Podcast

Decoding China's Rise In AI - Grace Shao

Grace Shao is a Financial/tech journalist turned analyst based in Hong Kong. On her Substack, she writes about AI and its relationship with energy, big tech, and society.

TIMESTAMPS:
00:00 - Trailer & Intro
01:30 Understanding the Chinese Internet Ecosystem
06:06 The Evolution of Chinese Entrepreneurs
08:21 China's Smartphone Revolution
10:45 The Rise of Super Apps in China
15:32 AI Adoption in Chinese Industries
19:16 US and China – Difference in AI
24:07 Comparative Advantages: US vs. China in AI
26:16 Chinese Cultural Attitude Towards AI
28:11 AI in Manufacturing
32:17 America's AI Advantage
36:08 Navigating the Chip Ban and China's Self-Sufficiency
37:43 Funding Challenges for Chinese Tech Companies
43:23 The Evolution of Tech Entrepreneurs in China
49:20 The Role of Education in Shaping Tech Talent
55:21 China's Tech Role in Southeast Asia
58:25 Advice for Fresh Graduates

Keith 01:30

Today, I'm joined by Grace Shao, one of my favourite Substackers and analysts. She writes a wonderful Substack called AI PROEM.

The thing that I really like about Grace's writing is the fact that she's a third-culture analyst. She's able to understand both the Western and Eastern technological ecosystems really well. With that, Grace, thank you for coming on.

Grace Shao 01:52

Thank you so much for having me.

The Rise of China's Internet

Keith 01:53

In Western media, people dismissed the Chinese internet ecosystem for a really long time as an ecosystem that just copies whatever Silicon Valley does. Whatever spawns out of the US, China just tried to ruthlessly imitate it. But somehow there seems to be a missing piece of that puzzle. They miss some very crucial facts in how China has developed as an internet ecosystem. Maybe let's start there. What do you think they're missing when they have that kind of narrative going?

Grace Shao 02:25

I think it stems from just not understanding each other or getting to know each other well enough. At the end of the day, there aren't that many people who really travel and work and live between the two worlds. I think it's really up to people like you and myself to bridge that knowledge gap.

But in terms of the internet era and how that's transformed and evolved into the AI era, the entrepreneurs we see in this generation are really different. Number one, on a very macro level, many of these people are quite young. Born in the '80s and '90s, many of them actually grew up in a pretty well-off situation. Economically, they were not struggling anymore. They rose with the Chinese economic rise in the last 30-40 years. Because of that, they've had access and abundance, at least extra information globally, but also the abundance of not having to worry about food and shelter. That's a significant change compared to entrepreneurs we've seen from the last generation.

The reason I bring this up is because entrepreneurs in the internet era, many of them born in the '60s and '70s, actually struggled growing up, even just to think about the bare necessities. So, what drove them was money. In many ways, the West wasn't wrong. Many of the starting points of the businesses, even if you think of the big tech in China, they found a business model that already worked in the US and they brought it to China. Many times, they did improve it or make it adapt more naturally to the local market. But fundamentally, the early stages of their business model were Western.

The AI era is very different. Much of the innovation is very grassroots, very new. It's completely driven by innovation from China internally, and then it goes outward. Going forward, we can talk about how the mentality of the entrepreneurs has shifted within these generations.

The Evolution of Chinese Entrepreneurs

Keith 06:06

One of the things that I picked up in Kai-Fu Lee's book on Silicon Valley and China and the rise of AI is that he points out that the copycat mentality in China creates gladiators out of these entrepreneurs. Because everyone's creating a copycat function of the hottest new feature, to a certain extent, you basically create a bloodbath where everyone's trying to out-innovate each other to fulfil a consumer market that was ever-changing. Whereas in the West, the approach towards scaling a business is that, "I've done it well in America, and the world should follow suit." Other than the prevalence of copycats in China, what other features or entrepreneurial functions have allowed an almost alternate Chinese internet ecosystem to emerge that's completely different from the US?

Grace Shao 07:12

If you want to shift back to the earlier stages of China's internet era, exactly as you said, because there's so much competition domestically, we saw more players in each vertical within China's own market. Even recently, Uber's founder talked about reminiscing back about when he was trying to enter the Chinese market and essentially got defeated by Didi Chuxing eventually. He was saying he was quite impressed by the speed of copying. There's no shame, almost, in copying as long as you can copy, copy, copy, and then innovate. You can copy in various iterations, but eventually there is going to be a breakthrough point, and they're going to be able to engineer something quicker and faster and more scale. That's something quite special and unique to the Chinese market, and it's really similar to many other industries and use cases in China.

For example, everyone knows about TikTok, but in China, people don't know that Kuaishou [快手] was founded much earlier than Douyin [抖音], although it's now maybe the second or third largest short video player in China. There was domestic competition, including when there were the e-bikes like Ofo [共享单车]. There were four or five major players at one point. When you walked around the streets in Beijing or Shanghai, you would see five or six different coloured bikes. That naturally created domestic competition, and it basically made things as cheap as possible for consumers.

Going back to the internet era, it was very much focused on serving the consumers, and that was the peak of China's consumption. This was the peak of China's GDP growth, the peak of China's rising middle class. Much of what these big techs served was a consumption upgrade. That really fundamentally changed as we saw the slowdown of China's economy these days, given that it's no longer in a hyper-growth stage.

China's Smartphone Revolution

Keith 08:21

In 2013, I think that's the turning point where many Chinese smartphones were being manufactured. Most Chinese people actually went online for the first time, and it wasn't through a desktop; it was through mobile phones because mobile phones were cheap, and you could access the internet cheaply. How did that then shape the consumer economy within China?

Grace Shao 08:42

That's really notable to point out. When we talk about why China is adopting or integrating AI really quickly without reluctance, it's really similar to what happened then. Basically, there's not much legacy holdback or mindshare held by the legacy players, because China's economy grew so quickly that actually, the majority of the population did not have access to PCs and desktops. Similarly, Chinese consumers didn't have access to Visa and Mastercard. When the technology shifted so quickly, everyone actually started picking up relatively affordable smartphones. Adoption of technology and digital payment was extremely fast in China because everyone just basically went from nothing—well, cash and no internet—to a smartphone. That was really transformative. You can definitely see that in the huge rampant adoption rate of using digital payment systems or even just being more open-minded to social media, even in the older generations.

The Rise of Super Apps in China

Keith 10:45

China's alternate ecosystem has created a different kind of environment and, therefore, creates a different kind of product ecosystem. For example, in the West, or maybe even in Singapore, we're much more used to SaaS, business, software as a service. We're much more used to using paid software products. But in China, that doesn't seem to be the case. There really isn't much prevalent office software. The first instinct for entrepreneurs is rarely to go into SaaS; they're thinking more about consumer software, or perhaps robotics and AI, but hardly SaaS. Why do you think that's the case?

Grace Shao 11:15

We can talk about two points. One is that the separate ecosystem, essentially what those big techs eventually evolved into, are super apps. Everyone has heard of the WeChat [微信] and Alipay [支付宝] of the world in China. The ecosystem is completely dominated by these massive platforms that allow consumers to do anything, from paying bills to booking a ride, to even renting a car, or doing day-to-day payments. That goes back to my point about the high adoption rate of these applications. They then expanded, and this is due to, frankly, in some ways, less regulatory controls in the beginning for their expansion and many acquisitions within the different verticals. They were able to build this ecosystem.

In that case, you can actually see, I think there was a study, most consumer mobile phone users use about 10 apps per month in China versus about 30 in the West. If you think about it, when I'm in New York, I would have to use Uber for my car, I have to use Seamless for ordering, I have to use four or five different apps for my daily activities. And then you have to go into the actual utility bill, like Con Ed's actual website, to pay for my utility bills. In China, you just use WeChat Pay [微信支付] or Alipay [支付宝] for everything; they're all integrated. That's a different ecosystem.

Because of that, it brings us to the second aspect of what you mentioned: internet players or even tech entrepreneurs have largely focused on tapping into the consumer space versus the enterprise space. With consumer applications, usually you don't really charge, but the monetisation strategy often is in ad revenue or in value-added services or premium versions. What you really need is scale. Everyone competed on DAU/MAU. The ones I just mentioned, such as Alipay [支付宝] or Douyin [抖音] or WeChat [微信], these are upwards of a billion MAU; this is a huge number, a huge base.

In the US, if you think about the recent 20-30 years, the economy is very much driven by knowledge workers, and in that sense, there was also huge respect and legal protection around copyright issues. Much of the business was trying to make money off of other businesses because many professional services were in the US. We think about Salesforce and Monday.com and whatnot, these are the SaaS software that almost every single company will adopt. In China, there aren't really notable huge SaaS software, at least not at the same scale as the American ones. But because there's also fluidity and almost a cultural adoption that people would even use WeChat [微信] for work communications, there were attempts, such as by ByteDance [字节跳动], they had a Lark platform, as well as Alibaba , they have DingTalk [钉钉]. These were all Slack-like applications, but they never scaled the same way as they did in the US.

It's also because of the makeup of the economy. China's economy makeup is not predominantly driven by knowledge workers; we know that it's very much manufacturing-driven. Many people frankly still work in farmlands and rural areas; urbanisation is a relatively new concept. Knowledge workers still make up a very small percentage of the whole labour force. For these knowledge workers, frankly, many of them still work for SOEs, which are state-owned enterprises, or small to medium-sized businesses, many mom-and-pop shops. Their goal, in many ways, is sometimes not about streamlining processes or increasing efficiency. It's about what the boss feels like, what the boss wants, about cost efficiency. There's just a completely different work culture.

But I think that's definitely changing, especially in so-called first-tier cities [一线城市] in China.

The adoption rate of respect for knowledge workers, as well as the understanding of tapping into these professional verticals, is definitely more prevalent. We can talk about that later as well when I spoke to a few AI companies where they're really trying to actually tap into that for value capture versus completely scaling and competing on the consumer side.

AI Adoption in Chinese Industries

Keith 15:32

I spoke to Professor Ke Yujin in the past, and she said that China's aspiration was never to be like the US, but to be more like Germany. In that sense, there really is a drive to become much more industrial, much more manufacturing-oriented in its economic makeup. So far, we've really framed the discussion on China as a consumer market, a very large consumer market, and therefore, it has created certain incentives that drive consumer tech adoption. But at the same time, it is also a very large industrial and manufacturing economy. How does that shape the way tech develops, maybe in areas such as robotics?

Grace Shao 16:13

I think this is a really interesting topic, and the idea of the physical AI space, or even just robots, is getting traction and attention right now, given the recent conversations about Donald Trump wanting to reshore manufacturing jobs back to the US—the discussions around whether the US could potentially use industrial robots to help fill in some of the gaps. It's interesting because there are statistics now going around on China and how they're adopting industrial robots. For sure, China has become a huge industrial manufacturing hub. This is something we all know: everything from home appliances to car parts to textile and retail products and toys. China actually has the world's largest industrial manufacturing economy, and they have the most industrial robots in factory floors. In fact, I think by 2023, there was a study by the International Robotics Federation that there are more than 1.75 million robots already being used. Now, I'm not talking about those sexy robots we see these days in the videos where they look like humans or dogs. These are not humanoid robots, but they're essentially designed to help in manufacturing studies and on the front-line factories, and they usually have six arms, so they're called six-axis arm robots, and they make up more than 50 to 60 percent of global installations.

It's really interesting because this isn't a sudden thing that just happened; it's been decades of planning that's really helped China propel to this stage. Even before the 2010s, China mostly imported these robots from leaders such as Germany, Japan, and some from the US. But in the 2010s, there were a series of national policies and ground-up entrepreneurial pushes to really adopt more robotic support in the manufacturing era. That leads to what you talked about, AI-empowered robotics. From the 2010s, we also saw China's rise in EVs, in battery solutions, and lidar sensors in various related sectors that really propelled all these technologies forward. Basically, by the 2020s, all of these technologies converged together. Being the manufacturing hub of the world, China has the cost advantage of producing essentially cheaper goods in these technologies, as well as really leading the front-line R&D. And now they're really, in a way, leading in humanoid and contributing to AI robotics.

US and China – Difference in AI

Keith 19:16

With that kind of context in the back of our mind, which is that you have a huge consumer market of almost a billion plus people, close to 1.4 billion people, and also being the largest industrial and manufacturing economy in the world, how do you see AI being adopted in the markets? Because in the US, what we see is there's much hype oriented around the LLMs, around your DeepMind, around your OpenAI, about your Claude, your Anthropic, Google Gemini, things like that. And people think of AI today, they really conflate it with, or immediately think about, the big tech products. But in China, the way people are thinking about AI now, I think, is different. In what way do you see the AI adoption in China across the different sectors play out?

Grace Shao 20:08

For sure, the big tech players are still a dominant part of the AI narrative and the AI direction, so we can go into that. But the difference is, for one, in the physical AI space, I'm sure you've seen headlines, many of the EV players such as BYD [比亚迪] or NIO [蔚来], they're quickly integrating DeepSeek into their smart cockpits. It's really enhancing the consumer experience. They're really thinking of it in a very savvy consumer way: "How do I let consumers and drivers, or even passengers, utilise this technology that's existing and open source to us?" So, they're bringing it into home appliances such as vacuums and fridges; they're actually integrating DeepSeek.

It's very consumer-savvy in that sense. That's quite different from, I think, the US or even Western markets where many non-big tech companies are still approaching AI with caution and some fear.

I think what we also touched on is the commercialisation strategy being very different, and that brings me back to the big tech players. The big tech players such as Alibaba ByteDance [字节跳动], and Tencent [腾讯], in many ways, are still really pushing forward on their commercial strategy. That commercial strategy for Alibaba in many ways, actually mirrors what many of the US big tech are doing, which is offering cloud infrastructure, compute support, and then obviously an open-source LLM, and then you can use it as an enterprise to adopt it and then tweak it as you want.

ByteDance [字节跳动] and Tencent [腾讯], they're really battling it out on a consumer application. For example, Doubao [豆包], which is the standalone application for ByteDance [字节跳动], they spent hundreds of millions trying to convert users onto their platform. Essentially, it's like a ChatGPT-like chatbot. I think at one point, they gained maybe 400 million MAUs, which is pretty decent, but that was at a cost of hundreds of millions in marketing. Tencent [腾讯] was very interesting; I personally thought it was very smart. They just integrated DeepSeek into their existing WeChat [微信] platform. I think the reason why it was so smart is, to really try to outcompete on frontier models is extremely hard, and we have the world's leading research labs trying to compete on that front. Tencent [腾讯] just said, "Alright, we're not going to do that. We're not going to be the best at LLM R&D and LLM development." But what they're going to do is leverage their existing 1.4 billion users globally. The beauty of integrating DeepSeek into WeChat [微信] is because you basically have the functional adjacency. What I mean by that is, if you open Douyin [抖音] and you're looking at entertainment videos, you're not going to think, "I'm going to ask a chatbot a question." But if you're actually on WeChat [微信] and you're chatting with a friend, and you suddenly have a question, it's really easy to open just another chatbot to ask them a question. There's a very similar user experience, a functional adjacency, to go into that.

Another beauty of Tencent [腾讯] that people are underestimating in the West is that they actually have a similar advantage to Google, which means they have a walled garden content database that no one else has access to. All these LLMs are scraping data from across the internet. But no one can get into YouTube; that's Google's advantage. Now, in Tencent's [腾讯] case, no one can get into Tencent's [腾讯] chats, all the WeChat [微信] chats, all the WeChat blogs (like Official Accounts [微信公众号] or Moments [朋友圈]), which is probably one of the biggest self-produced influencer content platforms in China. There's advantage in that, and I think that's again quite different from what some of the Western Big Tech are doing.

Number three, the last thing I want to bring up that's quite different, is again, it goes back to the culture of accepting AI being quite high. People from across different sectors are willing to try it. I think there's less of a pushback on trying it, or less of a fear, or even, frankly, less of a fear of data and privacy issues. People are just willing to pick up these free apps and try them. You're seeing people from university or high school students to a ninety-year-old grandma trying these tools out. I like to bring up one example which was really interesting: I went to Beijing for some family matters one day, and I stopped by a TCM doctor's office to get my back checked, as I had some back pain. It's not exactly the most tech-advanced industry you think of, but this lady, who is a TCM expert, said she was also talking to Kimi [Kimi 聊天助手] and DeepSeek on how to better understand some of the symptoms. It's very interesting, the adoption rate and the daily life integration across different sectors are very high. Therefore, many of these startups in China are actually finding vertical use cases for AI. Instead of trying to build up their own or compete on the LLM front, they're building startups and apps that are maybe just healthcare-related, or just used in the legal sector, or just used in fashion design. That's something really interesting. I think you're seeing more of that in the West, but China is very sophisticated, and in a way, very quick in adoption on that front.

Comparative Advantages: US vs. China in AI

Keith 24:07

The part that's always culturally interesting to me is that in China, the users tend to be more okay with trading privacy for convenience. Every society has different attitudes. Maybe in the West or in Europe, for example, they favour privacy much more than convenience. So, you have, for example, the GDPR, you have legislation, regulation around it. But within China, there isn't as much; there's almost an experimental approach towards it. Not just at the consumer level, but even at the government level, where you don't see governments trying to regulate first. They let companies do it on the ground up, and then based on what they see in the market, then they adjust their policies accordingly.

Grace Shao 24:56

I think on a very anecdotal way, yes, I would agree with that. I think you interviewed Robert Ross, and he wrote a very interesting piece, actually, not even long ago, about how there's actually trust for the government and for society amongst Chinese people, citizens, versus not much trust between people to people. That's quite interesting. Anecdotally, you might not trust your neighbour as much, but you do trust that society at large is very safe. There's definitely less of a top-down narrative around protecting your privacy and your data. And people are willing to trade using face recognition, fingerprint recognition, all of these things for convenience, for payment systems, or for trying out new apps, for sure.

Chinese Cultural Attitude Towards AI

Keith 26:16

So far, we've focused on AI adoption on the consumer side of the economy. If we could touch more about how AI is being integrated, maybe on the industrial side. You shared previously some of the tech products, consumer tech products, like your BYDs and other EV brands in China, are already integrating AI into their consumer products. But take me to the manufacturing side of the house. Where they're producing goods, where do you see AI being applied now? Where are the exciting commercial frontiers they're exploring?

AI in Manufacturing

Grace Shao 28:11

Full disclosure, I'm not a policy expert or anything, but I think one policy that's really notable for people to know is that there's the Made in China 2025 [中国制造2025] grand ambition from the government, and that's really propelled and put robots and AI development at the front and centre. What it has also done is set up these pilot zones across the nation, I believe across more than 11 cities. This is from not just first-tier cities [一线城市] or anything; this is from north, south, west, east, across the whole nation. It's really bringing in the best technology and talent into these pilot areas to try out innovation. Less regulation, let them run free and try out what they can first and then see what happens.

I think in terms of industrial adoption, it's really hard because, in some sense, AI is still very nascent; it's in very nascent stages. If we're bringing it back to humanoid robotics and whatnot, they're definitely not sophisticated enough to be operating in factories or anything yet. But that said, when I spoke to supply chain experts in Shanghai or even the startup I was speaking to in Hangzhou [杭州], they are using AI into predictive AI, not generative AI. But they are using this for predicting sales numbers in the future, seasonal upticks, or they're using AI for trends and issues that they would need for supply chain management.

Other areas, like the fashion side, are very interesting. This startup I talked to, they said previously many international brands would use their software—this is actually a SaaS company, a B2B company—and they would help them with 3D moulds, 3D design. And then obviously, they have the connections on the ground in the Hangzhou [杭州] manufacturing hub, like the textile manufacturing hub area, and that would seamlessly connect from the 3D design to sourcing the actual textiles and the fabrics needed, to the actual manufacture of the products, and then they'll ship it back to the designer houses out in Europe or the US. What they're doing now is using AI, and how are they doing that? They already have had 10 or 20 years of work experience with these companies, so they have a massive database. They can help designers use AI to better estimate how fabrics, different textures, will work together, or even as small as buttons and zippers, how they would hang on certain fabric. They would integrate that into their 3D design so you can actually lean on AI's knowledge and AI's prediction to help you design things. That has really helped them speed up the efficiency of designs as well. Many professional in-house designers are actually using their products.

America's AI Advantage

Keith 32:17

They still seem to have an advantage in terms of AI. I think primarily because of their access to high-end chips and also because the tech conglomerates there have worked on AI much earlier as compared to their Chinese counterparts. Where do you see the Americans holding more of an advantage in the coming few years?

Grace Shao 32:47

Obviously, GPU constraints and GPU access. We all know that there's going to be—it has been—an Nvidia H20 chip that Chinese companies can access, but the higher-end ones are banned. I think it's definitely been a scare. There's been a scare in China that companies are worried they won't be able to access the most high-end chips anymore. Even H20s were not the most advanced chips. For sure, as you said, the US has that advantage. The chip control, in some ways, does work to contain China's growth in the short term.

And then I think the advantage is also again in enterprise commercialisation. If they can find or build strong enough enterprise products, I think people will still be adopting US enterprise AI products for safety, for development reasons, for sophistication reasons. And it's very different from how consumers think. Consumers want things for free. Enterprises want things for safety reasons and productivity reasons. There's definitely that.

That said, I think there's also an interesting argument I'm hearing on the ground here: more and more people think that if we continue to see the chip ban, it's actually going to push China to be more self-sufficient in their own chip production and chip designing. Obviously, we currently know that the design is nowhere as close to Huawei's chips are nowhere as close to sophistication as the chips produced by Nvidia. But I just spoke to Ray Wong recently, who is a semiconductor analyst in D.C., and he was trying to explain to me the technologies, because I'm not the technical person. But he was saying that it's not that simple because once there is a real cutoff, the best alternative for all these Chinese big techs and startups is the Ascend 910C right now. They are maybe not the best performing ones, but they're actually able to process and they can still deploy highly scalable, high-bandwidth optical interconnect architectural support for these companies if there is enough space and energy power to support them. That means if there's enough infrastructure support. Frankly speaking, if you want the performance, Chinese companies have the money and the land; they can support that. It's obviously not the most environmentally friendly or cost-efficient way of doing it.

And then he was saying an interesting thing: he was saying that this is actually a really good opportunity for Huawei [华为] because currently, no one can actually access more Nvidia GPUs. So, they're all freaking out and looking for alternatives. Huawei [华为] can actually use this opportunity to really build their foundry and make advanced chips possible for China. And if they start actually building—not just designing, but actually making these chips—it would also create more competition for SMIC [中芯国际]. That would actually mean that there will no longer be as much monopolisation by SMIC [中芯国际] in the Chinese chip-making market, and competition creates progress and innovation. This could also potentially mean that Huawei [华为] itself could become more self-sufficient in the whole production capacity and increased production capacity in general. The key thing is that this is not going to deter China's AI progress, and if Huawei [华为] plays its cards right and does its R&D correctly, it can really push and propel China's reliance away from the US and actually become completely self-sufficient.

Navigating the Chip Ban and China's Self-Sufficiency

Keith 36:08

Do you see the government or maybe the companies in China actively trying to work around this constraint in the meantime? I know, for example, Jensen Huang [黄仁勋] was invited to Beijing for a huge talk just a few weeks back. Other than in the meantime using Huawei [华为] chips, are there workarounds that they're trying to use, or is it just a fundamental constraint that they're operating within?

Grace Shao 36:35

I think it's, again, I'm not a policy expert, but I don't think there's anything that's trying to work around it; it is what it is. If there's an actual ban, it is a ban. And in fact, I think as I speak to people in the industry, they say Jensen is freaking out because it's such a big market for him, for Nvidia. As a businessman, I mean, rightly so, if you're not thinking about geopolitics, he wants to maintain that sales. He's still trying to push the US government to actually allow him to sell redesigned chips. Nvidia has to redesign various chips for China to basically make it not as advanced as the ones that he sells to US companies. He's saying, "Look, if you don't allow me to sell to China, then, similar to what Ray was saying, someone domestically will fill in that space, and one day, we'll realise that we lost that market share as well." So, it's not a good strategy for even a US company either. I think right now everyone's just watching and seeing what happens because we also see the trade war playing out in front of us right now.

Funding Challenges for Chinese Tech Companies

Keith 37:43

Outside of the fight for chips, there is another interesting aspect of US-China tensions that might deter or might spur innovation, depending on how you see it, which is access to funding. In the past, if you look at Chinese internet companies, many of them in their early years had access to USD funds. They were able to build or get a subsidiary outside of China and get funded by American companies. In recent years, especially, I think even as recent as this year, there's been a clampdown on US VCs trying to fund Chinese companies because they don't want to enrich the competition. The most recent example I think about is MiniMax AI's funding by Benchmark, which I think is getting investigated by the US authorities. In that context, how are Chinese tech companies, specifically, or even Chinese AI companies, trying to gain access to international capital?

Grace Shao 38:51

Going back to the internet era, if you think about the Alibabas [阿里巴巴] and Baidus [百度] or ByteDance [字节跳动], whatnot, their initial pot of gold—many of them are not Chinese; many of their investors are actually Americans or from Western nations. But things are tense. I think partially it's geopolitical, but partially it's now that AI, semiconductor, and robotic industries have been categorised as sensitive industries. It's a two-way thing where, frankly, there are restrictions for the US government where US funds can no longer invest in Chinese semiconductor-AI-robotics companies. But there's also sometimes a reluctance from the Chinese companies to actually accept US capital.

I think, how do I put it? Back then, they were all super rampant in China. They backed a bunch of consumer companies, a bunch of internet companies, but they did withdraw much of their presence within China. And that was before even this AI thing blew up; it was pre-COVID, even around COVID times. I think the geopolitical narrative has not helped for many of these companies. From a PR perspective, frankly, it's easier to just not make the money than make the money and be put in the front and centre, under the limelight and scrutiny by the US government or the public.

Now, on the Benchmark deal, it's quite interesting. I was quite shocked that MiniMax [面壁智能] was able to obtain and receive a huge investment from Benchmark. But I think, when I spoke to people in the industry, they were saying that MiniMax [面壁智能] actually did still separate their China and non-China business, their entities. That was one of the reasons why Benchmark was allowed to invest into MiniMax [面壁智能]. Will it get worse? I'm not sure. But right now, I think from a pure capital perspective, investors want to get into the space because they know that there's innovation happening in China. Definitely, there's scrutiny coming from the US government, so there's fear of an obvious regulation to not allow them to do that.

Now, on the other end of the debate or conversation, what I brought up earlier, sometimes they don't want US investment because, for example, companies like Unitree [宇树科技] or DeepSeek , they've really become national heroes, national champions. And I'm not saying they're nationalistic, but I think they're quite patriotic. In many ways, they're not actually lacking capital. And in many ways, they understand that they're in a sector that is quite sensitive. So, it might even just be easier for them, again, from a PR perspective, easier for them to just take Chinese capital, be the Chinese hero, and continue to progress their innovation.

Keith 41:55

In the past, the way VCs were set up—with so much money or a huge part of their LPs coming from city or state governments—made it very hard for them to properly function as VCs. With that being said, how are investors in China looking at investing in AI companies?

Grace Shao 42:16

I think people are still definitely looking for fundraising, depending on how you want to position your company. For the ones who do sell within China, whose main product is within the domestic market, I think they're more open to VCs that are backed by states, whether it's municipal government or federal government level agencies. And they will work with local governments oftentimes, such as Hangzhou [杭州], which is extremely supportive of AI development right now. There are many different sponsorship programmes and investment opportunities, matchmaking happening within the government level. So, companies might be able to do that.

But for some companies I have met, they are Chinese, as in the researchers are Chinese, their headquarters are still based in China, but they actually sell to the US or even Singapore. Their main market is not China. Then, maybe they would actually be looking for foreign or Western investors to help them. That's not just a capital access thing; I think it's also about connection, reach, and credibility in the Western markets they're trying to reach.

The Evolution of Tech Entrepreneurs in China

Keith 43:23

I want to quickly revisit a point that you alluded to earlier, which is on the talent front: that the culture has evolved over the years, especially from the early internet era in China to now, which is primarily AI-driven, for AI entrepreneurs. How has that shifted?

Grace Shao 43:49

One point I mentioned earlier is that this generation actually grew up with much more abundance in terms of day-to-day necessities, and because of that, I think they had the privilege to dream. You can think of them as more similar to anyone in the US—think of Evan Spiegel or Mark Zuckerberg—when they came out as young entrepreneurs. Much of their talk was very much mission-driven or vision-driven versus purely money or business-making driven. I think you can see that in many of the entrepreneurs in this generation.

Again, I like to bring up Unitree [宇树科技] and DeepSeek , specifically Liang Wenfeng [梁文峰], Unitree's founder, and Wang Xinxing [王兴兴], DeepSeek/MiniMax's founder. For example, Liang Wenfeng [梁文峰] has openly talked about that for him, embracing open source was really a personal philosophical choice. He really wanted to do two things: one is to allow Chinese tech players, industry players, or innovators to have access to the most frontier technologies that they could possibly have. And to do so by doing open source was one of the best things you could do for the ecosystem. Number two, he was saying that he wants to attract and retain the best talent because many of these researchers are even effective altruists, or they are really mission-driven. They're not purely money-driven; they want to work for the best company. We even see this in the US: many people leaving OpenAI to join Anthropic or other competitors because they don't really believe in how OpenAI has shifted from an NGO to what they are now. I think that's why he really embraced open source. You can see how much of his work is very much mission-driven.

Another overarching theme that I've pieced together when speaking to different founders is that for decades, as you said, the West has blamed China or pointed fingers, saying, "You're such a copycat," or "You copy everything you do." Embracing open source and this mission-driven ideology has been the best way to showcase, "Hey, look, literally, you can see everything I've done; everything's open source. You can't even question if I'm copying or not; I'm genuinely innovating." It's a soft power flex, or trying to address the chip on their shoulder, whatever way you want to put it. I think this is a very big difference from, I'd say, the generation of entrepreneurs born in the '60s and '70s.

Keith 46:25

With regards to the technical development pipeline, I guess, if you look at these founders, obviously they need to have a very strong STEM background outside of just being motivated by a mission of sorts.

Grace Shao 46:40

I think fundamentally, even generations ago, people have always said 学好数理化，走遍天下都不怕.

Essentially, back then it was physics, chemistry, and there's that saying in Mandarin where it's like, "If you learn math, physics, and chemistry well, you can do anything in the world."

There's definitely that framework by even just the average parent in China—you think about people pushing their kids to be engineers and doctors. That's the typical path.

That cultural aspect and joke aside, I think institutions have also been really overlooked. When DeepSeek came out, one of the biggest shockers for the West was that everyone was actually Chinese-trained. I live in Hong Kong now, and I meet many Chinese graduates in my day-to-day work, even at these big companies and everything, and no one would ever question them in Hong Kong that they would be subpar. Everyone knows they're amazing talent coming out of Peking University [北京大学], Tsinghua University [清华大学], Zhejiang University [浙江大学], Renmin University [中国人民大学]. But I think they just weren't as relevant or prominent and understood by the West.

If you look at Liang Wenfeng [梁文峰], he's educated at Zhejiang University [浙江大学] in Hangzhou [杭州]. Many of these talents, Wang Xinxing [王兴兴] himself as well, from Unitree [宇树科技], also educated domestically; none of them went to study in the US. This is again a bit of a two-factor thing. One is a personal choice. I think with the rise of geopolitical tensions in the last decade, many families are choosing not to send their kids to study abroad anymore because they are in fear of even just headline stuff about racism or xenophobia. That's a legitimate fear that many parents have talked about. And then there's just actually, because of sensitivity, if you are studying certain industries such as AI, or machine learning, or certain mechanical engineering routes that are perhaps more related to weapon design, or even aeroplane design, you don't really get visas these days. That's another thing I've been hearing on the ground. All of these reasons have made many strong talents stay domestically. And domestically, obviously, there have been really great institutions to help facilitate this kind of STEM education and talent development.

The Role of Education in Shaping Tech Talent

Keith 49:20

Just as you were talking about the different founders, it reminds me that Zhang Yiming [张一鸣] is from Nankai [南开大学], and then I think Lei Jun [雷军], who founded Xiaomi [小米], is from Wuhan [武汉大学]. It seems to me that the talent is much more dispersed than in the US, where you think about all the tech founders; they all come from either Stanford or Harvard, or they're very concentrated in a certain geography. Whereas in China, it seems to be much more diffused in the way technical founders or technical talent emerge. I don't know how that spells out in terms of its business implications, i.e., where do tech startups choose to congregate or aggregate? Do you have an insight on that?

Grace Shao 50:00

I think in some ways it's as you said, but I think in China, for sure, a couple of schools you just mentioned are also still top 10, top 20 in this country. Many of them, for example, Tsinghua [清华大学], Peking University [北京大学], Renmin University [中国人民大学], Zhejiang University [浙江大学], they also rank top 50 in the world. They are really strong. I think people in the West might have just not known about it, but it doesn't mean that they haven't existed for literally more than 100 years. Just like in Singapore, you have National University of Singapore, Nanyang Technological University [南洋理工大学]. You have many great institutions. People just don't really know about them in the West. For educational purposes, I think maybe these Chinese schools are just having a moment. They can finally shine and feel that their entrepreneurs are coming out and speaking on behalf of them and representing them on a global scale.

In terms of location, there definitely still is a breakdown in where startups are congregating. One is traditionally the Shenzhen [深圳] area, what is called the GBA [粤港澳大湾区], the Greater Bay Area. That connects Hong Kong [香港] to Guangdong [广东], Shenzhen [深圳], everything. You can think of many of the last generation of entrepreneurs being based there, for example, Huawei [华为] is out there, Tencent [腾讯] is out there, DJI [大疆创新], SenseTime [商汤科技]—the list goes on. There was support from the local government to bring in talent, and there's organic talent attraction in that region. I think even when Jensen Huang [黄仁勋] visited HKUST last November for a fireside chat, he was saying that the GBA [粤港澳大湾区] has such an advantage in having access to talent that understand software and hardware integration. He was alluding to physical AI development, actually. That definitely is where much talent wanted to be based because capital access to Hong Kong meaning international capital, was also close and nearby. For context, Shenzhen to Hong Kong is less than an hour train ride. That's really easy.

Now, much talent, or now the current hot buzz, tech companies or AI companies, are actually based in Hangzhou . That's where DeepSeek is, Unitree [宇树科技] is. And for context, the big players there are Geely [吉利汽车], the big car manufacturer, OEM, and they own various EV companies, EV brands as well, as well as NetEase [网易], a gaming company, and a few others. I think in that region, there's extremely top-calibre talent. They have very good academic institutions. The city of Hangzhou [杭州] and the whole province has been very supportive of entrepreneurship. Their geographic location advantage is that they're really close to the textile manufacturing hub of the world. That whole region is where they produce all the clothes and toys and bags you think of in the world, as well as actually Shanghai. It's about a two-hour drive to Shanghai. You have some of the most savvy consumers in the world, have access to international capital and talent and resources and R&D. That region is now really coming through, and they're really incubating more AI startups.

Keith 53:28

Are there any regions outside of this "golden triangle," as they call it, that people should be paying attention to?

Grace Shao 53:36

I was just doing research on industrial robots and AI robots. I was quite shocked because in Dongbei [东北], northern China, people always think it was a big agricultural or industrial area, but because of this industrial heritage, many of the robotics design and industrial robotics companies are up there. Now that area is actually starting to shift some focus and leveraging their existing know-how and resources to really double down on physical AI. I think that would be really interesting to see as well.

Keith 54:45

They say demography is destiny. In a certain sense, when you have an ageing population like China, you have a much more rapidly urbanising population; you're really driving up tech adoption in what used to be your traditional low-cost, labour-intensive industries because now you're essentially switching from low-cost labour to low-cost technology. I think that's been one of the huge drivers. Do you see any other drivers in terms of tech adoption across the board, even in places as far as Gansu where traditional industries are starting to be more open to tech?

Grace Shao 55:21

In general, the culture has been very open-minded to technology. Even as far out as in Gansu [甘肃] or the rural areas, everyone is on their smartphones and adopting digital payment, and this has been really rampant. I would say across all industries, you're seeing companies trying to figure out how AI or advanced technology can speed up efficiency, productivity, and, as you said, create more cost-advantaged products.

China's Tech Role in Southeast Asia

Keith 55:51

I'd like to quickly turn our eyes outside of China. I'm in Singapore, we're in Southeast Asia, where China has a huge economic influence as well. Where do you see the future of Chinese tech in the region?

Grace Shao 56:04

Southeast Asia, especially Singapore, has always played a really big role in Asia and connecting China and the West, in some ways being the neutral ground for China and the US. For example, even in the internet era, you can think of Netflix, Google's Asia headquarters all being in Singapore. But equally, you also have Tencent [腾讯] and ByteDance [字节跳动] and Lazada [来赞达], which is owned by Alibaba all based in Singapore. Singapore has a really vibrant tech ecosystem in terms of talent, in terms of its strategic importance for these companies. It's almost seen as the gateway to enter all the Southeast Asian countries and markets. I think sometimes in the West, people forget that Southeast Asia is huge and it's so diverse. Each market has a different language, each market has a different currency, each market is a different actual country, and they see it as a whole. But Singapore is a soft landing spot for all these companies to first land. For Chinese companies, many talents also use Mandarin. For Western companies, it's easy to land because everyone speaks English. On a language and efficiency level, it's very easy to first set up corporate headquarters there. Singapore will continue to play that role for AI company development as well.

And I think on a startup level, I've seen many companies—if they're Chinese but they actually sell predominantly not to the Chinese market—many of the founders would actually move to Singapore or sometimes set up a very strong sales team or R&D team in Singapore, but then it's facing the rest of the world and not China. In other ways, I think we're seeing that ByteDance [字节跳动] and Tencent [腾讯], and obviously all the big tech with the Magnificent Seven in the US, they're all also building up data centre capacity rampantly across Malaysia,Thailand and even in parts of Singapore. But again, Singapore being the hub for much of the management, the execution is all done through Singapore to the region. Singapore continues to play a very big role in that sense, and obviously, it's still a very big financial hub as well. I think as much as Hong Kong [香港] is seen as the capital market centre and obviously the gateway to China and outwards, Singapore now is strategically positioned to be a neutral financial centre for the region and the rest of the world. I know many European investment institutions that actually all set up and are continuing to build up shop in Singapore.

Keith 58:25

It seems to me that Chinese companies really do learn in Singapore to internationalise. I think that's where they really bifurcate their Chinese operations and their international operations, and usually, it's done in Singapore. My final question for you is that if you have a piece of advice that you have to give to a fresh graduate entering into the working world, what would that be?

Advice for Fresh Graduates

Grace Shao 58:43

My humble opinion is just to get out there and try everything. I'm someone some might say is a bit jumpy, but I've really done everything. I've interned at a hedge fund that people don't know about. I actually started my career as an equity analyst covering the TMT sector in Asia. I've interned at an accounting firm. I've worked at a PR firm. I've worked in an ad agency. Obviously, the majority of my career was in journalism and broadcast and writing and audio digital media. I've worked in corporate strategy and PR. I really believe when you're young, you've got to try because I just listened to this podcast recently, Patrick O'Shaughnessy's "Invest Like the Best" and "Founders Podcast". I really liked one thing he said: he was saying, "You don't pick the passion, the passion picks you." And until you try things out, until you go through trials and errors, you don't really know what you're good at, what you're bad at, what you sometimes are good at but you don't want to do for a living. But sometimes you need to find a thing that you're good at, but you actually want to do every day relentlessly.

For me, after doing all these different things, I'm finally bringing everything together, and as you said, I run my own newsletter, and I can bring in my experience as a journalist, I can bring in my experience as an analyst, and I write deep dives. I obviously enjoy speaking to different kinds of people, listening to them, and learning from them, and that obviously comes from having experience interviewing and engaging with different kinds of people from all across the world. And then I think the accounting and finance background, even the creative background, has allowed me to really lean into writing in more depth in financial terms. That might be a bit more than someone who's completely from a liberal arts background. Versus my experience in broadcast and writing has allowed me to perhaps write in a more creative way than someone that's completely from a financial background.

I think you have to do a bit of everything to find yourself, and we're so lucky in this time and age that careers aren't just seen as so linear anymore. You can be hyphenates, you can be whatever, you can be founders, you can be anything you want, and society allows you to do that. It's very different from, I think, the last generation where you're expected to join one company and be there for the next 30 years of your life and just retire and chill out. You really have that flexibility. And even yourself, you've done a few transitions too when we talked offline. I think it's really great that now you're finding your passion and being that knowledge gap provider or bridger on your podcast. I really encourage people to just say yes to everything when you're young, especially when you don't have mouths to feed and you have other responsibilities.

Keith 1:01:34

With that, Grace, thank you so much for coming on.

Grace Shao 1:01:37

Thank you so much, Keith, for having me.