The State of AI: Is China about to win the race?
However, these same restrictions have pushed Chinese companies Towards different rules of the game: Clustering computation, improving efficiency, and releasing open-weight models. For example, the DeepSeek-V3 training program used only 2.6 million GPU hours, which is much less than its US counterpart. But Alibaba’s Qwen templates are now among the most downloaded open templates in the world, and companies like Zhipu and MiniMax are building competitive multimedia and video templates.
China’s industrial policy means that new models can move from the laboratory to implementation quickly. Local governments and large companies have already begun to introduce logical models in management, logistics, and finance.
Education is another advantage. Major Chinese universities They implement literacy programs in the field of artificial intelligence in their curricula, and integrate skills before the labor market requires them. The Ministry of Education also announced plans to integrate AI training for children at all school ages. I’m not sure the phrase “engineering state” fully captures China’s relationship with new technologies, but decades of infrastructure building and top-down coordination have made the system extraordinarily effective at driving widespread adoption of technologies, often with far less social resistance than you might see elsewhere. Naturally, widespread use allows for faster iterative improvements.
Meanwhile, Stanford High Artificial Intelligence Index 2025 Chinese respondents were found to be the most optimistic in the world about the future of AI, and significantly more optimistic than residents of the US or UK. This is striking, given that the Chinese economy has slowed since the pandemic first broke out more than two decades ago. Many in government and industry now see AI as a much-needed spark. Optimism may be a powerful fuel, but whether it can survive slower growth remains an open question.
Social control remains part of the picture, but a different kind of ambition is taking shape. The Chinese AI founders of this new generation are the most globally minded I’ve seen, moving seamlessly between Silicon Valley hackathons and pitch meetings in Dubai. Many of them are fluent in English and in the rhythms of global venture capital. Having watched the last generation grapple with the burden of Chinese branding, they have now been quietly building transnational companies from the beginning.
The United States may still be a leader in speed and experimentation, but China is capable of shaping how artificial intelligence becomes part of everyday life, both at home and abroad. Speed is important, but speed is not the same thing as mastery.
John Thornhill Responses:
You’re right, Caiwei, that speed is not the same as superiority (and “murder” might be too strong a word). You are also right to exaggerate the point about China’s strength in open-weight models and the US preference for proprietary models. This is not just a conflict between two economic models of two different countries, but also a conflict between two different ways of spreading technology.
