The U.S. - China Deep Tech Arms Race

Vivek Tuakuri: The risk of being reductive. It's really a battle for bottlenecks. And the US Principal bottleneck is energy and China's principal bottleneck is chips. We're still in a very strong position. We shouldn't take it for granted. We should certainly not do any harm to our innovation and talent ecosystem because China is paying attention to this, investing in this. Anyone who just imagines that China just steals and scales, they don't innovate like Deep Seq is a great rebuttal to that point. Right? Scarcity is the mother of invention. They didn't have compute, so they advanced. True, impressive and algorithmic efficient.

Michael Marks: Hi everyone. This is the TechSurge deep tech podcast presented by Celesta Capital. Each episode we spotlight issues and voices at the intersection of emerging technologies, company building and venture investment. I'm Michael Marks, founding managing partner at Celeste. Uh, if you enjoy Tech Surge, subscribe and leave us a review on your favorite podcast platform. Visit techsurgepodcast.com to sign up for our new newsletter and find our video episodes on YouTube. The US China Technology Competition is often described as a race. What is actually happening is a lot more complex. It is a multi front battle over chips, software, talent, infrastructure and global alliances. And the United States, despite holding the strongest hand of any player in this competition, is not always playing it well. Our, uh, guest today is Vivek Tuakuri, who has spent years inside the State Department and the Senate working at the intersection of technology and national security. He now leads research on US China competition at the center for New American Security. In this conversation, he helps us understand this important relationship where America leads, where China is closing the gap faster than we realize, and what is at risk if the US takes its advantages for granted. Well, Vivek, welcome, uh, to the show. Pretty excited about this one. This is, uh, an area that lots of people are interested in. So this is going to be fun. For the last 30 years or so, let's say before the last couple of years, China's dominance of manufacturing is what everybody talked about. And now nobody talks about that. Now what everybody talks about is that is the technology company competition between the US And China. When did this start happening? When did we switch from sort of mundane discussions about, you know, population sizes and manufacturing prowess and start talking about tech dominance?

Vivek Tuakuri: I think there was a couple sort of wake up calls, uh, and I would argue that it really happened towards the beginning of the first Trump administration, maybe towards the end of the Obama administration. But I think there were a couple sort of wake up moments. For the United States. One, we realized that we didn't make chips anymore. We were deeply dependent on Taiwan for chips, which brought China into the dynamic for obvious reasons because she has instructed the PLA to allegedly be ready to reunify Taiwan with mainland China by 2027. So we realized that whether we liked it or not, China was going to have a say in our technological innovation with respect to chips. And also just with, you know, chip manufacturing, we went from 40% of a global share to 10%, you know, in a generation. And China has taken up a lot of that with, with SMIC and with legacy chip manufacturers. That's one story. The other one is we woke up one morning and realized that we had lost the transition to 4G and 5G networks in a lot of emerging markets all over the world. And the US didn't really have a player. And I think we recognize that this wasn't just about market share. This was about espionage. This is about data security. This is about the ability of Chinese companies to, to reinvest that revenue back home in innovation, uh, and double down on sort of this, this self reinforcing cycle. And I think, you know, as these technologies have grown more sophisticated, their dual use nature and national security implications have become clearer, which is inevitably brought in the national security community in a different way. I think, you know, the combination of those things, uh, and of course AI has accelerated all of this significantly. We can talk more about that later. But I would say the, the early canaries in the coal mine were chips and telecommunications.

Michael Marks: Okay, well we'll talk a little bit about uh, tsmc, which I, uh, we can all agree is one of the most important companies in the world today. But let me do a little separation here. So we have the AI stack which we're going to spend a bunch of time talking about, but there's the hardware stack. One of the things we talk about at our firm is that, is that AI has caused a remake of the entire hardware stack. So that's, you know, semiconductors, which you've just referred to, but it's also, you know, energy data centers, you know, all the little chips that make things work. So there's this, and of course the, the major AI chips which we all know about from Nvidia and Broadcom and others. Can you talk to us a little bit about what are the issues between those two? On the one hand it's, you know, software development will dive down on it. The other it's, it's the hardware stack. You talked about tsmc. So these things Collide. But, but how do you see those two issues? Are they just connected? Are they separate? What's the story there?

Vivek Tuakuri: If you're only good at the software side and you don't have trusted players on the hardware side, you're not actually able to meet people where they are. Right. I think one of the things that we're seeing with, that we saw with, just to go back to telecommunications in the Huawei story is that they were able to provide bundled packages of the internal operating systems and software for Huawei that's proprietary to them alongside the antennas and the radios and the phones. And, and, but because they were able to provide innovations in the digital side with scalable, efficient, like genuinely good products on the hardware side. I think Americans in the west comfort themselves sometimes into thinking that China doesn't innovate, they just brute force their way into stuff. Like they are good at producing good technologies and services. And I think we're sort of lulling ourselves into complacency if we, if we believe that. Not to say that they don't subsidize stuff, but I think the short answer to your question, Michael, is that if you're just focusing on a world of bits and not the world of atoms, you are missing the opportunities for adoption, you're missing the opportunities for integration, and you are not able to scale in the way that China has been able to do in a lot of other domains. And I think, you know, even with chips, right, like we, we control, you know, obviously the United States has a huge lead in chip design, right? And the actual architecture of training frontier models and again, the chip design, the design of the chips used to train those models. Um, but a lot of this stuff is happening at the software level, right? And we found that in the physical world we were still dependent on the assembly and packaging for those chips. Abroad, we are finding that, you know, if you're talking about edge AI or the application of, you know, smaller models and less compute intensive, uh, workloads that don't necessarily need to run back to a data center and probably shouldn't because you want a lower latency, uh, service, you need to have the, the hardware for that, right? And you know, there are American and Western companies that are doing that, but China is very good at that. And you know, I think if we neglect that physical dimension of the competition at our peril, and that's even before you get into robotics.

Michael Marks: So yeah, okay, we're going to come back to this. Obviously this is such a core thing because we talk about putting export controls on the hardware and then we talk about how to, how to develop a global, you know, uh, a group of associated countries on the software side. But let's talk about, about the innovation side of China because you know, this wasn't a major part of the debate until the last four or five years. So we had Raphael Rif, a good friend on uh, our show last year, was a former MIT president. One of the things he talked about, if you go back 10 years ago or 15 years ago, the Chinese would say they can outproduce us and I would cause this but, but they'll never be as good as we are at innovation. And he said that's completely flipped around. I mean he looks at the, at the research output, the quality of papers coming out of China, the amount of patents that are being created and they're dwarfing what we're doing in the US today. And this has happened pretty quickly. So how would you characterize the strength of China's innovation system? I mean, is it true? Is this accurate that it's gotten way better lately and if so, why?

Vivek Tuakuri: Yeah, no, I think it's, it's, it's fair to say that it's gotten measurably better. And this isn't just speculation. If you actually look at, you know, I think people in the U.S. uh, often say, well, they're just producing a lot of papers and it's just volume, it's, it's quantity, not quality. And there is some truth to that. But now if you actually look at the number of papers coming out of China that are actually cited and in reputable publications, it's not where the US is, let's be clear about it, in some domains. But, but it's going up, the trajectory is going up. You know, the share of institutions that appear in global rankings in a lot of domains, whether it's life sciences or computer science, mathematics, uh, like is going up. To be clear, the US has formidable advantages. And then when you broaden the aperture to include uh, you know, Korea and Japan and Europe, sort of Team America if you, if the extent there is a team and we can talk about that later, but is, is in a much stronger position but China is producing both volume and quantity equality, uh, uh, of talent is, the trends are definitely going up and I think as I said it would be um, a huge mistake for the United States to think that our innovation ecosystem and our talent ecosystem is sufficient to, to secure a durable long term advantage over China in a lot of these areas. With that said, I think structurally I would still bet on the US innovation and ecosystem. And I think we have, the United States has the best hand of any country in the world to play in a technology competition. Now you can argue whether we were playing that hand. Well, I would argue we're playing it terribly in some respects. But we, you know, we have world class research institutions. Most people all over the world, whether graduating from a university and in, you know, Johannesburg or Delhi or Kuala Lumpur, um, are not waking up in the morning the day after they graduate and say, I want to move to Moscow, I want to move to Tehran, I want to even move to Beijing. Although that's starting to change a little bit. They want to move to the United States and specifically to Silicon Valley to start companies. And that's just an enormous advantage that the United States has and should squander. And I worry we're doing that a little bit. All that's to say we're still in a very strong position. We shouldn't take it for granted. We should certainly not do any harm to our innovation and talent ecosystem. Because China is paying attention to this, investing in this also, you know, with their Thousand Talents program, trying to bring Chinese nationals who've been studying abroad back into China to support, uh, their own ecosystem. And as we've seen, even some of their startup, uh, founders in the AI context, like with Manus AI, you know, they get acquired by Meta and all of a sudden they can't leave. You know, China's also trying to hold people in their jurisdiction and keep that brain drain from happening.

Michael Marks: Okay, so again, lots of things to unpack there. But let's talk a little bit about the China system. I mean, we know that China picks winners and losers, they subsidize, they put a lot of effort behind the certain areas that they think are important. They've been quite successful at that. And that is not the way we do it in the United States. So are there parts of the way they do it, a version of their approach that we should be thinking about changing in order to compete with them, or are we good with the open system that we've got?

Vivek Tuakuri: There are advantages and disadvantages to both sides. And our secret sauce is we have the most efficient, risk tolerant capital market in the world. Like that is an undervalued, not by your audience and definitely not by you, Michael, and your firm, an under appreciated superpower of the United States. There is no country in the world that can deploy the scale of capital that the United States can deploy and to drive innovation. And uh, you know, if you look at any country in Europe, if you look in Asia and including in China, it's just nobody touched, nobody can touch the United States. We're seeing it in sector after sector. I mean none more prominently than AI. How that has been an enduring source of advantage for the United States. And that is a function of our, you know, our rule, uh, of law, of our more free market economy, of tax and regulatory system. I mean there's a culture, of course, that I don't need to lecture your viewers about that you know very well, um, that is very hard to replicate. That's a huge advantage where we have disadvantages. We cannot do long term planning for the life of us. Right. And that is not to say that we need to have, you know, put ourselves in a straitjacket from a policy perspective. But you know, our democratic system at its best should be able to pull different perspectives around the country, manifested in Washington through the representatives, to create a durable policy that can actually endure for 5, 10, 15 years where you know, there's some disagreements at the margin, an ability to adapt to changing conditions and public input, which is actually a really important and comparative advantage compared to a more brittle system, an authoritarian system that can't change as conditions change. Um, but like that's how it should be working right now. We're not doing anything, like, we're not doing any serious, uh, not even short term policies. Right. We're in a world where we can't even keep the government fully funded for months at a time. So that is a massive disadvantage when it comes to sending clear market signals about what are, what, what are the strategic priorities of the country when it comes to procurement, when it comes to regulation, when it comes to like, you know, national investment. So that's a huge problem. Uh, and China's very good at that. They're on their 15th five year plan, very good at that 15th five year plan, uh, where, you know, they have all kinds of metrics that they put out there that they hold themselves accountable to. Uh, they have a tolerance for losses and it's a very, it's not very capital efficient. But you know, for technologies that matter for national security and national power, that don't have market returns, that kind of state direction, uh, can be useful. And the US has done stuff like this before. I mean, uh, you know, I think there's a question of just getting the calibration correctly. So ideally I would take the, the openness and the adaptiveness of the American system with some medium to long term clarity and stability of planning and that would Be a real, a real way to get the United States sprinting here.

Michael Marks: That sounds like something we'd all like to see. Uh, it's a perfect segue into my next question. So that Made in China 2025. Talk about the five year plans. Made in China 2025 was supposed to be China's master plan for techno, you know, technology, self sufficiency. Now we're in 2026. Can you talk us through that initiative where they've made progress, where they've fallen behind?

Vivek Tuakuri: Sure, yeah. So there's a lot of things behind our Made in China initiative. This, you know, started as you mentioned, about a little over 10 years ago. They're trying to do a few things. One, they're trying to diversify their economy and move away from this idea that, you know, China is just the manufacturing hub for cheap toys and disposable consumer electronics and you know, almost reclaim the prestige of Made in China. And the term, you know, diversification, moving up the technology value chain is another huge part of them. And also achieving greater levels of economic security. I mean that's sometimes, sometimes we forget that it's not only the United States that's trying to de risk and in some ways decouple from China to minimize our exposure. China's trying to do the same thing.

Michael Marks: Right.

Vivek Tuakuri: Uh, especially as they've seen us weaponize choke points against them. And so they've listed a bunch of technologies. I won't bore you by going through all of them, but you know, it's everything from EVs to biopharmaceuticals to aviation advanced rail, where they say these are the priority technologies that we want to mature, uh, improve sort of domestic procurement. We want to like have better, more globally competitive technologies that have a larger market share. They have all kinds of metrics that they, you know, basically bring experts from industry, from academia and to actually develop these indicators. And if you look at how they're, and they're quite ambitious in some cases and if you look at how they're doing, it depends on the tech. But if you look at clean energy, for example, or let's look at EVs, for example, they've uh, clearly hit it out of the park. Right? Like, I'm pretty sure one of their goals was to, you know, produce about 3 million EVs, uh, or sell about 3 million EVs. Few years after the, the Made in China initiative started, they're selling 15 million EVs a year. Right. At least. And they're totally crushing it. 90% of their domestic market is satisfied by Chinese EV companies, which is exceeding what they had set out to do. That's probably the area of greatest success, if you look at like chips, you know, the integrated circuits and electronics is area that they focused on to more of a mixed bag. Right. I think with legacy chips, the sort of foundational chips that are in our cars and smart appliances and everything like that, uh, they have dramatically increased their production capacity. I think half of all global new sort of legacy chip capacity over the next three or five years is coming online in China. And there's big worries about this as, uh, a version of China shock alongside the EV story. So they've been very successful there with advanced chips. The stuff you were talking about, tsmc, right. Struggling, making some progress, but the export controls have really bit from the United States, Japan and the Netherlands. So the story's mixed. You know, some areas they've done really well, but the point being is that they, they, they have a clear goal of what they're trying to achieve and metrics to hold themselves to it, and that itself is super valuable.

Michael Marks: So let's talk about, about our response to all of this. I think I know what your answer is going to be, but we have tried to a mix of tools. We have export controls, company sanctions, we have investment programs like the Chip Act. Chips Act. You know, you said we haven't been particularly good at this stuff, but is any of that stuff working or do we have other things we can do there?

Vivek Tuakuri: Yeah, well, I mean, look, I was working in the Senate as a lowly, uh, staffer when we were passing the Chips and Science act. And you know, I had a lot of different names. And I actually remember when somebody from another think tank came by and pitched this idea of we're going to pick five technologies and we're going to spend $50 billion over 10 years to propel American tech leadership. And I was like, that sounds like an awesome idea. It's never going to happen. Uh, you know, maybe we'll think about it another time. And of course, a version of that became the Chips and Science Act. And, and so maybe the lesson there is run with a good idea when you hear it instead of saying no from the outset. But, but anyways, that's a, that's a long preamble to your, to your question, Michael. The Chips and Science act was successful for what it was, even if it wasn't enough. Right? So I guess it depends, like it was insufficient in terms of its scale to do what we needed to do. Right. If the goal was we want to have a tsmc. We want to have the ability to be mostly self reliant on advanced chips, uh, and packaging in with the United States. We don't want China to have a chokehold over our ability to have advanced chips in our military and our economy. Like a reasonable concern, like $40 billion of subsidies was never going to be enough to do that. And subsidies alone are not sufficient. You need to have a broader ecosystem. Uh, you need to have a regulatory apparatus, you need to have the workforce to be able to do what TSMC does. It's not like you can just snap your fingers and put TSMC in Arizona even though they're trying to do that.

Michael Marks: Right.

Vivek Tuakuri: They've built an entire ecosystem in Taiwan around this industry. And so it's really hard, you know, with the resources that the CHIPS program had at the Department of Commerce, they did a pretty good job. Uh, and I think the bigger story here, and I really commend, uh, any of your listeners to read this series that another think tank has put out called Factory Settings, which is, you know, the Institute for Progress is another think tank. They have a substack called Factory Settings where they go through sort of the toil, uh, of having to basically create a $40 billion startup in the US government to you know, reshore advanced chip making in the United States. And I, uh, I won't go on too long about it, but the point is that what that experiment reveals is that you can do big hard things within the government, but it is way, way, way too hard to do them. Right. Yeah, uh, like the bureaucracy, the regulatory structure is not set up to do industrial policy, which doesn't always make sense, but in case like this, it does make sense. It's we've made it really hard to do it well. And that's before you even get to questions about just like building stuff in America. Right. This is just like getting the inner agency in Washington to coordinate. And instead of like, you know, you should be thinking about how do we make sure that Xi Jinping doesn't have a stranglehold over the American, America's military's access to chips. Not is this endangered bat going to, you know, be threatened by one plant in one state. I mean, not to, you know, I love an endangered bat more than the next as much as the next person, but it's just an illustration of like priorities that are not quite, uh, harmonized

Michael Marks: was a small step in the right direction. There's two other steps I want to discuss. The next one is export controls. We've Set some export controls, particularly on especially advanced semiconductors. The question is, is that working? And what has China done to adapt? Because it seems that one of the things they've done to adapt is use what they're good at, which is picking their battles and fighting them, you know, to sort of double down on self reliance. So is this working for us or is it a problem?

Vivek Tuakuri: I think they are working. I think there was a debate that. There's a reasonable debate to have a few years ago before we implemented these export controls, which was, well, right now China is heavily dependent on American chips, mostly Nvidia. That means Nvidia is getting revenue, that they're able to reinvest and maintain our innovation advantage in chip R and D. Okay. And it's sort of inhibiting the incentive, the sting of China to indigenize its own chip design and fabrication capabilities. Fair enough, you could have that argument. However, now that we have rung the bell essentially of cutting China off not only from buying advanced chips, but buying the equipment through ASML to make the chips, I don't think you can credibly argue that China doesn't have every incentive, an existential incentive, arguably, to indigenize its chip making capabilities. Even if we continue to sell them advanced chips, they're still going to do that because they know that even if we loosen export controls now, we could always tighten them up in the future. Right. And so that's one question. Uh, but I do think, like, if you listen to people working within China's AI ecosystem, they will tell you that the compute constraints are one of the biggest challenges that they face. Right. Deep Seeks founders talk about this openly in the press. Now that doesn't mean there are consequences. Right. It has meant foregone revenue for US industry. It has meant retaliatory measures against American tech companies. It has created frictions with allies and partners. And it is definitely, um, accelerating an indigenization of China's chip ecosystem. With that said, the big question, uh, is how long of a lead has this bought us? Right? I think that's a really hard question. I do think it would be naive for Washington or anywhere in the world to assume that China will not figure this out at some point, point, they will not overcome this bottleneck by virtue of their own capital and talent and ability to scale and other illicit means as well. So the question then is, what do you do with the lead that you bought yourself? And to the extent that I have a criticism of the export controls, I do think that, uh, Washington has focused probably too much of its energy on export controls at the expense of any of the run faster side of the equation. I think for bureaucratic and institutional political reasons, it's way easier for Washington to trip China through unilateral export controls that the executive branch can do with its on its own versus actually investing in the pillars of American innovation by doing a long term policy, by doing serious investments in R and D, by doing reforms at a legislative level. That stuff is hard, but that is how you run faster as a country. And that work isn't happening frankly, uh, because it's harder. And so I do think our approach is, even though people, it's easy to give a speech about how we need to do both in practice we're much more weighted on the protect and restrict side. Hey, we shouldn't do that. But I don't think it's balanced.

Michael Marks: Yeah, on the scorecard side, let's say the CHIPS act was a small win. You know, export controls, maybe a bit more compute denial. Maybe a bit more. But you mentioned Deep seq, so let's talk about Deep seq. So without access to, you know, all this incredible, you know, compute technology that we've been restricting, they still managed to come out with a pretty significant product. So totally.

Vivek Tuakuri: Yeah.

Michael Marks: So, so how does that fit in? I mean, so we have compute denial and they all complain about it, but then they developed a pretty, pretty successful LLM without it. So what does that all mean?

Vivek Tuakuri: Completely, I think, Look, I mean, uh, this gets to your question about innovation in China, right? Like I think anyone who just imagines that China just steals and scales, they don't innovate like Deep Seek is a great rebuttal to that point. Right? I mean you could add byd, you could add all kinds of companies to that list by the way. I mean I think it's very dated view. I don't know if anyone is making anymore, but they shouldn't be. Uh, with Deep Seek, scarcity is the mother of invention. They didn't have compute, so they advanced true impressive innovations in algorithmic efficiency that enabled them to have a much more compute efficient model. With that said though, there's a, there's some important caveats here because I think sometimes the response is, well, the chip controls aren't working, we've imposed all these costs on ourselves and China can still produce a pretty good model. That is only part of the story because it's one thing to produce a model, it's another thing to be able to actually provide the underlying compute infrastructure, to deploy the model at scale within your own economy. And so like, if China nationwide is still massively compute constrained, which they are, one, they're not able. It will, it will limit their ability to diffuse AI within the economy. That's one. And two, it dramatically limits their ability to actually go around the world and sell their AI products. Right. I mean, you can have open wave models that the world can just download on their own. Whatever compute infrastructure that they have. We're downloading deep seq here in the United States. Right. Uh, all the time. And. But if you're actually trying to do what American AI companies are currently doing, which is building AI data centers abroad in Europe, in the Gulf, in India, China's not really in a position to do that right now because their own market, it comes first, and they are controlling compute constraints. So there's a global competition dimension of this that I don't. That I think gets lost.

Michael Marks: We're going to come back and talk about that.

Vivek Tuakuri: But yeah, sure, I do think both of those things can be true. Like, the chip controls are encouraging, are spurring Chinese innovation, but they're also slowing, um, what China is able to do. And like, it's still true that like more chips equals more capabilities. And though. And like the other. Just point, Michael, not to beat this, beat this dead horse, but algorithmic efficiencies that China produces are also immediately diffused to the United States. Right. Like, so it's great they're producing stuff, but it's not necessarily like a net advantage for them from an innovation perspective.

Michael Marks: Well, I wanted to circle back to something we just alluded to earlier, which is about the manufacturing side, so we can all admit that China is about the best manufacturer has ever been on this planet. They've been really, really good at it. Is that going to help them catch up on the compute side or is it just altogether, you know, different issues?

Vivek Tuakuri: Uh, no, I mean, I think it. Look, I think once they figure out how to recreate what ASML does and do extreme ultraviolet lithography, they'll be able to scale it. I mean, I think it's really easy to overcomplicate the U.S. china, uh, AI competition. But on some, at the risk of being reductive, it's really a battle for bottlenecks. And the US Principal bottleneck is energy, and China's principal bottleneck is chips. Right.

Michael Marks: One thing that stands out in the AI race is the growing power bottleneck. China generated more than 10,000 terawatt hours of electricity in 2024, more than double the United States industrial power. There is Also estimated to be about 30% cheaper on average in the U.S. by contrast, a, uh, 44 gigawatt power shortfall is estimated for data centers between 2025 and 2028, largely because the grid was never built for this level of demand. It's another instance where centralized long term planning can offer a structural advantage in terms of infrastructure planning because China is

Vivek Tuakuri: adding more energy than we can possibly fathom. Right. I mean, I think 50% of all nuclear power plants, plants in the pipeline globally are going to be built in China. You know, not to say nothing of what they're doing on solar and wind and coal. So that's not their constraint. Their constraint is chips. So I think once they figure out the, the innovation challenge, they'll be able to scale production of chips in a serious way to be able to produce compute clusters that are comparable to what we're building in the United States. Right. So I think that manufacturing ecosystem, the state support the ability to build big stuff quickly at large, large volumes is going to be huge once they, once they have the thing to build.

Michael Marks: That makes a lot of sense. Okay, so, uh, at this detailed levels, some wins, some losses, plenty of stuff to talk about. Let's broaden this out a little bit and talk about the rest of the world. So you've been a big proponent of this American AI, uh, exports program where we're talking about providing alternatives to Chinese AI and other countries and so on. Can you talk about that a bit?

Vivek Tuakuri: Yeah, look, I think this, uh, uh, and you should feel free to interrupt me, Michael, if I'm giving you a really long, you know, answer, probably too long in the Senate and learned how to filibuster. Uh, look, I think that the AI export program comes from a bitter lesson that we've learned, which is that if you look at how China does, uh, tech promotion abroad, they, we talked about Huawei earlier in this conversation with their integrated packages. Right. That's been very effective. You know, we did this whole project on the digital Silk Road and countering the digital Silk Road, which I'm happy to talk about if you're curious. But yep, just one anecdote from that that has stuck with me. As we were in Indonesia, we did field research all over the world and we talked to somebody from Indonesia who basically said China is very good at the easy button. Right. You guys don't have an easy button. He didn't say it, but he was basically implying that we have a very hard button. And I would argue we have like seven buttons that people have to press. But not to butcher the metaphor, but just to make it real. If you're trying to build a high speed rail or telecommunications network in Indonesia, what China will do is they'll say, we're going to come and do your feasibility study, right? We'll pay for it, we'll run it to see if this project makes sense and how it would work. Then we're going to finance it, we're going to build it, we're going to do the technical assistance to maintain it, to operate it, and we're going to provide high level political pressure or support at the leader level, at the ambassador level, at the industry level. We're going to, we're going to be willing to bribe people in the way the US historically hasn't been willing to do. All of that has allowed them to lock up a lot of emerging markets here. And so the AI exports program is an attempt to provide a more integrated, comprehensive package in response to a demand from around the world, which is that, you know, instead of making a deal with one cloud provider, with the chip provider, with the model provider, with the orchestration layer services, the theory is that if we can put this together with some government support, with some political support and oomph, um, and even financing, it would help compete against China. So that's the ideal way that it would work in practice. It's obviously a little more complicated in practice. A lot of companies, uh, American companies are a bit leery of the government's involvement here. One, because the Trump administration is not always the best messenger for this stuff, given what's happened over the past year in other domains and tariffs and NATO and all the other things. And so having the Trump administration as part of your deal isn't always a comparative advantage for some of these companies in some of these markets. That's 1, 2. There's like not clarity yet. And they're working on it. I mean, I appreciate that they're working on it. There's not clarity yet. About like you say, we're putting, we're going to be promoting the whole AI stack. But there are tensions within the stack about what you want to promote. If promoting the stack is really just about selling chips to the rest of the world, you know, it raises questions about are you actually seeding the AI computing infrastructure abroad in a way that could enable competition at other levels of the stack down the road? You know, if you give Indonesia a bunch of chips, for example, and they develop the next killer AI app, are you actually introducing competition for American companies at different levels of the stack? They haven't really resolved these tensions. And then there's questions about, like, does the US Government actually have the people and the resources and the internal coordination to do this in a timely, effective way? Especially when a lot of people have left or been asked to leave the government over the last year? So there's lots of questions on implementation that have yet to be answered.

Michael Marks: Well, it's fascinating. So, uh, the Chinese are out trying to, you know, export their, their technology in ways that they can. We've talked about that. You know, the US is trying to do the same thing, trying to, you know, export our stack. This is a good time to talk about the rest of the world, and I don't want to turn this into an overly political conversation, but you've got European Union, you have Middle east, you have a group of four, and Japan and so on. What are these countries thinking now? Are they trying to play both sides? Are they picking partners? Are they just waiting to see? Uh, it's a complicated question, but what can you tell us about that?

Vivek Tuakuri: Great. It's a great question, Michael. I think it's actually one of the most interesting questions of the moment. Uh, look, even before the second Trump administration, there was anxiety in Brussels, in Delhi, in Tokyo, about dependence on the American tech stack generally. Usually that manifested at the cloud layer, right? So if you look at South Korea, if you look at eu, there's a lot of concern that the US through the Cloud act, which we don't need to get into, unless you're curious, allows basically US law enforcement and intelligence services to access data that is hosted on American cloud providers all over the world. Um, even if it's physically in a data center in France or in, you know, in Kyoto or wherever. So there was already this anxiety the second Trump administration over the past year because of all the stuff that's happened with Greenland and NATO and Ukraine and tariffs, has deeply intensified this anxiety about dependence on the American tech stack. Right. And it's coming at the moment where countries are thinking about how to make the transition to advanced AI, when the US controls a lot of the key levels of the AI stack. And so the result of all these trends coming together is the rise of what we're called, of so called AI sovereignty. Right. Or sovereign AI.

Michael Marks: Right.

Vivek Tuakuri: And we just put out sort, uh, of an index interactive website just this month that tracks about 140 of these projects all over the world and sees how they actually look in practice. Because there is no one definition for sovereign AI and complete AI sovereignty or AI independence, where you're controlling every layer of the stack from within your country is a fantasy. Even the United States doesn't have full AI sovereignty. We import most of our chips from Taiwan, as we discussed earlier. So the answer to your question is they are stuck between not wanting to be left behind in AI, uh, transition without wanting to become more dependent on American tech at the moment when American tech is almost the only game in town for AI. Right. Really tough position for them and they don't have a good answer. We were in France talking to some officials about this a few weeks ago and I think they are stuck between wanting to be a place that leads in AI and drives the investment and job and productivity benefits of AI without deepening their dependence on the United States. And it's very hard to walk both of these lines if you the same time. So. And a lot of the countries you mentioned are also not really interested in throwing themselves into the arms of the Chinese AI ecosystem. Right. In some ways India is more restrictive of Chinese tech than the United States. They've actually banned TikTok for instance. Right. I think they're going to be trying to figure this out over the coming years. I suspect, you know, if the tension in our bilateral relations continues in the way it has for other couple of years, maybe some of, maybe some of their, some of the assumptions we have in Washington about Europe and Delhi not being more open to Chinese tech will change.

Michael Marks: Okay, before we get to talking about the future a little bit and wrapping up, let me just talk a little bit about talent and what that means because it plays into the answer you just gave. So you know, America and you said everybody wants to come to Silicon Valley and all that, but now we've got all this stuff happening, you know, we're, we're not making this, uh, as such an open company, we're canceling student visas, we're making it harder to immigrate. Lots of entrepreneurs or scientists are leaving. So there's a two sided question. One is how bad is that for the United States? And the second is, does that help these other countries where their talent. Take France or whichever country you want to pick, where their talent has been coming to Silicon Valley now maybe it can't or they're leaving, does it give them more of an opportunity to create some value on their own shores?

Vivek Tuakuri: I mean, 100%. I just, I don't see, I don't see how losing smart people to competitors is good for the United States. Right. I, I mean it's a total own go. I mean as I said before, uh, Immigration is an asymmetric advantage that the United States enjoys. Right. People are not banging down the doors to go set up companies in Beijing who were born abroad, who were born outside of China. Right. Even within China, people were trying to go to Singapore, uh, for different reasons. Right. Because of the crackdowns and the regulatory relationship between the state and the private sector over there. So we're definitely losing people. And it's not just an abstract thing. I mean, you have, uh, even close allies and partners, the uk, Canada, Australia, who are, who have visa programs that have essentially been set up to take advantage of our sort of unconscionable H1B backlog. Right. They recognize that this is a, this is a moment of opportunity for them, uh, because hopefully we'll come to our senses and recognize that we need all of this talent. With that said, bringing talent into France or to Australia or to the UK or whatever is important, but it's not sufficient because you still have all the structural challenges in all these countries that, that doesn't magically resolve. You still have challenges of like, you know, regulations, uh, of uh, capital markets, as we discussed, being less risk tolerant of like, challenges of commercialization and scale. It's hard to put, with the exception of France, it's like these are not energy abundant, you know, jurisdictions. If it's, if there's anywhere in the world that might be harder to build energy than the United states, it's the U.K. right. Uh, where everyone has the shire to protect. I say that somebody, you know, adores the English countryside. But, but like the reality is that it's hard to build stuff over there like it is here. So I don't think talent's going to solve it for them. And the money is all still in the United States. Right. I mean our companies are spending $600 billion on AI, uh, capital investments this year alone, probably so. Right. People who want to make money and really be at the frontier of the, of the actual innovation that is in the United States. And I think that will remain an enduring appeal for people over the world.

Michael Marks: Yeah. One other question that I'm going to ask you to predict the future. So I know that you've done work in, we've talked about, about AI hardware and software, but there's also, you know, quantum and space and biotechnology and advanced manufacturing. You admitted a reference to robotics. We do a lot in our firm in uh, in the drug delivery space. China has been a giant player there. You know, all the big pharmaceutical companies around the world are having their products tested in China, where There's less regulation, so on. Um, and China's building massive genomics database, investing in gene editing, all these kinds of things. So I guess I'm just asking a big broad question. At the end of all this, we have kind of a cold war going on between us and China and lots of technologies. Do these other ones matter as much as AI are there? I mean, how do you think about the whole system of technology and competition between the two countries as opposed to just narrowing it down to AI specifically?

Vivek Tuakuri: Yeah, you mentioned a lot of domains of the competition that are super important. I mean, I do think AI is probably the most important. But I would put biotech quite close to AI for the simple reason that I just don't think people have appreciated what is actually happening at the frontier of biotech, which is biology is becoming a discipline of human design, of engineering. Right. And if you just sit with that fact for a second, it is kind of remarkable. Like, you know, this is not just about better drugs. This is about better plants, better fuels, better materials, better defense applications. This is about reorienting supply chains back home, using, presumably, you know, renewable. I don't mean that in like a clean green, although that's great sense. I mean, just like literally biological food stocks and sources that we can grow repeatedly in the United States, um, or another country can do it.

Michael Marks: Biotech is one of the next important areas to watch. As of February, China's biotech licensing boom is on track to double again over the next 18 to 24 months. With global drug makers like Novartis, Merck and GSK already signing major deals. That is a reminder that the next strategic competition is not only about AI or chips, but, but how these are applied to advance the discovery pipeline in medicine, biology and advanced materials. This sector of advanced biology is quickly becoming the next arena where industrial capacity, scientific talent and national strategy are starting to emerge.

Vivek Tuakuri: I think that the capabilities, not just in the health and pharmaceutical domain, which are considerable, whether it's about productivity or lifespan or curing rare diseases, or making rare diseases, you know, chronic rather than fatal, I think, uh, those applications alone are massive. But then when you think about potential for like, energy security, and, you know, the Department of Defense is now actually trying to spin up a defense, uh, bio, biomanufacturing industrial base which doesn't exist today because they recognize that, wow, if we're in a conflict and these key inputs that we need for our explosives, for our, like, protective coatings on a fighter jet or whatever it is, are reliant this, you know, we have Such, such intricate globally connected supply chains. And then we wake up and realize we don't have that one thing from China or from a market that's somehow no longer available where we're screwed. Okay, maybe we need to be able to make some of this stuff at home. Uh, so I think it's worth, it's not getting nearly as much attention and it's a huge, and China by the way is definitely paying attention to this and recognize value of it. And this really brings back home one of the points we started the conversation with which is because the US has been a, has, is the biotech leader has been a biotech leader and it's because uh, m largely it's a part of our pharmaceutical leadership. Right. And there's a lot of money to be made in American pharmaceutical innovation in the American market. And that's where the, the dollars have gone because we're a free market society. There's less money to be made in some of the other applications I talked about on the defense side, the national security side, even the energy and ag site. And, and so that's, there's a role for the state and China is able, is playing that role and is signaling that demand is making those investments and you know, they could reap some significant rewards from it. Uh, because we don't really have a long term biotech strategy at all to speak of and.

Michael Marks: Right.

Vivek Tuakuri: Um, quite the opposite. So that would be one, there's others I would add to that list. But, but I think that's, that's a big one to watch.

Michael Marks: My final question, it fits all this stuff. So you know, we're in the venture business over here and so if you know, talking about how vibrant the United States still is despite all these obstacles and creating new companies and so on. Any advice you would have for you know, founders, entrepreneurs who are getting started now as young people in this complicated world we just talked about. I mean do they, do they just avoid China if they're in the United States? Do they try to find places to play play that are acceptable to both sides but any words of wisdom?

Vivek Tuakuri: Yeah, look, I mean I, I, I went to a, I, I have a public policy degree, not an mba so I wouldn't presume to give advice to, to people starting companies. But you know, I guess from a geopolitical perspective I would be definitely leery of exposure to the Chinese market and anything that has a whiff of a dual use technology or technology with the national security angle to it and you know, the aperture of what is included in that is broadening and broadening by the year. So look, but on the flip side of it, there are, you know, if you're starting a tech company, the digital economies of the global south are booming. Right. If you look at Southeast Asia, if you look at even just the Gulf. Right. Uh, we did a whole deep dive on what was happening in Saudi Arabia with MBS's attempt to diversify and transform that economy. If you look at Latin America and the data center Bonanza and Brazil, there is enormous opportunity all over the world. Even if China is a bit of a question mark as, as an overseas market. I mean, not to say nothing of India. Right. And I think, I think there's tremendous opportunity for people who are thinking about how to part find markets and partner with, to solve problems, find markets and partners in some of these really hungry emerging markets that are really looking at tech as a way to leapfrog and catch up.

Michael Marks: Yeah, that's pretty much how we think about it. Well, Vivek, I'm going to wrap this up. I'm sorry having to skip a lot of questions I had for you, but this was very insightful. I learned a lot. It would be great to have you on sometime in the future and see how some of these things play out. But thank you very much for taking the time to enlighten our listeners.

Vivek Tuakuri: Michael, you're very kind. I appreciate the chance to talk to you and learn a lot from you as well.

Michael Marks: Okay, great. Thank you. Technology continues to become less of a vertical sector and more like national infrastructure. The countries who grasp this are focused on developing a sustainable innovation machine. How to finance, manufacture, power, export, protect and attract talent into the system year after year. America's strength is that it still has deeper capital markets, stronger research institutions and a talent magnet most countries envy. But those advantages require tending what Viveka has described. The talent flight, policy instability, lack of long term planning is eroding some of America's long term technology leadership. China's strength, on the other hand, is that it thinks in systems and is willing to stay focused for a long time. Some of those investments are already paying off. As of April 2025, China accounted for more than 80% of the world's 16,000 humanoid robot installations. Chinese student numbers in the US have already fallen by more than 25% from 2019 to 2024, from about 370,000 to 277,000. This race will be decided not by any single breakthrough by which country can build a system that keeps producing them and compounding its advantages year after year. Thanks for tuning in to the Tech Surge Podcast from Celeste Capital. If you enjoyed this episode, feel free to share it, subscribe or leave a review on your favorite podcast platform. We'll be back every two weeks with more insights and discussions of all things deep tech.