US vs China — The Race That Matters

Everything I've laid out so far - the convergence of AI, robotics, and energy, the disruption of legacy industries, the barbell reshaping of society - it all assumes something that needs to be said explicitly. It assumes that AI development continues to advance primarily in environments where competition, individual achievement, and market forces remain intact. And that assumption is far from guaranteed. I want to be direct about something most people dance around, or simply don’t think about: it is extremely important for the United States to win the AI race against China. I know that sounds nationalistic, and I know there are people who will bristle at framing this as a competition at all. But the stakes are too high to be polite about it.

Why This Race Matters

I want you to think about AI not as a product, but as a force multiplier for

whatever system deploys it. AI in the hands of a democratic, capitalistic society will be shaped, ideally, by competitive forces. Multiple companies, multiple approaches, multiple people all pushing in different directions, with the outcome determined by markets and consumer choice. AI in the hands of an authoritarian regime will be shaped by whatever that regime wants it to do.

The way I see it, this is the most important variable in determining whether

the Age of Abundance I've described actually materializes as something good for humanity, or becomes a tool for control that makes previous authoritarian regimes look primitive. Think about surveillance. Think about social scoring. Think about the ability to monitor every transaction, every conversation, every movement of every citizen. Now multiply that capability by a thousand. That's what AI-enabled authoritarianism looks like. And it's not hypothetical. AI will be powerful - we've established that. The real question is: powerful in whose hands, deployed according to whose values, controlled by whom? The Jack Ma Lesson I think the clearest illustration of the difference between the US and Chinese systems is what happened to Jack Ma. Jack Ma built Alibaba into one of the most valuable companies in the world. By any measure of entrepreneurial success, he was at the very top. He created massive value, employed hundreds of thousands of people, and became the most prominent businessman in China. And then he gave a speech criticizing Chinese financial regulators. What happened next should be instructive for anyone trying to understand how these two systems differ. The Ant Group IPO - which would have been the largest in history - was pulled at the last minute. Jack Ma essentially disappeared from public life for months. When he reemerged, he was diminished. His companies faced antitrust investigations. His influence evaporated. Now think about what this tells you about the ceiling on achievement in China. In that system, no one - absolutely no one - can be perceived to be more powerful than the Communist Party. No one can threaten the power and decision-making ability of Xi Jinping. The moment any individual's influence threatens to rival the state's, that individual gets cut down. Contrast this with Elon Musk in the United States.

The Elon Comparison

Elon Musk is easily the most powerful private citizen in the world. He

controls the dominant electric vehicle company in the US, the dominant commercial space company in the world, a major social media platform, and one of the leading AI companies. His wealth exceeds the GDP of many countries. At the time of this writing, his net worth is somewhere around $800 billion. His decisions affect global markets, geopolitics, and technological development at a civilizational scale. And the US government can't just make him disappear. And trust me - they’ve tried. I'm not saying Elon is above the law. He has faced SEC investigations, regulatory battles, and constant media attacks. But he remains in the arena. He remains powerful. He remains able to compete, build, criticize, and push forward with his vision regardless of whether the government approves. This is the fundamental difference. In the United States, there is no ceiling on individual achievement. You can theoretically become a centibillionaire. You can become arguably more influential than the President on technological matters. You can build companies that reshape entire industries and the government's response is not to eliminate you, but to try to regulate you within a legal framework. In China, the ceiling is the Party. And the Party will always protect itself first.

Why This Affects AI Development

Now you might be thinking - okay, but why does this political difference

matter for AI specifically? And the answer goes back to what actually drives AI advancement. AI is developed by people. The best AI researchers in the world can choose where to work. They're highly mobile, highly educated, and have options. The question is: where will the top talent want to be?

In a system where your success can be confiscated the moment it becomes

inconvenient for the ruling party, the incentives are warped. Why push boundaries? Why challenge assumptions? Why build something that might disrupt existing power structures? The rational response in an authoritarian environment is to play it safe, to avoid attention, to make sure your work serves the state rather than challenges it. In a competitive environment, the incentives are the opposite. Push boundaries. Challenge assumptions. Disrupt existing structures. That's how you win. That's how you get rich. That's how you change the world. This is why the best AI talent has historically concentrated in the US. Not just because of salaries or resources - China can match those. But because of freedom. The freedom to pursue ideas wherever they lead. The freedom to fail without state consequences. The freedom to succeed without the state taking it away.

China's Open-Source Strategy

I want to be clear about something: China is not sitting still, and they're not

stupid. They have a strategy, and it's clever. China has been releasing major AI models as open source. DeepSeek and other Chinese AI labs have put out models that compete with the best American offerings - and given them away for free. A lot of people look at this and think China is being generous. That they're contributing to global AI development. That open source is inherently good and we should celebrate this. I think that completely misses what's actually happening. China's open-source AI strategy is designed to weaken America's private AI development model. Think about it from their perspective. American AI companies like OpenAI, Anthropic, and xAI have raised hundreds of billions in capital on the premise that they can build proprietary AI systems that will

be worth trillions. That capital funds the R&D, the talent acquisition, the compute infrastructure. Now imagine if comparable AI capabilities were simply free. Available to everyone. No need to pay for ChatGPT or Claude or Grok because an open- source Chinese model does the same thing. What happens to the American AI business model? What happens to all that capital that was premised on building something differentiated? It evaporates. This is economic warfare disguised as generosity. China is trying to commoditize the layer that American companies are trying to monetize. If they succeed, they undermine the entire economic engine driving American AI development. And the uncomfortable part: in some ways it's working. DeepSeek's releases have already affected valuations and strategic planning at American AI companies. The threat of free-good-enough AI models hanging over the industry changes the math on what's worth building.

The Manufacturing Reality

China's manufacturing capabilities are genuinely impressive, and will be a key

variable in the development of humanoid robots. China has built the manufacturing base that America used to have. They can build things fast, at scale, and at costs that American companies can't match. When Apple needs to ramp up iPhone production, they do it in China. When Tesla needed to expand capacity quickly, they built a factory in Shanghai that went from empty lot to producing cars in under a year. Try doing that in America. This manufacturing expertise extends to everything AI needs. Batteries, chips, components, devices - China either makes them or can make them. The physical infrastructure of the AI era runs through Chinese factories.

Now, some people look at this and say we should decouple completely. Bring

all manufacturing back to America. Reduce all dependencies on China. And I understand the appeal of that argument. But I think it misses the practical reality. You can't rebuild 40 years of manufacturing infrastructure overnight. You can't train a workforce. You can't develop the supplier relationships. Even if you started today with unlimited money, it would take a decade or more to replicate what China has built. So we're in this awkward position where we're competing with China in AI while depending on China for much of what goes into AI systems. That tension isn't going to resolve quickly, and anyone who tells you otherwise is selling something. The question is how to manage that dependency while the competition plays out. How to reduce the most critical vulnerabilities without blowing up the entire global supply chain. How to build alternatives without pretending we can snap our fingers and bring everything home.

Why China Moves Faster

There's another uncomfortable truth: China can often move faster than we

can on major projects. When China decides to build high-speed rail, they build thousands of miles of track. When America decides to build high-speed rail in California, we spend decades on environmental reviews and legal challenges and still don't have a working system. When China decides to deploy 5G infrastructure, they deploy it nationwide in a couple of years. When America tries to deploy 5G, we have endless debates about which frequencies to use and how to handle the rollout. This speed advantage matters for AI. Training AI models requires massive data centers. Those data centers require power infrastructure, cooling

systems, buildings. The country that can build this stuff fastest has an advantage. Elon has talked about this. The buildout speed of xAI's Colossus cluster was remarkable by American standards - just 122 days from start to operational. But that required Elon's particular ability to cut through obstacles. Most American companies can't move at that pace. In China, if the government decides something gets built, it gets built. There's no environmental litigation. There's no NIMBY opposition. There's no endless permitting process. The state says build, and building happens. Now, this speed comes with costs. China has built ghost cities. They've built infrastructure that didn't make economic sense. The lack of checks means bad decisions get implemented just as fast as good ones. Authoritarianism is fast in all directions, including wrong directions. But for AI specifically, where the race is measured in months and years rather than decades, China's ability to build fast is a meaningful advantage that we need to acknowledge.

The Chip Battleground

Beyond software, there's hardware - and the implications are even larger. The United States has been trying to restrict China's access to advanced AI chips - the Nvidia H100s and similar hardware that power the largest AI training runs. The logic is straightforward: if you can control the hardware, you can control the pace of AI development. But there's a tension here. NVIDIA wants to sell chips wherever they can. American semiconductor companies have huge revenue from China. So you have this push and pull between national security interests that want to restrict chip access and commercial interests that want to maximize sales. But I think the chip restriction strategy might not even work the way it's intended.

China could have inferior hardware but beat us on algorithms. They could

develop more efficient training methods, better architectures, smarter approaches to solving AI problems that require less raw compute. History has shown that constraints often drive innovation. When you can't brute force a problem with resources, you get creative. I'm not saying China will definitely overcome the chip restrictions. But I am saying we shouldn't be complacent about hardware advantages. The race is going to be won on multiple fronts, and China is competing seriously on all of them.

What America Gets Right

Despite my concerns, I remain optimistic about America's position. And let

me explain why. First, capitalism drives innovation in a way that central planning simply cannot match. When you have multiple companies competing fiercely, each trying different approaches, each with skin in the game, you get a diversity of experiments that no government committee could ever design. Some of those experiments fail. Most of them fail. But the ones that succeed create breakthroughs that planned economies cannot replicate. Tesla, SpaceX, xAI, OpenAI, Anthropic, Google DeepMind - these are all pursuing different strategies for AI and related technologies. They're competing viciously. They're stealing talent from each other. They're trying to out-execute each other. And that competition is what drives progress. Second, democracy provides error correction that authoritarianism lacks. When policies don't work in a democracy, they eventually get changed. Leaders get voted out. New approaches get tried. It's messy and slow, but it's adaptive. Authoritarian systems can make decisions faster, but they can also make catastrophic errors that persist because no one is allowed to challenge them. Third, there's no ceiling. The most ambitious people in the world know that in America, their upside is unlimited. They can build the next Tesla or

SpaceX or OpenAI and keep what they create. That attracts talent and

ambition from everywhere. China has to work with whatever talent stays within its borders or can be convinced to return. America attracts talent from the entire world.

What America Gets Wrong

But I'm not blind to America's weaknesses either. And if we're going to win

this race, we need to be honest about them. The bureaucracy is killing us. Try to build a new factory in America and count how many permits, environmental reviews, labor negotiations, and regulatory approvals you need. Compare that to China, where the government can simply decide something will be built and it gets built. Our regulatory apparatus was designed for a different era, and it's actively hindering our ability to compete. The education system is failing. I've talked about this elsewhere in this book, but America's education system is producing workers instead of entrepreneurs, test-takers instead of problem-solvers. We're falling behind on STEM education while arguing about things that have nothing to do with preparing the next generation to compete. The political polarization is exhausting. Every issue becomes tribal. Every policy debate becomes about teams rather than outcomes. China doesn't have to deal with this. They can align around a strategy and execute. We struggle to align around anything. And the short-termism in American business and politics makes it hard to sustain long-term technological investments. Quarterly earnings, two-year election cycles - the time horizons are all wrong for the kind of decade-long pushes that winning the AI race requires.

The Taiwan Question

I would be remiss not to mention Taiwan, because it's central to this

competition. Taiwan manufactures the majority of the world's most advanced semiconductors. TSMC is the cornerstone of the global chip supply chain. And China claims Taiwan as its territory. You can see the problem. If China took Taiwan - through military action, economic coercion, or political annexation - they would control the hardware that powers AI development globally. They wouldn't need to develop their own advanced chip manufacturing. They could simply take it. This is one of the most dangerous scenarios for the AI race. Not because China would necessarily shut off chip supplies to America - though they might - but because it would fundamentally shift the balance of power in technology. Control the chips, control the AI. Control the AI, control the future. The United States has been moving to reduce this dependency. The CHIPS Act is funding domestic semiconductor manufacturing. Intel is trying to catch up. But building semiconductor fabs takes years, and we're starting from way behind. This is a vulnerability that won't be fixed quickly. And I think it's worth noting: TSMC is building fabs in Arizona, but they've faced challenges that illustrate exactly what I was saying about American manufacturing. Worker availability, cost overruns, delays - all the things that China seems to avoid by simply commanding resources to show up where they're needed. American workers are fine - the problem is systemic. The regulatory environment, the cost structure, the legal risks - everything is stacked against fast, cheap manufacturing in America. And that's a strategic problem when you're racing against a country that has optimized for exactly that.

The Data Advantage Nobody Talks About

Now, I want to bring up something that gets overlooked in the US-China AI

discussion: data. China has over 1.4 billion people generating data. The government has broader data access than in the US, despite China's comprehensive PIPL privacy law. They have massive platforms - WeChat, Alibaba, TikTok's Chinese version Douyin - that collect enormous amounts of behavioral data. In AI, data is everything. The models are only as good as what you train them on. And China has access to data about human behavior at a scale that's hard for American companies to match, especially as privacy regulations in the US and Europe continue to tighten. Now, there are counterarguments. The data is in Chinese, which limits its applicability to English-language AI systems. The censorship within China means certain types of data don't exist. The cultural norms are obviously very different. And the quality of data matters as much as quantity. But I think this is an underappreciated factor in the competition. American AI companies are building on English-language internet data that's increasingly being locked down by the platforms that own it. Twitter data, Reddit data, YouTube data - it's all being monetized or restricted. Meanwhile, China can mandate that its companies share data for AI training if the government decides that's strategically important. This is another asymmetry in how the two systems compete. American AI companies have to negotiate and pay for data access. Chinese AI companies have a government that can simply direct data to flow where it's needed.

The Competition Within Competition

And I want to add a nuance here that I think gets lost. This isn't a pure two-

player game between US and China. Within each country, there are competing factions, interests, and approaches.

In the US, you have the AI safety crowd who want to slow down

development. You have the accelerationists who want to push as fast as possible. You have legacy companies trying to protect their positions and startups trying to disrupt them. You have politicians whose expertise lies elsewhere trying to regulate technology they haven't had time to deeply study - which is understandable given how fast this field moves, but still creates a knowledge gap. In China, you presumably have similar internal debates, though they're less visible. There are factions within the Party with different views on technology, on openness, on how to compete with America. The outcome of the race depends not just on US versus China but on which factions within each country end up driving strategy. If the AI safety crowd in America succeeds in dramatically slowing development, while China's accelerationists push forward, that changes the trajectory. If China's more cautious factions prevail and they pull back on AI development for whatever reason, that changes things too. This is why I'm hesitant to make a definitive prediction. There are too many variables, too many moving parts, too many internal competitions within the larger competition.

The Stakes

Let me bring this back to what really matters. If the United States wins the AI race - and by "wins" I mean maintains leadership in AI development, preserves competitive market structures, and ensures AI is deployed in ways that respect individual rights - then the future I've described throughout this book becomes possible. The Age of Abundance, properly managed, with benefits distributed widely. If China wins - given the current state of the country, authoritarian control of AI becomes the dominant model globally - then we're looking at something very different. AI as a tool of state control. AI as a mechanism for

surveillance, scoring, and compliance. AI that serves the Party rather than the people. Now, I need to be honest about something uncomfortable: the United States isn't immune to this risk either. AI-powered surveillance is likely already deployed domestically in some respects, and it’s not well understood if these tools consider people’s freedoms and privacy. Facial recognition. Predictive policing. Social media monitoring. The tools of authoritarian control can exist in America too. And with AI capabilities advancing, the temptation to use them will only grow. Could the US devolve into its own version of an AI-empowered Big Brother state? Absolutely. The technology doesn't care about political systems. The same AI that China uses for social credit scoring could, in theory, be deployed by any government. What gives me (theoretical) confidence in American resilience is structural: democracy, free markets, independent judiciary, free press, decentralized power. These institutions create friction against authoritarian drift. They're not perfect - they've failed before - but they provide checks that don't exist in China's system. In theory, if an American administration tried to deploy AI for mass surveillance, there would be lawsuits, Congressional hearings, journalistic investigations, corporate resistance, and electoral consequences. In theory. I emphasize "in theory" because these safeguards are not guaranteed. They depend on continued civic engagement, institutional integrity, and public vigilance. The price of freedom is eternal vigilance, and AI makes that vigilance both more difficult and more necessary. So when I say it matters who wins the AI race, I'm not saying America is perfect or that American AI leadership automatically produces good outcomes. I'm saying that American AI leadership preserves the possibility of good outcomes in a way that Chinese AI leadership does not given its current state. The difference is between a system where course correction is possible and a system where it is not.

I don't think this is hyperbole. I think this is the actual stakes of the

competition. And I think most people in America don't appreciate how serious this is. They see AI as a consumer product, as a productivity tool, as something that generates images or writes emails. They don't see it as a civilizational fork in the road where the path we take determines whether humanity moves toward freedom or control.

What Needs to Happen

So what should America do? I don't have all the answers, but I have some

thoughts. First, we need to streamline the regulatory environment for AI development and deployment. That doesn't mean no regulation - I've been clear that I'm pro-regulation for ensuring benefits are distributed. But it means smart regulation that doesn't slow us down relative to China. Second, we need massive investment in semiconductor manufacturing and supply chain resilience. The Taiwan dependency is a strategic vulnerability that must be addressed. Third, we need to fix the education system to produce more people capable of contributing to AI development. This is a generational project, but it needs to start now. Fourth, we need to recognize China's open-source strategy for what it is and develop a response. That might mean more public funding for open-source AI development in the US. It might mean different business models. It might mean accepting that some AI capabilities will be commoditized and competing on what can be differentiated. Fifth, we need political leadership that actually understands technology and can make coherent long-term strategy. This is maybe the hardest one, given how our political system works. And last not but least, we need energy. SO much energy. ALL the energy.

The Talent Migration Pattern

I think one of the most important dynamics to watch is where AI talent

chooses to live and work. For decades, the best technical talent from China came to America for graduate school and stayed. They built careers at American tech companies. They founded startups in Silicon Valley. China trained them, America kept them. That pattern is changing. China has invested heavily in keeping its talent home and bringing emigrants back. They've built research institutions, funded startups, and created pathways for technical workers to achieve success without leaving China. The nationalist sentiment has increased. The cultural pull to return home has strengthened in some ways. Meanwhile, America has made it harder to stay. Visa restrictions, political tensions, and a general climate of suspicion around Chinese nationals in tech has pushed some people back who might otherwise have stayed. Not to mention the political climate around immigration, period. Every time we make it harder for a talented Chinese engineer or researcher to build a life in America, we're potentially handing that person to our competitor. This is a tricky balance. There are legitimate national security concerns about technology transfer and espionage. I'm not naive about that. But I also think we've swung too far toward suspicion in ways that hurt our competitive position. The ideal scenario is that America remains the most attractive place for ambitious, talented people to build their careers - regardless of where they were born. That's what made Silicon Valley what it is. That's what built the American tech industry. If we lose that, we lose one of our most important advantages.

What Victory Actually Looks Like

I've been talking about "winning" the AI race, but I should be specific about

what that means. Winning doesn't mean eliminating China as a competitor. That's not going to happen. China will continue to develop AI capabilities regardless of what we do. They're too large, too capable, and too determined to be shut out. Winning means maintaining leadership in the most advanced capabilities. It means ensuring that the norms around AI development reflect competitive market values rather than authoritarian control. It means being the place where the most important AI work happens, where the best talent wants to be, where the most valuable applications get built first. Winning means that when the AI era fully arrives - when artificial general intelligence becomes possible - it emerges in an environment where multiple players compete, where no single actor controls it, where checks and balances exist. Not in an environment where one government monopolizes the capability and uses it to entrench power. That's what's actually at stake. The structure of the system that will govern the most powerful technology humanity has ever created. National pride and economic dominance are secondary concerns.

The Optimist's Take

I'll end this chapter on an optimistic note, because despite everything I've laid out, I do believe America can and probably will maintain AI leadership. The fundamental advantages - capitalism, democracy, no ceiling on achievement, ability to attract global talent - are real and durable. They're not going away. China's strengths are also real, but they come with structural weaknesses that are hard to overcome. The people building AI in America are some of the most capable, ambitious, and driven people in the world. They understand the stakes. They're not going to let America fall behind without a fight.

And the race isn't over. It's just getting started. There's time to course correct, to address weaknesses, to double down on strengths. But we have to take it seriously. We have to stop pretending this competition doesn't exist, or that the outcome doesn't matter, or that technology development is somehow neutral with respect to political systems. The race matters. The outcome matters. And if you’re American, America needs to win. Because you know damn well the Chinese want the same for their country - as they should.