Artificial intelligence is reshaping ideas about which regions of the world carry strategic importance. Competition between the United States and China is unfolding simultaneously on technological and geopolitical planes, and an advantage in one sphere does not guarantee victory in the other. Bloomberg Opinion columnist Hal Brands writes about this.

"Technology is constantly changing, while geography is far more permanent. But innovation continuously reshapes the world’s strategic geography by changing the significance and value of key areas," Brands notes.

The paradox of the current confrontation is that Washington is trying to restrain Beijing’s technological rise, while American companies are increasingly actively using Chinese AI models. This dependence became the subject of a separate report by The Washington Post, which discusses the growing role of Chinese developments in the American AI economy.

From Steam to Silicon

Brands compares AI with earlier technologies that changed geopolitics. Sail power, and later steam, opened the "era of European global dominance"; the internal combustion engine turned the Middle East into a "gold mine and a powder keg"; and nuclear weapons reinforced the order of the Cold War superpowers. AI, he writes, is "once again changing the strategic situation." Bloomberg describes this article as a story about how technology is "changing the very landscape of geopolitics." It continues Brands’ previous columns, in which he warned that America’s lead in AI is shrinking.

Dependence on a Rival

As The Washington Post reporter David J. Lynch reports from Beijing, American officials claim that Moonshot AI created its models through the "malicious" misappropriation of developments from American technology companies. In addition, they view the company as part of a Chinese Communist Party program aimed at making American business dependent on cheap Chinese AI.

This dependence intensified over the summer. In July, Moonshot introduced Kimi K3, a model with 2.8 trillion parameters that the company calls the world’s largest AI system with open weights. According to the company, K3 is close in capability to Anthropic’s flagship Fable model. Later, the Trump administration accused Moonshot of using "covert," though not necessarily illegal, methods and of building K3 on the basis of Fable. In Beijing, the accusations regarding illegal "distillation" were called "groundless."

As Fortune writes, American startups and companies on the Fortune 500 list are quietly adding Chinese models to their operations in order to control AI costs. According to the publication, generation in Anthropic’s Fable costs $50 per million tokens, while in DeepSeek-V4-Pro it costs about $0.87. According to the Associated Press, Coinbase has announced that it is switching to Chinese models to reduce costs. In addition, AP reports that on the OpenRouter platform, all five of the most popular models over the course of a month were Chinese.

At times, Washington’s own policy was playing in favor of its rival. Z.ai released the GLM-5.2 model in June, shortly after Anthropic’s Fable and Mythos models were unavailable for more than two weeks because of export restrictions. "Restrictions on an American model can instantly open opportunities for a Chinese competitor," Arena platform CEO Anastasios Angelopoulos told AP.

Chinese Models Are Narrowing the Gap, but the Divide Remains

Despite the rapid growth of China’s AI industry, it is still too early to speak of full technological parity. In Angelopoulos’s assessment, if the full spectrum of capabilities is taken into account, Chinese models still lag behind America’s leading systems.

At the same time, the spread of Chinese developments is also raising questions within the United States itself. Congress is scrutinizing American companies that use Chinese AI models, since such practices concern not only business costs but also the country’s technological security.

Mozilla CTO Raffi Krikorian summed up the ongoing changes this way: "The open leading edge is increasingly becoming Chinese."

For Washington, this means that competition with Beijing cannot be limited to export restrictions on technology or the struggle for computing power. Even if American companies retain an advantage in the most advanced systems, Chinese models may prevail on the basis of price and accessibility.

As a result, artificial intelligence is changing the world’s strategic map in several directions at once. The United States continues to compete with China for technological leadership, but at the same time its own business is becoming increasingly dependent on the solutions of Chinese developers. And it is precisely this contradiction that is becoming one of the main challenges for American strategy.