The United States and China seem to compete in ia. The reality is that they play completely different sports

Everyone talks about the “career of AI” between China and the United States as if it were a competition where one will win and another will lose. But in reality there is no career because everyone is playing a different sport.

• The United States invested 100,000 million dollars in AI in the first half of 2025, pursuing AGI as a technological religion. His philosophy is clear: spending billions today to control the future tomorrow. Giant proprietary models, infinite parameters, the search for the Holy Grail of the AGI.
• China, meanwhile, has not allocated amounts equivalent to its technological sector. Its approach is as pragmatic as Chinese culture itself: to capture market now, monetize in applications later.

This divergence does not arise from abstract philosophies, but three concrete structural forces:

1. Chips scarcity. The OCTOBER US SANCTIONS 2022 They cut Chinese access to the most powerful NVIDIA chips. The paradoxical result: China was forced to innovate in efficiency. Deepseek He managed to match the performance of GPT-4 spending eighteen times less money. Technological restriction forced an economic advantage that the United States, with all its hardware, cannot easily replicate.
2. Capital scarcity. The Chinese risk capital collapsed after the regulations of 2022. Chinese the startups collect funds only after demonstrating product working and real metric uses. In this context, opening the source code becomes the smartest strategy: free marketing, viral adoption and developer ecosystem without acquisition cost. BAICHUAN AI AND ZHIPU AI They achieved financing from Alibaba and Tencent precisely after publishing their open models and demonstrating traction.
3. The concentration of applications. The average Chinese user uses less than ten apps per month, almost all channeled through Wechat or Alipay. The American is around thirty. This structural difference completely changes the penetration strategy. In China, who conquered those few points of mass distribution conquers the entire market. In the United States, you have to fight in dozens of different fronts. That is why it makes sense that Chinese companies open their models: they need adoption speed in those bottlenecks, do not own defenses.

The most revealing case is how Wechat chose Deepseek Above Yuanbao, Tencent’s own. The decision exemplifies well Chinese pragmatic mentality: the functional product triumphs over corporatism. Yuanbao was his own, but Deepseek was better. Wechat needed quality already, not in two years.

Twenty Chinese car manufacturers and more than one hundred hospitals have made similar decisions, integrating Deepseek into their systems. Each integration expands the ecosystem and increases change costs.

Behind the Chinese strategy there is also an emotional component that should not be underestimated. After decades being labeled Copycats, Opening the source code has become a way of demonstrating real innovative capacity. Publishing the weights of a model is equivalent to saying: “Here is our work, examine it, improve it if they can.” It is national pride turned into business strategy.

The United States continues to build the most powerful AI in the Universe. China is already selling functional to half the planet. Sam Altman talks about the AGI As a civilizational destination, Chinese executives apply the proverb “interests cannot be eaten” and seek immediate monetization.

Both can win their respective sports, but it would be absurd to ask who is winning the World Cup when one plays football and the other to basketball.

The interesting question It is not who will win the race, but what will happen when in five years One has built the perfect AGI and the other controls the AI infrastructure that uses half the planet.

In Xataka | The Chinese government does not trust Nvidia: its survival in China right now hangs from a thread

Outstanding image | Xataka