I'll spoil the article for myself, because this is noteworthy, in my opinion. Listen: the head of the company that supplies artificial intelligence chips to the entire planet just said that the United States will lose the AI race. He didn't say "could lose," and he didn't even say "risk losing." He actually said "will lose."
Jensen Huang He doesn't mince his words when he talks about China: in the red corner, there's essentially free energy for data centers, streamlined regulation, and half the world's AI researchers are concentrated there. In the other corner, the US with its "cynicism" (his words) and 50 states discussing 50 different regulations. The AI race has a favorite, Huang says. Financial Times. And that's not what Washington wants to hear.
Free energy versus bureaucracy
Huang's words come on the sidelines of Future of AI Summit of London, a few days after the meeting between Donald Trump e Xi Jinping in which the White House confirmed the ban on chip exports Blackwell more advanced towards China. The American strategy focuses on technological containment, but Huang does not seem to be convinced of the effectiveness of this idea.
“In China, energy is free,” he said (in a chilling but effective simplification) referring to Chinese government subsidies for data centers. ByteDance, Alibaba e TencentBeijing has increased incentives after tech companies complained about higher operating costs due to domestic chips. Huawei e Cambricon, less efficient than Nvidia's but still functional. And then, I ask quietly so only you can hear me: HOW MUCH less efficient, exactly?
In short: while the United States is debating potential new state (not federal) regulations on artificial intelligence, China is eliminating energy costs and accelerating. The contrast is stark: regulatory chaos and fragmentation versus centralized support. Caution, or slowness, versus recklessness—or, if you prefer, speed.
Nanoseconds, not years
Huang has changed his tone dramatically compared to the past. In October he declared that China was “nanoseconds behind” the United States, no longer years as we thought. Now it goes further: China will win the AI race. Why has it upped the ante?
Approximately 50% of the world's AI researchers work in China, and Washington is losing access to this talent pool with policies Huang calls counterproductive. "We want America to win this race. But we need to be in China to win over their developers. A policy that causes us to lose half the world's AI developers isn't beneficial in the long run; it hurts us even more," he said at the Nvidia conference in Washington in October.
AI Race: All the Competition's Numbers
Il 2025 AI Index Report from Stanford confirms that the United States released 40 significant AI models in 2024, compared to 15 in China and just 3 in Europe.
On the quantitative front, Washington maintains its advantage. But China has deposited more than 38.000 patents in generative AI between 2014 and 2023, six times more than the US over the same period. And the performance of Chinese models is rapidly closing the quality gap with American ones.
DeepSeek, a Chinese startup that emerged this year, has developed models that compete with OpenAI at dramatically lower costs, proving that chip restrictions aren't enough to stop innovation.
US investments in 2024 were 109,1 billion dollars against 9,3 of China, but Beijing compensates with a centralized national strategy and direct government support. The plan Made in China 2025 It has made the AI race a state priority, with world-class universities and dedicated infrastructure fueling a rapidly expanding ecosystem.
AI Race: Optimism vs. Cynicism
Huang, as I was telling you, identified Western "cynicism" as one of the main obstacles. "The West, including the United States and Great Britain, is held back by cynicism. We need more optimism," he declared. It's not just about technology or capital. It's a question of approach: China can move quickly because the government directly controls private innovation, data access, and infrastructure development. The United States must contend with fifty states that could adopt different regulations, slowing down the entire system.
Huang's statements take on particular weight because Nvidia still controls approximately 95% of the global market than artificial intelligence chips. No one has a better view of where this competition is headed. And the picture he paints is not reassuring for Washington.
Trump responded that Blackwell's most advanced chips will remain an American exclusive, but Nvidia hasn't even requested export licenses to sell them in China, citing Beijing's stance toward the company. This unstable balance reflects an uncomfortable truth: the AI race is being played out on multiple fronts, and blocking chips may not be enough if the other side has unlimited energy, motivated researchers, and a state that sets priorities without discussion.
Let's be honest, I spoiled the article for you right away. What did I get? American technology is superior. It is, for now. But how long can this superiority last when the other side runs much faster, spends much less, and has far fewer constraints? Come on, you can spoil the ending.
Huang also gave his answer. It remains to be seen whether anyone in Washington will listen.
