Taylor Ogan: But, but that's frustrating, because you saw people from open AI who read the paper who were saying they must have, they must have copied us, basically. This is not what it appears to be. They knew the answer. I mean, I, I hope. And so they were trying to really trick people into buying into this narrative that nothing that comes from China is how it really is.
And, and so yeah, I think that that's something that's really frustrating. I mean, one of, one of my, the core framework for understanding China is if the numbers or the tech appears too good to be true, they probably aren't false.
Welcome to Manifold. Today, my guest is Taylor Ogun, the CEO of Snowball Capital. This is his second time on the show. We'll put a link to the earlier episode in the show notes. We are excited to have him here all the way from Shenzhen, China. Taylor, welcome.
Taylor Ogan: Thanks for having me. I'm excited about this one.
Steve Hsu: Yeah, me too.
So we were just saying, some things have changed since weI was in China in the fall since we recorded the first episode, but I didn't get down to Shenzhen. I hope to fix that maybe later this spring. Tell me what the vibe is like where you are.
Taylor Ogan: I mean, Shenzhen is such a different place than the rest of China and, and maybe to a fault in some ways. When you come to Shenzhen, maybe, yeah, you weren't seeing the real China, but you're also not seeing the future the same way. You know, you can, you can go to the suburbs of Beijing. And ride some robo taxis around.
You can see like one random drone delivery in a suburb of Shanghai, but that's really normal in Shenzhen. And even like the EV landscape, you just see all of the cars, some of the EVs that shouldn't still exist. They're on the roads, they're driving around. When a car is launched, you're guaranteed to see it that week in Shenzhen.
so it's, it's just such a different place. And I think actually in a lot of people's understanding of China kind of post COVID, they're not coming to Shenzhen and like state visits,Gavin Newsom did come to Shenzhen. That's the only U. S. politician that I know of who has come to Shenzhen since, and he just did the most random route.
So I think that, I think politicians, other business leaders, we track private jets, landing in major Chinese cities. And some of them are coming to Shenzhen. So that's really promising. Jensen has been here, on some kind of secretive trip. These guys like to use the same exact VistaJet. So it's pretty easy to check and they're pretty lazy in terms of the origin airport.
So, Tim Cook comes here quite often, but yeah, I think in terms of others coming to China, Shenzhen still isn't on their list. And, I think that should change if they want to get a real feel for how the future is developing.
Steve Hsu: That's totally crazy. That's like somebody not knowing where Silicon Valley is in the United States.
Taylor Ogan: is. Yeah, yeah. That's how it feels.
Steve Hsu: For my listeners, Shenzhen is just across the border from Hong Kong. And it's in a way like the newest. Well, there are a lot of, in a sense, new cities now in China. This one really didn't exist at all 40 years ago, and now it's in a way, as we were just saying, the Silicon Valley of China.
It's the place where people who are really future oriented in China go. I would say Shenzhen and the other places are probably Shanghai and Beijing. But Shenzhen is really like a tech city. It's the center of a huge supply chain for electronics, advanced manufacturing, all those things.
I, I wanted to ask you, one of the impressions that I've been getting just following things from afar, but also having a lot of conversations in November when I was there is a sense of inevitability or irresistibility of this domination of the manufacturing supply chain.
Like, it just doesn't seem like anybody can slow it down or stop it. And, you Shenzhen any, any thoughts on that?
Taylor Ogan: Yeah, it is. It's a weird feeling. It's a really exciting feeling. I think the turning point was late August, early September, 2023. That's, you know, when Huawei came out with the Mate 60. And that I like to talk about things in terms of future technology museums, like what gets an exhibit and a permanent exhibit.
And I think that is one of them, at least in the China domination in broader tech. That is the first stop. and, and that really just ignited something in people. And, and it's very evident, something that's really unique about Shenzhen is you really do just hear people talking on the street about DeepSeek. About Huawei. Like. All ages, all genders, like everyone is talking about this. Everyone is aware of this. I mean, I just moved into a new apartment. My landlord brought her husband to work at Huawei. She is the kind everyone does in the city. It seems, she brings me this, this. Robo vacuum for the windows because of a brand new apartment.
So the windows are still kind of dirty. and I was like, I, I feel like I'm a tech guy. I don't even have this. And she was like showing me how to use it.
And so, yeah, I think that's definitely unique in Shenzhen. this This this inevitability of, of domination, I think with that, and we talked about this last time, the Chinese engineers need to be more confident for them to really fully understand and put their efforts towards something like this. And they are that since we last spoke, November, 2023, I just listened to it last night. and I was like, man, a lot has changed, but that is something that, that we were talking about that exact thing.
And it really, it, I thought maybe it would take five years to get there, but it has happened. I think DeepSeek, especially right before Chinese New Year, when everyone goes home and, you know, shows what they did to their family that year. The timing of that was largely due to that, which was kind of awesome.
when.
When everyone shifted over to Xiaohongshu with the TikTok ban, the founder of Xiaohongshu in his, his, like, basically status in WeChat was, this happened, like, basically on a Friday. Don't talk to me all weekend. I'll be back in on Monday. Like, so it's kind of like this, you've never felt this.
The Chinese have never felt this in terms of technology where they know that when they release this paper and this model, they close their computers. And they go home and they know that tomorrow, you know, is going to be crazy. So that kind of confidence is how I would really summarize it. But it's a clear shift.
Steve Hsu: This is great. This is one thing I wanted to focus on because. Out here in America and in the West, we talk about a DeepSeek moment. We talk about it mostly in the way it affected NVIDIA's valuation. But I wanted to get the vibes from you for what the DeepSeek moment meant for engineers and investors in China.
So that's something I'd love to go deep on today just to unpack some stuff for my listeners who are not as, maybe not as up on what's going on in China as you and I are. You mentioned the Mate 60 Huawei phone launch. And the reason that was big for the Chinese was that the U. S. government really literally tried to kill Huawei, right? They tried to deprive them of the chips they need for their 5G base stations and infrastructure. They deprived them, tried to deprive them of the chips they needed for their cell phones. And their phones were really inching up in quality compared to Apple. You know, they were pretty competitive with Samsung and Apple and then they got kneecapped because suddenly they couldn't buy chips from Qualcomm anymore.
And so the Mate 60 was the first model that came back where it was a domestically made CPU for and and modem unit everything for that system on chip for that phone, and it was a world class phone. So, that seems like ages ago, but that was a big moment right in China. And now you're talking about a situation where something that DeepSeek does, for example, is noticed everywhere in the world, right?
Everybody in the world, people in Paris are talking about DeepSeek, right? Talk a little bit about, were, were you surprised? So had you been tracking DeepSeek, before the V3 R1? I mean, V3 came first and then R1, but how did it hit you and how did it hit people around you?
Taylor Ogan: So it's kind of funny how certain technologies, rumor spread in different cities. DeepSeek is a team from Hongzhou, not Shenzhen, it's quite rare. And it's kind of like when this Wukong video game came out, that was kind of another moment where a lot of people were really, really excited. Like, we have the best game out there, and it's Chinese.
And very heavily Chinese characteristics. But it was, it was made by us. So like these rumors do spread, the DeepSeek thing I happen to be, it's one of those things that after it all explodes, you go back and you search through your messages to see who, who is really trying to turn me on to this. And the earliest one I have is April, 2023.
And it was someone saying it wasn't DeepSeek at the time.But they said it was a screenshot from one of the founders, saying benchmarking Open AI. And, and I mean, you know, that was many models ago, but that was, so that was the first time that it was really on my radar. In December, things of 2024 really started ramping up and I mean people from very different circles where we're talking about it. And all that we know is right before the new year, when everyone goes back to their hometowns, something big is going to happen. And that, that's kind of all we knew. I mean, we assumed one of the big questions was, was this going to be available only in Chinese or, or really good at just Chinese and then not so good in other languages. and I, I think that actually is, is pretty important too. It's an explosion, because had we talked about this last time, you kind of Chinese still need the West to validate things. And that's unfortunate, but that's the reality.
I think that will change but in terms of culture and fashion and music and technology still like that's why Tesla is still pretty popular, you know in China. So yeah, that was kind of the question was how like worldwide will this get and but also one a huge part of the rumor was that its fully open source and um Reinforcement learning was, was definitely stressed in the rumors.
Beyond that, I mean, we just were kind of waiting and then when it came out, it wasn't. I don't know what people expected, but it didn't explode. Like within, I mean, now we're measuring in terms of hours. It didn't explode within like the first 12 hours, then 24 hours. And then it was like, but you can, you can use this.
They got rid of the, the login page even, so you could really just type in, you know, DeepSeek and then have it, and use it immediately. And so I think it kind of, honestly, I think we had to wait until the West kind of woke up and really fully processed it. And then it was just people, you know, mostly academics who were saying this thing is amazing.
And then that next day, like day two, fully day two was when it just went crazy.
But at that time, the DeepSeek employees were, I mean, I have a picture from a friend of their office and it's just, the lights are off and the door is locked. There's no one there. So, so they, they knew what they were doing and that's kind of part of the excitement.
Yeah, I guess. If there's a movie ever made about China's tech rise, that will be a huge scene.
Steve Hsu: Right. So, you know, for the really geeky people who actually read the DeepSeek papers, we kind of knew they were for real all the way going back to, you know, maybe two or 2. 5. And then when V3 came out, which was, I can't remember exactly when it was, it must've been over the summer, well, before our one came out already, if you read the V three paper, they hinted at some of the things they did to train V three.
They basically admitted they had an advanced reasoning model that they used to do some of the training of V three. But the V three paper, of course, they didn't release our one. Our one came out months later, but some of us already knew it existed. And so when I was planning my trip to China, I was mainly in Kunming for personal reasons, and then Beijing and Shanghai, you know, to really see, take the temperature of the country.
I had been in contact with one of the DeepSeek engineers, a PhD. And, he was trying to get me, you know, a pass so I could come visit Hangzhou. I was going to stop in Hangzhou after I gave my university seminars in Shanghai. I was going to go out. It's only like a half hour, 1 hour train ride. I was going to go to Hangzhou and hang out with them.
And then somebody there put the kibosh on it because they're like, no, we don't want any outside visitors right now. So, I would have been a great Twitter scoop if I had gone there. In the run up before they released R1. So,
Taylor Ogan: So look, since I did your podcast last, I don't think I've done any, I've turned down all podcasts because it's just, it's such a different world here that I don't think people really fully understand it. You need a really dedicated podcast.And I think your listeners probably have the highest IQ of any podcast.
And so I don't really care. I'll say some things that like, it's, it's not, I'm not uncovering some crazy huge thing, but something that I probably wouldn't maybe tweet. and one, the first thing is that I, I just heard that the mayor of Hong Joe has to personally approve all meetings with DeepSeek.
Now, I don't know if that's post everything, but that could be why, and I, I also don't know if they're saying that kind of like this hyperbole, but, yeah, that it's kind of funny when, when these technologies are ahead of the government and we're seeing that more and more. In China, you used to never see that. It was always the government that knew exactly what was coming.
Here's another thing when Huawei released the Mate 60, right? There was no press release. They just quietly released it. and a friend who's very in the know, text message me on WeChat, I'm in your lobby. And I was like, why, why are you in the lobby? Like I have things to do. And, he had the phone and it had only been out for like two hours.
And, and so, I mean, so that's how kind of some of these things start. But I have been asking so many people about that moment, just because we're, we're going to replicate that in, in different ways throughout the years now. And I wanted to know, did Huawei need approval from the government to do that?
Because Gina Raimondo was in China and her whole thing was her dad, you know, his job was displaced by a Chinese factory. And, and so it was very personal to her. And the Chinese knew that. and the timing, we can admit it was not accidental. It's just, I wanted to know, did Huawei have this phone ready?
And then the government said, well, actually Gina Raimondo is coming for a visit. How about, you know, I just needed to know that. And I have determined through various sources, who would know the answer to this, that no, they did not, they did not consult with the government. Now, the Shenzhen government, maybe. But they did not.
And, and so that was something that really shaped my worldview. Because I just assumed, like, I think I know what's going on in China enough, but I do assume that the government has, you know, this overreach in, in some areas. And that was one that, again, just like DeepSeek, they knew that this is people were going to freak out about this, this phone, most obviously the chip.
But same thing with SMIC. Like, did they, did they consult on this? Did they have a say in, in yes, this would be released like when Gina's here? And, and so, yeah, the answer is they just did that.
And so I, I think that's also Shenzhen of them to do. You have companies like BGI who, it's kind of funny when, when people in the West talk about BGI, those who know it, they, they talk about it like it's a, it's a puppet of the CCP, when in reality they were the first company to leave the, you know, Academy of science and come to Shenzhen. That was basically, they left Beijing. They came and you definitely know the story better than I, but they came to Shenzhen, basically in protest of the government after they signed them up for, you know, everything.
So, Shenzhen companies are like the bad boys. And they just do things as they want, whereas in other cities, companies. it is more tied to the government. So back to when these companies are ahead of the government, it's very interesting to see how the government tries to take credit for it.
And so sometimes the government does kind of incubate different industries. Like in Shenzhen, they're really obsessed right now with the low altitude economy. So basically drones in every form flying around. And that is like, they are opening up literally airways for drones. And then, so they're, they're very hands on with that. But with large language models, I mean, look at Kimmy, like the government definitely has some, some influence in this. But with DeepSeek, I bet the government was not consulted ahead of that launch either. And, and it's kind of interesting. It's not like it's not censored in the same way. Like you can ask some very sensitive questions. And I think that was kind of a risk that they took. and you could track this back to maybe Jack Ma, as maybe the start of testing the waters.
but, and I think the crackdown on, like online learning platforms overnight, kind of reset things for people, but now we're kind of seeing people experiment again, these companies. And, and so for the Hangzhou government, they, after the post-deep sea, wanted to really be proud of it and say, yes, this was our creation.
So yeah, I think it's just interesting to see how that works.
Steve Hsu: Yeah, you know, on, on those different things. So for Huawei, I would have, I could have gone either way as to whether they sort of coordinated with the central government for that release. I was kind of like you. It could have, it could have been either way. For DeepSeek, I would say 200 guys. You know, a quant trading fund.
Very likely they did it. They just did it. They didn't ask for permission. They just, hey, we're, we're releasing a technical paper. It's a great model. We're releasing the model. Hey, we're just going to do it. That would have been my bet in the case of BGI. You're absolutely right. They were in the Chinese Academy of Sciences in Beijing, and they were recruited to Shenzhen.
The Shenzhen government offered them. factory land, all kinds of stuff. And they wanted to bring genomics technology and research to Shenzhen. So I think one of the things people here in the U. S. get really, really wrong, and we can use the case of DeepSeek as an example. So you and I are pretty sure they didn't coordinate with the government.
They just did what they did, and it was fine. And now the Hangzhou government wants to take, obviously, credit for having such a great company in their town. But, if you read on Twitter, not just Twitter, but even, like, leading people who follow AI in the U. S., or even follow the U. S. China AI race, they all assumed that DeepSeek was like hand in glove with the Chinese government and like, oh, they didn't develop this for just $6 million.
The Chinese government probably smuggled tons of H100s into the country for them. They probably have like billions of dollars. They probably have a data center like Elon's or something, right? So there were just serious people, not just the rando. on X, but serious people had completely the wrong picture of what I think is probably the actual relationship between the company, this tiny company, DeepSeek and the government.
And so people out here just, I think, have it, you know, it's just very tough to understand how China works. If you don't spend any time there.
Taylor Ogan: But, but that's frustrating, because you saw people from open AI who read the paper who were saying they must have, they must have copied us, basically. This is not what it appears to be. They knew the answer. I mean, I, I hope. And so they were trying to really trick people into buying into this narrative that nothing that comes from China is how it really is.
And, and so yeah, I think that that's something that's really frustrating. I mean, one of, one of my, the core framework for understanding China is if the numbers or the tech appears too good to be true, they probably aren't false. And I'm not saying that everything is true, but if there are so many checks that no one in the West can comprehend, If you have, let's say an EV that you're selling a hundred thousand units a quarter and you consistently are doing that in the west. They say well, there's no way that you're doing that profitably or you're faking the numbers.
Once you start talking about the government, you should stop yourself and realize okay I'm clearly talking about something that has been successful. And then maybe the government was responsible or had a huge part in this Maybe they had no part in it or maybe it's somewhere in between oftentimes.
It's somewhere in between but the point is that once you get into that conversation, it's kind of game over. Like you're talking about a product that's awesome. So yeah, I think that is kind of the default, of people in the West when talking about Chinese technologies. And you need to, if you have that, stop it.
I mean, that, that really is, is what I encourage people to think because then you can, you can take the time to actually find more opportunities instead of trying to disprove one specific thing. There's no way that DeepSeek did it for that much. There's no way that, you know, if you were too focused on why LIDAR shouldn't be used in cars, you missed the whole LIDAR wave.
I mean, now 40 percent of new cars, luxury cars sold in China have LIDAR. I just re-listened to it last time on here, it was 10%. So that's a monumental wave.One company yesterday just sold their 1 millionth LIDAR unit. Now some cars, it's quite common for cars to have 3 LIDARs on them. So like, while you're trying to disprove that, you missed it all.
people did that with EVs, with, you know, they got too obsessed with trying to prove fraud and In, you know, Nikola and companies like that, and then they completely missed the whole EV trend. So, yeah, I think that's a first step in trying to really follow the Chinese tech landscape.
Steve Hsu: I think, you know, this is obviously not something you need to know, but for our listeners, I would say the simplest test is very easy. You go on YouTube, like, say, say you heard about some new product in China. It could be like humanoid robots, right? You think, oh, this could be BS, like, they, surely they can't make humanoid robots, only Tesla can make humanoid robots or something, right?
So, just go on YouTube and try to find a credible Western expert, don't trust any of these, you know, shifty Chinese guys. Just look for a credible Western guy and you'll be able to find a Western guy who's actually gone there and tested it or has something to say about it. And, it will take you like 10 minutes of searching to find it.
And, and, you know, it's just nobody even investing that amount of time before they have some conspiracy theory about how the tech is fake or the, it's all part of a, you know, government project and they don't make any money off of it. So, yeah, I think that most of the anti China people, even the China skeptic people are just lazy.
They're not very rigorous.I think they just went through for me personally. We just went through another one of these. It's like, for every vertical you can name where there's one happening.
And so humanoid robots, you know, kept seeing these unitary robots, you know, robot dogs, humanoids and, and, you know, I'm sort of thinking these things are really good.
And I remember people saying online, like, oh, this must be CGI. The robot can't really. Right. It's got to be CGI. So I just did what I did, I just followed the algorithm. I just gave you, I just started doing some Google searches and there was like a robot competition in Japan where any company can send robots.
And I'm like, okay, Japan, we can trust Japan, right? That's above board, right? They're not in the pocket of the CCP or something, right? So, so, you know, you go and watch competitions like this in Japan where different robots compete in obstacle courses and stuff. And the unitary dog won by a mile, right?
So you can just, you can just verify these things very quickly. Now, I tweeted out a, I think it was a Morgan Stanley report on, you know, what the global supply chain looks like for humanoid robots. And they were having trouble identifying Western companies that really mattered in that supply chain other than Tesla and NVIDIA, partially because the supply chain is very similar to the EV supply chain.
It's electric motors. control systems and yes, it's centered, you know, right where you are basically. So, you know, that's another one where you could easily be fooled into thinking like Boston Dynamics and Tesla are leading the humanoid robot revolution. But no, actually, like everything else, it's centered in China.
Taylor Ogan: Yeah, and I mean, that was Adam Jonas. So it's, if, if Adam Jonas is picking up on it, then it's probably pretty obvious. He's kind of a known Tesla fanatic. and anything that's not Tesla doesn't exist. And, but, but yeah, I mean, that, that definitely is what's happening. I'll, I'll drop something else.
There will be, I think, a DeepSeek moment in robotics, meaning a company that we don't know their name right now, will come out of nowhere. And I think that probably won't be as big, as, as widely, understood as something, you know, but that's a really big deal. And that's what we're working with here.
In the U.S. you can't, you would need to have just a constant plane back and forth, bringing parts for you to have a real, real robotics startup.
And so now, I mean, there was just a Shenzhen company that yesterday released a video of its robot doing a front flip, a standing front flip, and not standing on a box and doing a twist front flip like Boston Dynamics. Not that that wasn't impressive, but, and that has the most impressive gate. I've never seen a humanoid.
Now if maybe we should talk about humanoids right now, because it is something that literally even last time we talked about, it wasn't a huge industry. Now it is. So we've met with all the large, humanoid robotics companies. And it's kind of funny. A question that we ask every single time is why does it need to be bipedal? And this is where I, I'm curious to your take on this. They refer to Tesla. And they say, well, Optimus looks like, you know, this. And so in their own slideshows. And in their decks, they're showing optimists and I'm like, you guys know that that thing is tethered and controlled by someone right behind it.
And like, why are you obsessed with this? And they said, because investors are and, and so it kind of comes back to this tiny group of people with a lot of money and frankly, aren't that intelligent, especially when it comes to the space. Who is saying it needs to be like Tesla. And what's more is once we get past the Tesla part, and by the way, they all agree that they're years ahead of Tesla. So why are you putting it in your deck? The other part is that they say well it has to, the people who are buying this, they have to understand that it's easier for them to process if the thing looks like a human that this is designed to replace a human.
However, other people that I've spoken to and not at these companies, they say, I've asked the same thing. I don't get why it can't have rolling, you know, legs or, or other means or the trunk even, you know, that's where the battery is. And so that's the most important part. And, and that's a pretty high center of gravity.
And, and so it would make more sense to have a bigger base and especially they're going around mostly factory floors that are already flat. And, and so really just different arms. And we are now seeing that, but still there's a huge obsession with humanoids. So I would imagine that humanoids, as we think of them, bipedal will actually not be as big of a thing compared to robots with extensions.
And you're already seeing some companies just not putting the head on, because pretty useless, except just optically. So yeah, I think it's, it's a very interesting space. And it's changing really quickly. And something that's also interesting are the people who are working in this space.
The smartest people in the world, from my point of view, are working, who are born like, you know, in the 90s, are working for either an AI company at a very high level or a robotics company and it's funny because they can jump between them. And so some of these people at these leading Chinese robotics companies, I talk to them, they say growing up I wanted to work at Huawei or I wanted to work at Alibaba. And now they're working on, you know, basically an extension of their high school, you know, robotics competition team. I don't think a lot of people expected to be in that space, but that is where I think it used to be the autonomous driving space. We're kind of seeing that fade away just because it's, it's more robust and kind of more or less figured out. Among the companies who have figured it out.
So yeah, we're seeing this the shift. of really smart people and in five years, I don't know if they'll still be working in robotics, but right now they have some of the smartest people in the world
Steve Hsu: I would agree with that. Now, I think something that's maybe not obvious to an outside observer is that if you actually look under the hood at the architecture of what a language model is doing, the multimodal language models, instead of processing like a chain of tokens, which are like text from a sentence, that chain of tokens could be little chunks of the visual field.
And it's the same architecture, the same transformer architecture that's really good at looking at a visual field and saying, Oh, there's a TV over there, and there's a cat over there, and that's a dog. And, ultimately, that is what you want to integrate with your robot. Because your robot is, instead of like, having a crude way of sensing, what's in the world.
You want it to actually ascribe meaning and map what it sees into that same embedding space, which has all the concepts that we use like cat tree, you know, fire and that. So, people who are interested in AI, like large language models, are interested in robotics because robotics is the next evolution of how transformers are going to be used.
And in particular, robots are really good because they give you a really good RL environment. So, the big breakthrough that DeepSeek made was they realized, oh, if we just give the model well posed problems, like where there's definitely a right and wrong answer, but it has to think through multiple steps to get there.
And we just reinforce, we modify the connections in such a way that gets really good at these well posed problems. The real world gives you well posed problems. Like if I would, if someone says, hey, pick up the cup and lift it above your head to the robot, it either does that or it doesn't. Like maybe it smashes the cup or it doesn't get it, or it picks the wrong thing up, you know?
So the real world is a great RL space. And a lot of these AI people realize that. So they realize it. Once the robot is good enough on the basic mechanical stuff, they have an infinite RL system. And we now know that these models can get really good really fast once you put them in a good RL environment.
I'm actually going out to Silicon Valley in a few weeks to hang out with one of the biggest video game companies. And they have internal physics engines for the video games where stuff happens in the game world, but it follows the laws of physics. And we're looking at like, that is our, an RL environment for training, you know, ultimately training robots, but in that virtual world.
But all of this is actually connected under the hood. And so you're right, people can go back and forth between these different companies. Now, the guy who's really like a mechanical engineer trying to figure out like, oh, how do I make the fingers of this robot really sensitive. That's a little different than building AI models. But at the control level for these things, there's tons of commonality.
Taylor Ogan: And something I keep hearing from these people is that these robots are not yet smart. They're not yet intelligent. And so I think that combined like if I were a Silicon Valley engineer listening to this, I would be terrified by the fact that one, they're not yet smart. And two, that China is now in the position, these Chinese companies are in the position, where they're just releasing stuff without any governance. Because that's something that was not the case a few years ago. That would terrify me.
And that for me excites the hell out of me. But that is something that is like seeing humanoids walking along the street in Shenzhen, like when enough people send me videos of that, that that's, you know, but but.
But a few years ago, I think there was this understanding that we need to have this huge press conference, we need to invite the mayor of the city, we need to have the press like, you know, frame it in a certain way, and some companies are still doing that. But other companies, these companies with, I think a younger team, and I think a lot of the people were educated in the West, so maybe that's where they get this from except DeepSeek actually, but no one was educated in the West, which was interesting and they still do this. So maybe that's not the cause.
Steve Hsu: Could just be generational.
Taylor Ogan: I think it is totally generational and I think that is one of the biggest. And that's why a company like Huawei is so complicated because you have a really structured company and they follow like the IBM, you know, blueprint and, you know, Ren is really calling the shots in many ways and, but then you do have these really, ambitious young engineers at the bottom and they can't work their way up as much as they could at some of these startups.
And yet so many people still want to work at Huawei, beyond just getting paid enormously well. So, yeah, I think that's, it's just an interesting company to see that.
Steve Hsu: Quick, quick comment on Huawei. In their internal, I've watched some of their internal like videos and stuff that are really made for like a kind of internal consumption. And they really do lionize, they have a thing like young geniuses. Like Huawei recruits young geniuses. And the young geniuses are given a certain amount of leeway to do interesting stuff. So I think it could be both. It could be like heavily controlled at the top level, like an army, like, you know, like,
Taylor Ogan: Yeah,
Steve Hsu: This is because people in the West are confused. Like, you know, they run it like a military thing. They confuse that with its part of the PLA.
Those are like actually
Taylor Ogan: Yeah, right, right. Exactly.
Steve Hsu: But it could be a well structured big organization, but they are deliberately trying to recruit the smartest people they can and let them do like have the freedom to do stuff and then
Taylor Ogan: That's exactly. Yeah.
Steve Hsu: Yeah, how to deal with it later after they invent it. I'm pretty sure that's the actual structure of the company right.
Taylor Ogan: That is. And also those videos you're talking about, that they have their own company newspaper and, and yeah, other companies have this, but they've always had this. So, that company morale is very unique. I mean, they obviously, they literally build their own cities that are kind of cool or kind of weird, however you want to look at it, but they do encourage everyone to, to be here for life.
And, and yet, yeah, they are attracting probably not the top 100 people, but maybe the 500 to a thousand smartest. and, and yeah, they, but that's why they're doing things kind, they're, they're releasing things kind of like a startup would. and, and I, I kind of wonder, like, did, did people in the West just start asking random people in China, what kind of company Huawei is? And they said, yeah, well, it has like a, a military kind of structure and then they thought as a military company, but, um,
Steve Hsu: I think there are two, there are two things that happened, which were sort of accidental. So one is, you know, the founder was in the military, but of course he left the military for a long time. Yeah. And he had been out of the military and been a scrappy entrepreneur for a long time before, you know, he really got anywhere with Huawei.
So there's that meme. And the other meme is like, they did have some weird, like, what's the right way to do it? Team building retreats where they actually brought a team to a military trainer boot camp, but I think that was more just like goofy Western like team building stuff, but they happen to choose a military thing.
And that, but that sort of took on a life of its own. So, so people. And also, I guess in the West, they wanted you to think this is part of the P. L. A. And that's why you shouldn't install their networking gear and stuff like this. But I don't think they really are at all, like, very connected. Yeah,
Taylor Ogan: No, no, not at all.
Steve Hsu: Yeah.
you know, one area, though, I think where Huawei is getting, like, really, the top people is. If you look at the chip war, the hardware chip war, Huawei is playing a big role. Like, obviously, you've got SMIC on the front lines, but there's a deep collaboration between Huawei and SMIC. And like, I think, for example, if you look in that supply chain, there's so many pieces in that supply chain, including, like, EDA design software, so the software they actually use to design the chips.
Well, I think Huawei, those guys basically just built their own, maybe first pirated. Then recreated their own EDA. So if you're a smart, like super ambitious, smart person, and you want to, like, be at the frontier of the chip, you know, the semiconductor industry, Huawei is not a bad place for you to actually go, right?
And they had one of the things that caught my eye about them before the chip ban, was they were designing CPUs that were competitive with what Qualcomm could build or what Apple could build. Well, you know, before they were fabbing them at TSMC, but, but that would attract a lot of smart people. It's like, if you want to be on the cutting edge of chip design in China, well, Huawei is not actually a bad place to go.
So, I think there are areas where like, wow, if that's your interest, they, Huawei could, do have a call on literally the best people in that particular vertical.
Taylor Ogan: Yes. And, and I think as we talk about the U.S. adding these companies to entity lists and, you know, blacklisting them, the response from Chinese companies, they like to ask me about it just because. They're not talking to Americans that often or ever,but I also like to ask them their personal reaction to it.
And it's kind of funny. Recently we visited BGI's new headquarters, which is awesome. and, and definitely probably the only Americans to, who are not in, you know, the scientific community. and, and I asked them, I said, do you like to personally feel the attacks or impact? And also, as like, has your corporate culture changed because of this?
And they said, most of the people I talked to, they said, no, not really. That's just politics. And I was like, but you guys are like, you're really like that, that's a, there are more direct companies that I could understand, but you guys are not one. So they didn't feel that way.
But then Huawei. They, as we're talking about kind of a military response, they treat it like it's a war. And I think it would be hard to really figure out how to quantify this, but I think every time Huawei was added to another U. S. government entity list that sped up their chip development 2x. And I was just talking to someone who worked at Huawei out of college. It was his dream job. And now he actually worked at BYD, which is interesting because BYD was always the stepping stone to Huawei. but he said, and he was someone who would be in the know. He left in early 2023. I said, did you guys have a sense this phone would be ready by basically mid year and he said, no, that was supposed to come out 2025, 2026. So think about how much happened in such a short period of time, but also they are recruiting people who were not working at the company when all of this happened, because they, this is kind of China's fighting, you know, warrior, and that attracts a certain type of engineer.
And so there are, I mean, there are teams at Huawei in two different parts of China that are working around the clock who don't have to do this and they are literally cut off from their friends and family just because the work they're doing is so sensitive. And imagine, imagine how quickly they're going to accelerate, Chinese complete domestic chip development.
there,
Steve Hsu: You going to say that they're working on EUV lithography or?
Taylor Ogan: It's kind of funny, like the government sometimes they're, they're aware of this, right. When it comes to something this sensitive. There are some press releases that if you read between the lines and you know what to look for that hint at this, and one of them is, is in Shenzhen, that's the bigger breakthrough, lab and.
You can look on Google Earth. It was, there was nothing there and now there is something there and there are thousands of people working there. The other one is around Shanghai and that is more for the next chips in the next phones.so yeah, the bigger lithography breakthroughs are in Shenzhen. But these people, they, they're basically like, you know, the military, except not at all, like, and, and so, yeah, like they can, they can communicate with their families.
But they're, they're working their asses off. And that's because they want to, it's not, this is not like factory workers. This is like some of the smartest people in the space. And they're all Chinese.
Steve Hsu: The way I would analogize it, if you have a broader picture of the tech competition between U. S. and China or the West and China, like, EUV is really a significant bottleneck and if, in one scenario, if they don't get EUV for 10 years, you know, I'll buy some more NVIDIA stock because a bunch, a bunch of competitors that were going to come out and compete with NVIDIA suddenly don't have a way to fab their stuff, right?
So it is a super important bottleneck. If I were the government, I would be pouring, whether through Huawei or through government labs or a bunch of companies, I would be pouring tons of resources into that. And for the people involved, they would think about it like the Manhattan Project. It's like, well, I'm not joining the army.
This is super important for the development of my country. It's got to be secretive. The rest of the world is against us trying to stop us from being able to do this. They don't think we can do this. So I could easily imagine that mindset. It's like, not, not hard, right? So you could have like five Manhattan projects going on around EVE, right now.
Right. I would be surprised if there weren't multiple projects like that. Yeah.
Quick question before we leave DeepSeek. Are they raising capital? And would you know, like would you get a look at the term sheet that they're writing? No, it would. Is that because you're not, cause you're an American or.
Taylor Ogan: I don't think so.
Steve Hsu: So, so you think they would, you think they are raising, but you wouldn't get a look at it.
I don't think they are raising, from, they're not raising from the government. Let me just put it that way. And I don't think they're raising from smaller firms. Got it.
Taylor Ogan: If they're raising at all. And I don't think they need to. But I, that's, I wouldn't even try to throw my name in the hat. Yeah, not even the American thing.
That's just, that's, you
Steve Hsu: you might not have been shown the term sheet, but somebody you know might have. Okay.
Taylor Ogan: Yeah. So, I mean, maybe that's like how secretive they are about this, but, I think also that, that kind of distracts from what they're doing. They have partnered with pretty much everyone in about a week. I don't know how you have that many meetings. Yeah. So yeah, I think that they're probably not focused on that right now.
And there's a lot of revenue coming in this week, you know.
Steve Hsu: Interesting. Good. So, let's see some other things I wanted to talk to you about.
So, you mentioned this, this meme that there's, there was an idea that China can't keep up with the West because these companies don't have enough autonomy. They're not independent enough. They have to go to the government and get approval to do stuff.
And that's going to limit their pace and related to that, but not exactly the same thing, is the idea that, hey, the whole startup ecosystem has self-destructed because, you know, there's just less venture being invested and, you know, they crack down on Jack Ma. So that's a negative signal to startups that, hey, you better toe the line or you're going to be in trouble.
How do you read that? Like, I think you just said companies are acting independently and aggressively and not consulting the government. What about the general mood as to whether the government favors entrepreneurship and tech companies and stuff like that? It seemed to me when I was in China, it was kind of obvious the government is pro that kind of stuff, but people in the West seem to think otherwise.
So, what do you think?
Taylor Ogan: I definitely VC has, has dried in China. There's no question about it. but now these huge companies are basically VC funds. And, and they're not quiet about it. I mean, sometimes they are, but like Alibaba sometimes likes to be really quiet. But, and same with Tencent, but you, you do have
VC activity, around these, these big headquarters. I mean, some of these headquarters literally have entire, like little cutouts of their campuses for this. They have incubators, in different cities, you know, so there is still capital being deployed for sure. The government, so the government has their own like VC funds, right?
The Shenzhen government is probably the most successful. and so they're also not really shy about that. I think there has been some, there's a lot of resistance to taking money from these kinds of state sponsored funds, because then you are, I think actually people learn this from ByteDance and they don't ever want to be accused of this. Because I think entrepreneurs in China are now setting their sights beyond China. It used to not be the case. They never saw beyond China. If they want it to be a car company, they want it to be the third largest car company in China. They had no aspirations elsewhere.
Maybe like in other Southeast Asian countries. But now these super young entrepreneurs, they're, they're all born in the nineties or even early 2000s. That's the generation I'm talking about. They set their sights globally. And one of the building blocks to do so is to not have any suspicions of being involved with the government.
And I think the government knows that as well. And they're also cautious about that. So, it's different, it's a different space right now. But it's still exciting.
And I think next you have the smartest people in China returning to China in droves that you've never seen before. They're not leaving.
and I think DeepSeek that team really is. If you look at their backgrounds, like literally every single one of them, I believe at least the core 50, as they say, they're all educated in China and now, you know, what nature's rankings have eight of the top 10, you know, top universities. And so I think this also is coupled with the obsession with R& D. That is something that has stemmed from the government just in terms of rhetoric, but there is such an obsession with R& D in China right now. And that's driving the youth. It's inspiring everyone. And so now this, if you want to build a car, for example, like let's say Xiaomi, right, which deserves its own section, in my opinion, they had competing teams to build the whole car.
And, I mean, Apple kind of does this right but you know Apple sends like, you know 15 of their American engineers to Huaqiangbei and Shenzhen for a week and they see what they come up with. This is different. This is because they have complete marketing teams. They have designers within this and then they come out with this, you know, clay model and, and also the innards of it.
And they present it to late June and other board members and such, and they approve the winning team. The winning team then is assigned with that project and the other competing teams either go on to do something else or work under that new team. And so there is actually upward mobility for young people.
And Huawei is doing this as well. So if you want even to be a designer, like BYD has more auto designers and the rest of the industry has auto designers. They just have so many in the last two, three years, they've just recruited them all, even, even in Pasadena, like they have some, they're actually quite a few people working.
So they, if you want to be a really top designer or the head lead designer of a project, you can throw your name in the hat, your design in. And if it's chosen and you can kind of defend it, you are one of the leads on that. And so that's something that the structure of these companies was so regimented before, in China, especially.
And now that's, I think companies just have learned from each other. That's no longer the case. And, and so that's why we're seeing a lot of exciting activity. Across all industries, like that's the only way really to describe this across the board, because if you just saw it in one space, it wouldn't, that wouldn't really satisfy.
So it's, that is, that's been a huge corporate shift culture.
Steve Hsu: So I think you're saying that, okay, number one, there is just a big interest in R& D. In China, not just in the private sector, but also the public sector and what kids aspire to. And I would actually analogize that to the Sputnik era, which you're too young to remember, but, but when I looked at the textbooks and the various things that I was studying from in the 70s, when I grew up in the 80s, I could see that they dated from the Sputnik era.
So there was a huge, like, Revolution upending of the U. S. Education system in response to Sputnik. Like, we just suddenly decided we got to produce a lot more engineers and scientists and the curriculum in high school is going to change. And, you know, we're going to have like rocketry clubs and stuff. So something I think is similar to that is happening in China.
Now, I think you're also saying within the big companies, established companies, they're creating situations where there's competition and creativity within the company. So there's that mechanism. But I'm still a little concerned, like if a startup coming out of Tsinghua or whatever, it's a, it's a Zhejiang university, can't get like a 5 million seed round or 10 million seed round to do something, which is like totally out of the box.
Eventually there's a problem, right? Because you still want to have that part of the ecosystem. And I think what I'm hearing you say, and what's been reported is that part of the VC ecosystem is somewhat depressed right now,
Taylor Ogan: Yes, yes,
Steve Hsu: I did meet with a very big venture investor in Beijing when I was there.
And he said his ecosystem is still healthy, but in general, like the total amount of capital being deployed is less. And he even told me about his Western LPs. He has a few dots that are really like true blue diehard LP big, you know, pensions or university endowments that are still willing to put money to work in China.
But a lot of them are just like, you know, China, uninvestable risk off China or whatever it is. But yeah, so that, I think it's still a challenging time for that specific thing, even if there are other sources of innovation in the economy.
Taylor Ogan: Yes. But that's definitely true. What happens when that changes, when you have the smartest people in the world in huge concentrations that are given the cash to do that to really think outside of the box? The other part of this though is maybe it's coincidental, but they're still the kind of free spirit Entrepreneurs are not common yet.
I think they're in university right now probably In high school, people talk about where the next Steve Jobs is, right? Is he alive or is she alive? And I definitely am biased in who I'm asking, but they think that they're probably a middle schooler or high schooler in Shenzhen public school.
And by the way, these Shenzhen public schools are nuts. They like to have their own laboratories. They, they also do like, Huawei will sponsor their computer lab. And it's like nothing you've ever seen. So they're, they're very fortunate. And there is a huge discrepancy between tier ones and everything else in China.
But like, let's just talk about what is happening instead of, you know, is it happening in every single school?
Steve Hsu: Yeah, even if it's, even if it's only the tier ones, it's like 100 million people. 200 million people living in those cities, right? So it's like a nation of people,
Taylor Ogan: Exactly. and just the way they're wired, once they are able to kind of release their creativity, I think that there is, there's going to be a reckoning. I don't see how there won't be. And I think that, yeah, it is part of how they're wired. And that is changing. I think the schools are still probably encouraging them to be more conventional.
But there are risk takers and you can see this also in just fashion. I don't follow fashion at all, but I have heard that this is happening where it's so unconventional and this is something that maybe some crazy designer out of Brooklyn or London or Paris may do but now this is like a 25 year old Chinese girl who didn't go to college. So I don't think we're going to see a huge amount of like college dropouts, you know, waves. but yes, when there are enough people who are thinking completely outside of the box, it's going to be awesome. And I think the fund they'll find will be allocated for that, but that is not quite, we're not quite there.
It's something to look forward to, but yeah, so I think it's kind of both. But also something not to ignore is that these crazy, maybe not quite Sputnik moments, but huge breakthroughs are happening within these massive companies.
And I think in terms of motors,some inverters, like these are small things that maybe don't deserve a huge announcement that is happening at Huawei at BYD.
And so that's, that's also really exciting. There, you know, we could see DeepSeek moments in terms of just surprising everyone from these massive companies. But we also could see them from completely unknown companies that were started next month.
Steve Hsu: I agree with you. I think it's happening all over and it doesn't have to be through the venture capital model. I think there is like when the red note moment happened. You could see there's tons of fashion going on in China . It's got its own moment, its own vibe, and it's just totally different.
But a lot of Americans who saw it were really into it. So if you followed people on TikTok who were really into fashion, Americans or Westerners who are really into fashion, and then they discovered what the fashion looked like through Red Note in China, they were really into it. So it wasn't like, oh, this is shitty Chinese fashion . Wait, these people are really, this is really interesting stuff, right? So there's a lot of stuff like that. Even that robot that washes your window. I remember during the TikTok Xiaohongsu moment, there were some of the American TikTokers who were like, I was watching this woman on red note and I saw this thing on her window washing.
And I was like, This lady lives in the future because I've never seen anything like that, but some random company, probably not venture backed, probably just some company that is good at simple household robotics, just built that thing. Right? And so that, like, that's still a source of innovation, right? Even though it's not a venture capital model.
Yeah,
Taylor Ogan: And I think, I think sometimes these, these little innovations are, are born just naturally, it seems. And because you have such an advanced supply chain and manufacturing capability, you can just realize these things like I can't explain how easy it is to just make a custom chair for my balcony, like, even, you know, on the simplest levels.
But it's also exciting. As you say, fashion. Yeah, I, I've heard of that happening. I just went skiing in Xinjiang, for the second time, in this tiny village that actually is the birthplace of skiing. A lot of people think that it was in Europe, but there's evidence that it was actually in China.
And I grew up in Colorado, so I've skied, you know, world class resorts and mountains. skiing, now this resort in China is only three years old. It 's snowing more than I've ever seen. It has more terrain. The plant, this is just phase one. The next phases. it will be the largest ski resort by far if they build it out.
and yeah, it's, it's hard to, to get to. but last year when I booked this, I knew nothing about it. I couldn't research it in any way, really, except just where is it on a map? The hotel was still in trial operations. They didn't have a website and I got there and it's like a paradise and there aren't that many people, but there are enough, and I was shocked that the Chinese.
Well, it's like 90 percent snowboarders, but they're incredible. And I asked them, where'd you learn how to snowboard? And they said, indoor, a lot of them like in Guangzhou. and, and so, so then you see, they just have Xiaomi walkie talkies on their helmets and they're messaging each other, when they don't have cell service, you know, their Huawei's, which apparently T Mobile is coming out with something that has existed in China for years. Yeah. and they don't even have low earth orbit yet. So, but so then in just one year, I've seen so first of all, the Chinese now own all of the brands. They are classics. A lot of people don't know it's mostly Amra sports.
But the smallest little innovations, they're just, they're just being born naturally. the little gadgets they have in their helmets, the ways that they don't have their goggles fog up. I mean, the ways that their boots snap in, it's incredible to see this in every different element.
Now, all of these malls. Do they have these AI fire extinguishers and that just happened? There wasn't a huge, you know, there weren't Morgan Stanley papers about the future of fire extinguishers but so this stuff just in every aspect of life you can see it changing.
And so that is something that is hard to explain but that's setting China so much further ahead of the west.
Steve Hsu: I'm totally with you. You know, it's so far from, you know, there's this stereotype that we've been talking about. Oh, this is a centrally planned economy or it's going to be growth is going to be suppressed or innovation is going to be suppressed because it's an authoritarian state. But actually when you go there, it's like relentless competition in the marketplace. Lots of people who have their own factory can still build little things. They can inject mold. They know how to put electronics together with a little AI brain. They're just, they can do all that and there's just relentless competition for everything.
Even the way like the zipper or the Velcro like on a jacket is done. It's like done in some city where all they do is make jackets, right? You know, like down jackets or something. And they come up with some little better way to fasten, you know, your collar or something. I noticed that stuff. It's so opposite of what the picture that people here have of it. It's actually, I would say, like the most relentlessly competitive and innovative economy.
Yeah.
Taylor Ogan: And that's exactly what I was saying earlier. It makes my job a lot easier. Not that, you know, my due diligence cycle is shortened because of this, but it's, it's impacted by it when these products are, are big enough. There are so many checks and balances throughout the supply chain, their competitors, the Chinese are the first to try to disprove something.
They're the first to say that, you know, BYD's newest Yong Wong can't actually go through water. That's just like a hoax. And then, but then when, when you don't hear that about enough things. Then, you know, okay, that's legit. And, then the only question is like, does it overheat or, you know, what's the cycle ability of, of this, you know, crazy battery that they just released.
But that is, that's the new reality. and so I think still people in the West will just continuously just say that there's no way that, you know, X, Y, and Z exists. And if it does, it probably doesn't work. And I've bought things from China that didn't work in the past. You know, so that's why like even these Shein moments and, and, you know, all of these little tastes that people in the West get of China, I think maybe that will compound enough where they realize the person that I bought, you know, 10, you know, pairs of, of pants, custom made from China can actually do way more. And I just looked at their website, you know, on Aliexpress, and I didn't realize that I could have my whole wardrobe built out. Maybe that's a bad example. But, I think they are turning on to that.
And, and, you know, it's kind of grassroots even. Sometimes it's, it's farmers. Someone was just telling me that they, their favorite thing is to watch farmers on TikTok, in the West with DJI agriculture drones. And they're obsessed with them and they know all about them and, and they're even pronouncing some of the Chinese names correctly.
And some of them have even come to Shenzhen to like, you know, tour their factories. I'm really optimistic about this kind of grassroots movement, especially the younger generation. I think they're more open to it and they see the older generations as out of touch. But, but we're, we're getting there, but we're not even close to being there.
Steve Hsu: Yeah. I would agree. I think that awareness is increasing in that Xiaohongsu moment. It was just for a moment because now people have gone back to using TikTok. I don't know what's going to happen at the end of 90 days, but, but that was a kind of like microcosm of what could happen if Americans suddenly got VR and they could wander around Shenzhen, you know, they would realize, oh my God, there's all this innovation there that I didn't know about.
But I think that will gradually happen.
One question I have, which is directly relevant to your business is, are we seeing a change in perception that China is now invested, investable? Like, when are we going to see, because we just did see a run up, maybe that was all due to domestic investing.
But because of all these tech leaders who met with Xi, was it one or two weeks ago? Suddenly there was a run up in the market in some of these valuations. Are there now more Westerners who are like, Hey, I should, I should put 5 percent of my capital in that market because obviously stuff is happening there. And maybe it is investable and any thoughts on when that will happen?
Taylor Ogan: Yeah it hasn't happened. And, ,
I mean, there are also Chinese who are, you know, not in the investment space, but are connected enough and highly intelligent who are also still wondering if China is investable. And, and that's something that, that really shocks me, because you're living and breathing it.
You know, that, you know, it's very real now. But I think we see right now, with the Hong Kong stock connection, there is still, in the last month, a lot of southbound volume. So those are still shiny mainland Chinese investing in the Hong Kong, you know, version of the stock. So some of these dual listed companies, that's, that's, it's, there is more movement.
Definitely the Hong Kong listed H shares. Those are gaining traction, but the Asia market is really what's still so untapped. And even throughout this month, it kind of remains untapped largely. There will be a shift into it. I don't know what it will take, but these companies are hugely undervalued.
And they also, they don't know what they can accomplish when their stock prices do increase significantly. They can have a lot more capital to play around with, and really build out things. So you kind of, you can see it in like the Alibabas that are well known by Wall Street. That's where a lot of volume is moving the last few weeks. But that's like, that doesn't really satisfy me. You know, that that's not, it's kind of just a blip.
I think some hedge funds definitely have moved back into some of the bigger Chinese names, but that's not what we're talking about. You know, we're talking about like your friends who just invest on the side when they start, you know, they have 20 percent of their portfolio in Chinese companies. That's what we need to get towards. And we're not. I think we're moving in the right direction but again, that also has to happen in China. And I think now with the housing crisis and you know, the Chinese are awesome savers. They are wondering what they should do with all of these savings if they can't just buy five homes. So will that be directed, you know, towards stocks? I think so, but we still have to wait. But that's not to say that people really are following what's going on and they are realizing that people in the West are using, you know, four of the five top downloaded apps in the U S are Chinese. Like that's pretty self validating for them.
And then they can go and say, well, now I can use them as well. It's a weird cycle, but it is happening. So I think we're making progress.
Steve Hsu: Do you think the property bubble has hit bottom?
Taylor Ogan: No, but I and it differs by cities. And, you know, residential versus commercial. But it's near the bottom. And people don't seem to be as concerned by it. But they also feel burned, that's for sure.
I think it's, yeah, it's not nearly as much of a concern. Five years ago, the Chinese didn't, the average Chinese person didn't have as much to look forward to, regardless of, of, you know, class.
And now they have at least a lot to look forward to. They are seeing their cities improving in front of their eyes. The, you know, the, they're still building like crazy. I'm looking out my window and it's only construction basically still. and, yeah, so, I mean, last time we spoke, this floor didn't exist. And that was only like less than a year and a half ago. So, so there, there, everyone is seeing this happening.
And they're even the smallest things like, wow, China is now doing really well in the Olympics. I think everything is directionally up and to the right in China overall. So instead of focusing on these little things, everyone realizes that they're on the right track now.
And that wasn't the case five years ago. I don't think
Steve Hsu: Yeah, I mean, I could sense it when I was there. I mean, obviously, you could talk to people and, you know, you could sense, okay, there's, there is a problem with the property bust and stuff. But on the other hand, it was still obviously a very, very vigorous economy with lots of innovation.
I wanted to ask you about, I've been, you were speaking, you were talking about Alibaba and I've been kind of tracking them thinking about whether I should buy some and, I guess, they just explicitly embraced the AGI and, but the market reacted negatively. Did you, are you following that kind of stuff or yeah, so it's kind of funny because when you talk to Silicon Valley guys, one of the things that they're, especially the ones who are really AGI focused.
One of the things they cling to is like, well, Xi doesn't get, Xi Jinping doesn't get AGI, and yeah, these Chinese companies, there are some good AI companies in China, but they don't really get AGI either, they're not racing toward AGI. And so, that's what's going to win it for us. But, it's kind of weird to now have Alibaba come out and say, hey, we are an AGI company.
And we're going to put $ 50 billion in infrastructure in the coming years. But then the market didn't receive it well. So I'm curious, how do you interpret that?
Taylor Ogan: How the market interprets Alibaba, especially these days, is incredibly confusing. I think a lot of people did not know Jack Ma was, you know, let out of his cage, basically, when they thought he was put in it. He, I mean, he, we track his, his jet, and he is still involved in many Alibaba things, but also in his own projects, like farming and wine and things like that, that many billionaires are.
But in the last year I have been flying to China constantly. And he, I think him being in the front row, especially, that was a signal, you know, we talk about signals all the time with the government. And sometimes it's not a signal. Sometimes it really is a signal. And that was kind of both. I mean, they were just a relevant party to the party, I guess.
And, he was their representative. And the fact that he still is their representative is I think shocking to a lot of investors. So I think they are still focused on that. I also think they are calling themselves an AGI company while Jack Ma is now back in the spotlight, maybe it has to do with Jack Ma.
So I think they're also, you know, it's a recruiting tactic if you aren't saying that you are in the most cutting edge forms of, of AI, then you're not going to recruit the top people.
And there's been some flight from Alibaba. So yeah, I think it's, it's a bit complicated, I think, but what's more complicated is the stock price. Not to say that it's too volatile. It's just these like, you know, five day reactions, or announcements on a Friday China time. And the U S has a trading day too, to make sense of it. But then that wasn't taken the same way on Monday morning, you know, China opened like that's where some things can be complicated. But if we look month by month, there's excitement, in these, these types of statements from these big companies. We just saw the same thing. It's a very similar thing with BYD. And finally getting into, you know, level two, eight ass and level three. and yeah, so I think these still, these big companies have huge announcements and, and directional shifts. That's something else that just because you heard, I mean, I can think of so many examples, but maybe one of the best ones was, was Ren Zhengfei saying that Huawei will never make a car, you'll never see a Huawei logo on a car. And they're basically, you know, one of the leading car partners. and yeah, technically there still isn't a Huawei on the outside of the cars. Actually the new M9 does so, but they can change.
And so these founders are all born within the same few years. They all come from, you know, a very specific generation in China. And they think they know what, where they want to take their company, but they're also open to this new generational mix up. And that's, it's exciting because I, I didn't expect it to be such a fluid shift as it appears to be. And yeah, the same thing is happening within Western companies, but the fact is that they're happening in Chinese companies.
Steve Hsu: It's so funny because here it's the Mag 7, and you could even argue there's a Mag 7 AI bubble, depending on how you view it.And then like here I am tracking all the, the quality of all the Chinese. LLM models, right? So, you know, ByteDance has a world class model. Alibaba has a world class model. You know, Tencent has one, DeepSeek.
So there's actually, there's actually quite a few. And they're getting no credit for it, right? Whereas, the US companies are getting tons of credit for their AI activities. And so part of me when I saw Alibaba say that was just like them figuring out like, well, maybe the capital markets will give us a big valuation boost if we just mimic what Microsoft and Google and these other guys are doing, and actually our models are about as good.
But it backfired on them and they lost like, I think they were down 10 percent or something, right? So crazy.
Taylor Ogan: But it's like, why did, why didn't they just do that when everyone else was doing that? And, and I don't know, but those kinds of things I think are like, did anything change in their, in their actual internal corporate strategy? No.
Steve Hsu: I don't know. Well,
Taylor Ogan: But if they, if they suddenly said, no, we're going to be more aggressive with data center build outs and put the money there and Microsoft is maybe getting bonus points for doing that. Like, maybe we should get bonus points too. I don't know. But you know, here's a funny thing. When DeepSeek, when R1 came out, I think it might have been the same day Kimi launched the Moonshot, which is the company that makes the Kimi model.
Steve Hsu: They launched their reasoning model, which on the benchmarks is just as good as R1. And I've tested, you and I were just talking about Kimi before we started recording. I've used it a lot. It's really quite good. So, I think people in the U. S., like, first of all, like, they didn't think there were any good models in China.
Then, now they think, like, oh, DeepSeek. I've heard of DeepSeek. But they haven't heard about all these other guys. And the way they value these companies is not at all how they value equity companies with equivalently strong AI capabilities in the West. So, there's some, there's some big disconnect. Yeah.
Taylor Ogan: You can say, see this with. A lot of products, look at smartphones, like if, if you gave, you know, a high school or 10 phones and, you know, probably six of them would be Chinese companies, they would probably rank the top five phones, the Chinese phones, if they weren't, you know, in this Apple ecosystem.
And I mean, you know, I have Huawei's newest phone and it's, I can't believe that I used it. I mean, I still have my iPhone, but,yeah, like the Chinese though, I still would like to reference Western rankings that don't include any Chinese phones in them because they just, the reviewers can't get their hands on them.
So yeah, I, I think like how, these benchmarks are really interesting because you can't really ignore them. and it's, you can test it yourself, but plenty of others have tested it and here are the results and it's constantly updating. I still think that's changing, especially if the Chinese want to make something of any kind of product, whether it's software or hardware that they want to have global recognition for, they kind of have to make it global.
and so on. Like Xiaomi taking their car to, you know, Nürburgring, that's a good example of that. But then yeah BYD taking their, you know supercar to a Chinese equivalent track is not the same, you know, and it's still, they don't even have a Chinese name for the car. Like, and that's, that's fine.
That's their strategy for now, but there are companies that are trying to come out swinging globally and others are kind of doing it, you know, a different strategy. But we're seeing, we're seeing both. And at least we're seeing both because it never really was the case.
Steve Hsu: Yeah, I mean, arguably, if DeepSeek hadn't opened, if they weren't open sourcing their models, people here would just continue sleeping on them because Kimi is not open source, but it's freely available. You can use it, but you know, you can't customize it for your own needs.
Taylor Ogan: Yeah. And even making it convenient in English, like that, is a big aspect of this. And, yeah, I think, BGI actually just released a model and it was really complicated. Like I had my whole genome sequenced by them a few months ago. And, maybe we can talk off camera about that, but,you know, super exciting, a thing that, you know, under a thousand dollars is crazy.
But now you can upload that to their large language model. And, but yeah, the. Onboarding was quite complicated. So even though I have all the reason to find my, you know, file and upload it, yeah, it's, some of these are too complicated. So, Kimmy used to be pretty complicated. That's the other thing in China it wasn't, but in the West it was.
So yeah, if you can easily download it from the app store and the English is not broken. I think that's also a huge difference. And you also see the intent of these, at least in the large language model space. Are they trying to make this really global, or is this not that big of a deal?
And model to model, it changes. Like iFlyTech, it's also one of the leading AI companies. Sometimes it's like, you're so excited about your Chinese publication, but you don't even have anything in English on this. So maybe this is just, you know, a mid cycle model. But yeah, I think that's why DeepSeek was just so, so enormous, because they just did everything at once.
Steve Hsu: You know, under the radar, the Quinn models from Alibaba are extremely good and they have very small models that can run on edge devices like phones. Those are extremely good. And I think what's going to happen, because now there's a deal between Apple and Alibaba in the China market, they're going to be using these small Quinn models running natively on the phone.
And if they realize that works well, that's going to come back to the rest of the world. So, possibly it'll be Quinn models running on iPhones all over the world in a few years. Very interesting development.
That would be interesting. Yeah. So, okay. We've been, we've been on for 90 minutes. I think we should, I should probably let you go.
It's been fantastic. We're going to have to do it again. I wish you all the best and I'll probably be in your neck of the woods. Maybe April, May, something like that. So I'll definitely look you up.
Taylor Ogan: Okay. Perfect. Yeah. It's a good time to be here.
