions to kind of spread this scientific value which is a great service that AI2 does.

Host: Yeah. How many people do you have right now?

Nathan: Probably about 200 people.

Host: Oh wow. That's a lot.

Nathan: So most of them are researchers right? I would guess probably half of the talent is technical engineering on like the traditional AI research and then there's engineering teams that help support our clusters and um some of our products.

Host: How is like a NGO lab different from other labs in Silicon Valley?

Nathan: So AI2 is a full nonprofit which means most of its money comes from philanthropies. So the Allen is on the door which is from Paul Allen one of the co-founders of Microsoft and then also has large grants from u for example from the NSF. So, it's all nonprofit funding and doesn't really sell any products. It just gives everything it makes away for free to for people to learn from.

Host: Wow. So, you guys have more research freedom, I guess.

Nathan: Yeah. So, I think that on the spectrum of closed lab to academia in terms of resources, it sits in the middle in terms of freedom, which is like we have more resources and we feel pressure to build models that are actually good. So it like does suppress the freedom a little bit but much less than a frontier lab that is like racing towards AGI but if you go to a true academic institution you'll see that they have a diversity and kind of intellectual spark that you like won't get at a company just based on the needs of paying the bills.

Host: How is Seattle vibe different from the Silicon Valley vibe?

Nathan: It's so much calmer. Like there's not events all the time with various people meeting up. I personally wish that more people in Seattle. like there's major tech talent hubs at Amazon, Microsoft, and all the tech companies would have more of this, but it's what it is. You can't fight it. In some ways, I think it's a better place to like settle down and have a family because it's like a bit more spread out and it's a bit less intense and how bright this everything is in terms of talking about AI. Like I don't have AI in every conversation where in the Bay Area AI feels like it's the only conversation right now, which I understand why. Like I'm bought in as well. But you visited the Bay Area a lot.

Host: I go regularly. I say I go like three to eight times a year.

Nathan: Okay.

Host: I went to UC Berkeley, so I lived in the Bay Area for like seven or eight years.

Nathan: And then you have to go back to stay in touch with what's actually happening in the Bay Area. You don't have the view like this. A very beautiful lake. What's the name of it?

Host: Um Lake Union.

Nathan: Definitely my dreaming office for sure.

Host: Yeah.

Nathan: Yeah. So when you visit China, how do you think the office there, the

Speaker A: AI lab office there, they don't have this view, but they're also tech company offices where it's like they're pretty similar in that regards or like I go in them and it feels like I'm in a tech company and I've been into a lot of tech offices in my life and it's like, okay, the showroom culture is different.

Speaker B: Oh yeah, showroom pretty unique in China.

Speaker A: That is a very unique China thing that a lot of people just I think have to visit for their first time and then they're like, okay, this is a thing. Cool. Cool. Thank you so much, Nathan, for having us.

Speaker B: Yeah. Thanks for coming.

Speaker A: Bye.