Column: What makes China likelier than the U.S. to win the AI race?

by Steven Hoffman

As the Chairman and CEO of Founders Space, I spend a great deal of time in conversations with influential entrepreneurs and investors across China. Those discussions reveal a picture of the global artificial intelligence (AI) race that feels very different from the one often portrayed in Western media.

When you look closely at what platforms like DeepSeek, Qwen, Kimi Chat, and Doubao are building, you start to see a structural shift that gives Chinese companies momentum the United States is struggling to match. Three forces in particular stand out, and together they form a foundation that could reshape the balance of global AI power.

The first force comes from a surprisingly old-fashioned element of modern technology: energy. Training frontier models depends on enormous amounts of power. In China, the state views this as a national priority and has created an environment where energy costs can be brought down through direct support and long-term planning.

At the same time, the United States is experiencing rising energy costs, aging infrastructure, and policy decisions that have set back renewable investments. With the country moving away from clean energy incentives and stumbling on grid modernization, the gap grows wider.

Cheap and reliable power may sound mundane compared to algorithmic breakthroughs, but behind the scenes, it influences every important decision a frontier model company makes. Costs compound over time, and China is deliberately positioning its ecosystem to carry those costs with far more ease.

The second advantage is China's new relationship with data. In the United States, data remains locked away in private silos, guarded by corporations and fragmented across thousands of institutions. The government has limited authority to open these vaults, even when shared access would serve innovation.

China is taking the opposite approach. It has the ability to mandate that large companies contribute data to shared pools and is building a national data exchange system designed specifically to fuel the next wave of AI growth. This does not simply create larger training sets. It also allows Chinese startups to train finely tuned models for every major industry, from finance to logistics to healthcare. When that type of domain-specific data becomes widely accessible, it accelerates an entire ecosystem rather than a handful of tech giants.

The third force comes from Huawei, whose reach across both chip technology and networking gives China a structural edge the United States cannot easily replicate. Because Huawei controls the networking layer inside its data centers, it has redesigned those systems so they can extract more performance out of cheaper, lower-power chips.

Most Western companies rely on chips from one firm and networking gear from another, which limits how deeply they can integrate the two. Huawei is crafting data centers where the network fabric and the chips speak the same native language. The result is a path to future parity using less expensive hardware, and that creates an economic advantage that compounds at scale.

When you place these pieces together, a larger picture emerges. Cheap power reduces the cost of training and inference. Open and mandated data access strengthens the diversity and quality of models. Integrated networking and chip design narrows the performance gap even when using less advanced hardware. Combined with China's embrace of open source model development, these advantages create an ecosystem where AI becomes both powerful and accessible at prices Western companies will struggle to match.

This is not a story about who has smarter engineers or more ambitious founders. It is a story about national strategy, infrastructure, and the structural forces that shape technological eras. And right now, those forces are tilting in China's favor.

Editor's note: Steven Hoffman is Chairman and CEO of Founders Space, and is the author of "Make Elephants Fly."

The views expressed in this article are those of the author and do not necessarily reflect the positions of Xinhua News Agency.