"India and Southeast Asia are still struggling because we have different languages. English is not the same as Thai, Vietnamese, or Filipino. It's disaggregated—different languages, disaggregated materials, disaggregated market sizes and applications, and disaggregated GDP per capita. This makes it very hard to train AI every day. Chinese AI is being trained by a billion plus people in China, and Americans, all 300 million of them, are training the American AI along with Western-educated folks. So it's actually hard to build a pure play AI company out of Singapore structurally."  - Jeremy Au, Host of BRAVE Southeast Asia Tech Podcast


"Private equity versus venture capital, venture capital grew out of the private equity class. When you think about it, there's public equities, there's private equity, and private equity is private vehicles funding private companies. Venture capital is a specialized subset of private equity. From a media perspective, coverage tends to focus on venture capital because private equity buys stable, mature businesses that have already been built, whereas venture capital is more exciting to write about. You have heroic founders who go out there telling you that everybody's going to get married to AI soon. Don't worry about it, enjoy it, it's good for you. There are also many spicy startup failure stories for 19 out of 20 startups, which are far more interesting compared to some private equity fund buying Toys R Us and maximizing profitability from it. I think there's a different media exposure component to it." - Jeremy Au, Host of BRAVE Southeast Asia Tech Podcast


"So in America, there was a huge surge called quick commerce, which asked if you could get stuff delivered within two hours or one hour. In America, most of quick commerce fizzled out because they got killed by Amazon and other giants that had enough scale and service delivery to do it. The craze for quick commerce started there, then spread quickly to Europe, and now to Indonesia and India. In India, quick commerce managed to succeed to some extent by slowing down. Instead of delivering in two hours, they shifted to eight hours or next day, offering a cheaper price if it took longer. Quick commerce eventually became commerce champions in India."  - Jeremy Au, Host of BRAVE Southeast Asia Tech Podcast

Jeremy Au explored how talent, policy, and capital flows shape startup ecosystems across Southeast Asia, India, and China. The discussion covered talent strengths and weaknesses across countries, the role of industrial policy and government subsidies, the challenges of building large language models outside the US and China, and the impact of US China geopolitical tensions on venture capital flows.

01:18 Southeast Asia talent rankings: Jeremy explains Asia Partners’ view that Singapore and Indonesia rank higher in talent, while Vietnam, Philippines, Malaysia sit in the middle, and Thailand lags.

04:01 Quick commerce lessons: He compares how India’s delivery startups slowed timelines to survive, showing why founders must adapt to market development levels.

05:10 TSMC and industrial policy: Jeremy highlights Taiwan’s deliberate investment in semiconductors as a case study of state-driven strategy creating global champions.

07:30 Policy shapes business: He outlines the chain: “Policy leads to law, law leads to business, and business leads to everything else.”

15:08 Sequoia’s split: Jeremy explains how US–China tensions and new rules like reverse CFIUS forced Sequoia to separate its US, China, and Southeast Asia/India arms.

Jeremy Au (00:00)

The question is, why does Asia Partners say that Indonesia and Singapore is high talent, fewer for Thailand and somewhere in the middle for Vietnam, Philippines and Malaysia? First of all, obviously, this is an Asia Partners methodology and it's a qualitative assessment. I would say that in general,

makes sense in some pieces. would say, for example, I would say that Singapore obviously has very high technical talent. I think nobody disagrees that it is very high. And I think that nobody disagrees that Vietnam has actually a pretty good education system, is training lot of engineers. So actually you see a lot of Chinese companies obviously setting up factories and training talent in Vietnam. You see a lot of Singapore companies moving their engineering hubs into Vietnam as well. Because Vietnam is scared of their big brother China. So they have a little brother strategy, right? They focus on industrialization.

They're focusing on economic growth to be able to keep their country going quickly, right? So they're also learning from the Chinese model of, you know, communists focused on sciences and hard engineering. So I think there's a lot of technical talent in Vietnam. I think obviously English competency is a difficult component for them right now. Everything in between is a little bit contentious, I would say. I think that the Philippines talent, obviously there's a BPO and virtual assistant and core center talent. But I think everybody would agree that Philippines has a big brain drain.

problem where the Filipino talent is migrating to Singapore, the US or Europe. So I think the Philippines will acknowledge it as big problem for them. So I think where they stack up is pretty difficult. Indonesia as well, think it's large population, lots of actually entrepreneurial engineering talent as well. So I think there's a little bit of a debate. I was just meeting with, you know, kind of like large language model, AI side, and, know, from their perspective, they're looking at AI founders. From their perspective would be.

American and Chinese nationals are the biggest adopters of AI in terms of startup. they think that Singapore founders are decent as well. They think that AI adoption as a technical infrastructure is lagging in kind of like Thailand, Malaysia, Philippines, Vietnam. So I think it depends on what you want to define a talent piece. Singapore is only Qatar in terms of GDP per capita, right? So it doesn't have to deal with those next two layers, right? Which is, think, roughly corresponding to rural and tier

three cities. I Singapore is a tier one city by itself, right? So I Singapore policy is relatively straightforward in terms of industrial policies, focusing on biotech, shipping, aviation, nuclear engineering soon, a very high end manufacturing, pharma, pseudocalls, right? So I think that Indonesia and India, for example, have some similarities, but I think during the quick commerce wars that happened about four or five years ago. So in America, there was this huge search called quick commerce, which was, can I get stuff delivered within two hours, you one hour?

my stuff. And in America, most of quick commerce actually fizzled out because they got killed by Amazon and these other giants that actually had enough scale and enough service delivery to be able to do it. you must remember that the craze for quick commerce really started there. And then it spread quickly to Europe for quick commerce. And then now it's spread to Indonesia and to India. And think in India, your quick commerce has managed to succeed to some extent, but they succeeded by slowing down.

So instead of you two hours, suddenly they became, let's deliver to you in eight hours. Then next day, and you know, we'll give you a cheaper price if it takes longer to deliver. So basically what happened was that quick commerce eventually became commerce champions in India. They became less advanced in terms of it. So I think the short answer is, I think that as a founder, you have to be aware what level of development your country is, what kind of company you want to build.

And you can maybe build a more advanced version if you get government support, right? So for example, TSMC was a semiconductor play. Before that, Singapore used to be the semiconductor champion of Southeast Asia and Asia. I mean, alongside obviously Korea and Japan, right? But Taiwan did a very deliberate strategy to build a Silicon Shield and invest billions of dollars in energy, in land, in technical talent, in training to create one industry, which is a semiconductor industry, right? And so they were able to inorganically build that and build TSMC.

Of CSMC also had the right idea, the right skills, et cetera, but all of it had to come together. think that technically without, think, I'm just trying to say here is that you can imagine a scenario that if somebody tried to build a CSMC in another country like India or Latin America, or even in Europe, you could have struggled because you have. So I would love to tell you that, I mean, mean, another way saying this is like, if you want to be a comedian, you move to New York or Los Angeles. That's where the talent flywheel is, right? We saw that in the previous slide that we had.

Boston Route 128 used to be Silicon Valley of America, but it lost the crown of technology ecosystem to Silicon Valley. So I think you should be aware of what you're building, right? I think if you go to Silicon Valley today and try to create a, know, aircon repair business, people will just laugh at you. But I can imagine that if you build a, you know, but if you look at Indonesia and you look at Vietnam, you know, and you look at the GDP per capita difference between that and Malaysia and Singapore, you see that.

We all live in the equator. It's super hot. I think everybody's going to buy an air conditioner if they can afford the energy and the bill, right? So you could build, technically you could build that, from my perspective, an air conditioning unicorn in Southeast Asia, right? So I think it's also useful to think about it in terms of the country is that if you look at the amount of money that's going into ecosystems, this is a chart of the deal value on a of like month to month basis by Deustry Asia from 2021 to 2025. And it shows three ecosystems. It shows India, Southeast Asia and China in terms of the deal.

components. And what you see here is obviously that China is the highest. It entered a lull during the pandemic era and during the China crackdown on technology entrepreneurs during this time period. Then roughly around 2024, think the reopening of China after the pandemic, but also the reintegration of entrepreneurship and technology in the Chinese ecosystem has led to an increase in funding size deals in China. Orange in India was

and Southeast Asia were both higher roughly in 2021, around this time period, then both have dropped over time to a relatively low quantum that we have here. So I think this is a good way to understand at least the total amount of volume that's going in. And obviously when you think about it, you're like the differences in ecosystem, but I think there's actually quite sizable amount of value and money still going into startups. I think that policy leads to law, right? And law leads to business and business leads to everything else, you know. So

I think there's a very strong kind of like circular component to it. I would say that I think that first of all, the Chinese technology ecosystem is still fundamentally strong because it's the hardware, the big piece. And structurally, the Chinese government has put a lot of money and effort and focus on building a conveyor belt between the R &D, the universities, the production and the export industries. So I think they've created that dynamic where, and today I think the EU and America, for example, would accuse China

of overly subsidizing the solar cell industry. because most people don't know, but in the past, the solar cell champion manufacturer used to be Germany, right? But unfortunately, all the German solar cell champions have all died and now they're all Chinese. But the Chinese have also made solar cells much more cheaper than the Germans ever did, right? Whether it's a function of economy scale, whether it's a function of government subsidies, whether it's a function of talent, whether it's a function of business model, whatever it is.

But today is undisputed that the Chinese solar cell industry is the number one producer in the world, right? And the reason why that's important for us to think about is that we always need to think about it from the fundamental startup side versus the funding side. And this chart shows the funding from a private sector investment. doesn't show government funding into the ecosystem. Okay. So I think if we leveled in Chinese or other subsidies into it, I'm pretty sure that

The China chart would be much higher from my perspective. I don't think that the India ecosystem will say that the Indian government does a lot of subsidies for the Indian system. think in Southeast Asia, I think some countries, I think primarily Singapore is pushing and Vietnam is pushing ⁓ the technology ecosystem. So I just want to kind of like put a big caveat there ⁓ to be thoughtful about that. It's not just private capital is important, but also government support and capital as well.

I think that private equity versus venture capital, venture capital grew out of the private equity class. Okay, so I'm just going to say this, right? It's like, when you think about it, there's public equities, there's private equity, right? And private equity is private vehicles funding private companies. And venture capital is a specialized subset of private equity, actually. From a media perspective, tends to talk about venture capital. And the reason why is that private equity tends to buy stable, mature businesses that have already been built. Whereas venture capital is so much more exciting to write about. You have these...

heroic founders who go out there telling you that, you know, everybody's going to get married to AI soon. Don't worry about it. You know, enjoy it. You know, it's good for you. And then, and then there's so many spicy startup failure stories for 19 out of 20 startups. So it's just so much more interesting compared to some private equity fund buying Toys R Us and maximizing profitability from it, right? I think there's a different media exposure component to it. Yeah. Why did I choose budget capital and high growth startups?

I I love technology. I've always enjoyed science fiction. I happen to study in California and in Boston, right? Which are both hubs for technology innovation. So I'm a very good bridge between the Southeast Asia ecosystem, as well as the American technology and capital markets, right? They understand me. They come to me to understand the Southeast Asia market. And vice versa, I can work with Southeast Asia founders and explain and discuss in a very Americanized way to Western train capital markets as well.

That background could also make me suitable for multinational corporations, you can imagine. But obviously I prefer startups that are hiding the higher growth. There's more risk, there's more reward. I think that if you look at Singapore, again, it goes back to what other industries are doing well in Singapore, right? So if you think about it, Singapore as a government is focused on what? Biotech, logistics, aviation, events, you know, but also a stack of other things, right? Pharma and so forth. Fintech, obviously. So obviously there are clear verticals that Singapore is good at.

and technology is able to be supported by the Singapore ecosystem, not just in terms of capital, but also in terms of talent and networks and understanding. So if you think about it from an AI perspective, when you say AI, actually, if you think about it, are no large language model companies in Southeast Asia. So what do you need in order to be a large language model? First of all, you need a very deep capital pool. That's one. And that's either available in America or China, I think, to some extent. One. Two is you need a very wide data set. Obviously, the American script.

the entire internet, including, I don't know if any of you use blog posts or Twitter or Reddit or all that stuff got scraped and was aggregated into a large language model. And then for the Chinese ecosystem, most of it was not written on the public internet because the Chinese internet came into existence later and it was within walled gardens, right? So within WeChat, et cetera. So most of the training materials was in people's WeChat messages. And so that gave them the training data, you can imagine, and the Chinese language was big enough.

with, you know, over 1 billion people teaching the AI every day, specifically in Mandarin, how to use the AI, right? So what you see is that Southeast Asia and India are struggling to build large language models, right? I think there's also one in Europe as well. All I'm just trying to say here is that the Europeans are saying we're doing this because we believe in digital sovereignty. We're never going to buy a Chinese AI and we're never going to buy American AI. So they're willing to subsidize and support the European AI piece, right? And I think there's implicit understanding that

Once all these language models become fully scaled, you can imagine regulators in Europe may not be favorable to a Chinese American AI working on European applications. You can imagine, right? This is how we don't really believe that American AI will be deployed in China. Neither do we really believe that Chinese AI will be deployed super widely in America to some extent, although the Chinese have figured out that they can go open source. And so it's not Chinese if you donate it to the world, then everybody's using open...

source Chinese models. India and Southeast Asia are still struggling, right? Because first of all, as Lai Ping asked, we have different languages. So English is not the same as Thai versus Vietnamese versus Filipino. It's disaggregated, different languages, the materials are also disaggregated, the market sizes and applications are also disaggregated, the GDP per capita is also disaggregated. So it makes it very hard to train the AI every day. Because right now,

Chinese AI is being trained by a billion plus people in China right now, right? And then Americans, all 300 million people are training the American AI and all the Western educated folks are also training the American AI. So, so it's actually hard to build a pure play AI company out of Singapore structurally. So maybe another way of saying this is you could ask any one of us and be like, why isn't anybody in Singapore building large language models? And answer is structurally, it seems to be more difficult. I think there's a few people building it, but it also reinforces my point.

large language model for South Asia is being built with Singapore government support, which is called Sea Lion, I believe. But there's no private capital going to it. So I think if you look at the public news of Sequoia, they felt like they had a split because of the US-China geopolitical tensions, right? So Sequoia did make money in both the US startups as well as in Chinese startups. They both were doing well. And Sequoia was a name brand that was a US dollar denominated. And so it was an ally of the

Chinese tech ecosystem in the early days. But depending on who you say, whether it's the Chinese or the Americans, but the crux of it is there's a divorce, you know? And that divorce basically means that many American limited partners and also we'll say the US government under the first Trump administration, as well as the Biden administration now views American capital going into Chinese technology companies as a national security risk. So they don't want American capital to go into China.

And there's an interesting rule called CFIUS. So it is an investigating committee that allows America to veto investment deals by China into America. So that's one. But they have now implemented under the last year, the Biden administration, which Trump has allowed to execute, it's called the reverse CFIUS, the reverse CIFUS. But basically what that rules means, if an American limited partner on American fund invests in a Chinese

national led or Chinese startup, it must be declared to the US government and the US government holds the right to veto. In other words, American funds will find it very hard to comply to invest in Chinese startups in the knowledge that any time their transactions can be unwound. So effectively that caused that bifurcation into Chinese-only funds versus American-only funds. LPs are not able to, under this new regulatory regime,

may not be comfortable taking on the geopolitical risk. And also the general partners of those funds may not feel comfortable with that risk. I think Sequoia Chinese partners decided that it's better split from the American side and then they went their separate ways. And they both agreed that Southeast Asia and India can go a separate way as well. So I think that at the end of the day, I think that partners and teams all work in their teams, right? So I think the Chinese team felt comfortable with the single market.

ecosystem. I Americans feel comfortable with American and the European deals. And then I think they felt like Southeast Asia was an independent ecosystem with its own different set of return profile for Southeast Asia and India. Now, the last slide here is talking about Southeast Asia. But one thing to note is that Southeast Asia VC funds, this is a chart showing from 2018 to 2025. But this shows, and from Dew Street Asia, the amount of capital that Southeast Asia VC funds have been able to successfully raise. And this, what this clearly shows is that

There's a hump between 2018 to 2019, right? Which was also driven by the Singapore ecosystem and the government support to drive increased venture capital flows. There's a dip in the 2020 piece because of pandemics, so everybody didn't know what to do. And so people didn't stop writing checks into VC funds. And then during the zero interest rate era, they got accelerated by the kind of like Trump and Biden administrations. A lot of money flowed out from America and a lot of capital went into the US technology ecosystem.

zero interest rates also went into the Southeast Asia funds ecosystem between 2021 to 2023. And then since then, it has dropped quite a bit because now there are very high interest rates to control for inflation that happened after the pandemic, fly chain, tariffs, the Inflation Reduction Act by Biden. was a huge stimulus in America, whatever it is, but very high interest rates have caused a sharp dip in this. This is important to know because

The capital that is raised by all these funds, they normally have a mandate to deploy that within two to three years. So the capital that was raised in 2022 are supposed to be finished deployed by 2025 effectively. So the fact that we have very little capital being raised in 2024 to 2025 indicates that Southeast Asia deployments of capital will probably be low between 2026 to 2027. But there's some cyclical component of it. this is a leading indicator that

We will probably see more bad news or we'll continue seeing a start-up winter in Southeast Asia for few more years.