The United States wants the world to take sides on AI. But for many countries, China’s narrative could be more convincing

Hong Kong —

Rapid advances in Chinese artificial intelligence over the past year appear to have alerted Washington to the real possibility that the world’s second-largest economy could pull ahead of the United States in an increasingly tight technology race.

“We’re leading China in AI…and frankly, I want to keep it that way, because whoever wins in AI will win,” US President Donald Trump said earlier this month.

Trump’s comments highlight the challenges of a technological competition between the United States and China, increasingly framed in existential terms. The United States has a clear advantage, with pioneering AI labs, the most advanced chips and deep-pocketed investors. Taking advantage of this dominance, Washington is also forcing countries around the world to choose sides.

But China has a vision that, for many, might be more compelling. Although it lags behind the United States in cutting-edge AI models, it is rapidly closing the technology gap while championing a more open ecosystem focused on accessibility and cost reduction.

In July, Chinese leader Xi Jinping launched the World Artificial Intelligence Cooperation Organization — a rival group to the Pax Silica alliance that Trump formed last year to reduce reliance on China for AI supply chains.

While more than two dozen countries and the European Union have signed on to Trump’s Pax Silica, Xi has recruited 29 countries, including Russia, Indonesia and Pakistan, to his alternative vision of open models, which allow users to freely download, customize and run without paying high fees to U.S. companies like Anthropic and OpenAI.

For developers in the Global South, an inexpensive Chinese model of DeepSeek or Moonshot may be more useful than a slightly more capable system requiring an expensive subscription and access to a foreign cloud provider, said Eric Olander, editor-in-chief of the China-Global South Project, a research agency.

Amid concerns about AI’s potential threat to humanity and growing calls for a global watchdog, the technology emerged as a key focus of the Trump-Xi summit this week.

The right thing to do on AI, Xi said in his talks with Trump, is to “build on each other’s strengths, not protect against each other” — a reference to Beijing’s concerns about U.S. containment, from existing controls on technology exports to potential restrictions on AI.

“The two sides can continue their dialogue on AI, exchange views on its risks and benefits, and jointly prevent abuse and misuse of AI,” he added.

But the summit yielded little progress on AI beyond a formal dialogue and a two-way communications channel, proposals the two leaders discussed ahead of the summit – underscoring the deep-seated mutual distrust amid contrasting visions on AI.

China’s embrace of open systems has not always been a top-down strategy on Beijing’s part.

Restrictions on access to the most advanced chips due to U.S. export controls, coupled with tighter capital markets, have pushed Chinese developers toward open models as a way to compete with major U.S. proprietary systems.

This change has proven effective.

In one year, global usage of Chinese models increased from less than 15% to more than 54% last week, driven by DeepSeek, according to AI ranking data from OpenRouter, a model marketplace.

Even U.S. companies, from Airbnb to DoorDash to Shopify, have adopted Chinese models, exploiting the benefits of open systems, including lower costs and greater customization flexibility.

Rather than choosing a side, companies – American or not – generally prefer to diversify their technology suppliers, Olander said. Chinese AI could spread across existing telecommunications, cloud and hardware networks, particularly in emerging markets where Chinese companies already have large distribution networks.

“DeepSeek or Moonshot might not make a lot of money, but the companies around them — Huawei, ZTE, the others that provide support, provide technology, provide all the services around AI — could do very well,” Olander said.

The meaning of winning the AI ​​race is also poorly defined. Different measures, from profitability to market share, could give a different result to the question.

“Is Apple winning the mobile race? Fewer people are using Apple products, but Apple is gobbling up 90% of the profits,” Olander said, adding that the likely outcome of the AI ​​race could mirror Apple’s case, where the United States holds the profits while China hoards the users.

As AI capabilities continue to advance, the differences between the most advanced models and slightly less capable systems could become less noticeable to most individuals and businesses, experts said. Some companies may need pioneering American models, while others may find Chinese models effective enough for their needs.

“For most people in the world, the question will ultimately be what kind of trade-offs they want to make,” said Chucheng Feng, founding partner of consultancy Hutong Research.

“Do I want a model that costs 10 times as much but is six months early, or a model that costs a tenth of the price but is only six months late?

However, the appeal of Chinese models could depend on the extent to which they remain accessible without restrictions, experts say.

At the China AI conference in July, Xi called for greater openness and collaboration in AI, while emphasizing the need for regulation to keep it under human control.

The better China’s models become, the less likely Beijing will leave them unrestrained, said Alex Colville, a technology and security analyst at the government-backed Australian Strategic Policy Institute.

“Beijing wants to export the country’s technology to strengthen its international influence, but national security always comes first,” he said, explaining that restrictions could be imposed around AI with Anthropic’s Mythos-like cyber capability, for example, which could pose a threat to China if misused.

“But it won’t be something as black and white as a total shutdown or a total opening. What I think we’ll see in the future is a mix of open-source and closed-source AI from China, because both serve the country’s national interests,” Colville said.

Instead of a blanket ban, Beijing could use licensing rules to restrict where certain designs could be used, he added.

Global developers and some governments have rushed to take advantage of China’s open models to promote sovereign AI development and avoid choosing sides. But they must now worry about how long these models will remain open, as China may be forced to limit the deployment of its most powerful open models to “avoid cyber anarchy”, said Reva Goujon, China director at research and consultancy Rhodium Group.

“China’s strategy is only more compelling to the world as long as the models remain available and open so that they can be adopted and refined for national and business needs,” she said.

As Chinese companies unveil increasingly efficient models that approach the performance of the most advanced systems from Anthropic and OpenAI, Washington is increasing accusations that China is engaging in what it calls “industrial-scale theft” of American trade secrets.

Anthropic and OpenAI have released reports claiming that China’s leading AI labs trained their models on results from U.S. systems, a process known as distillation. In a lengthy report released earlier this month, Anthropic detailed how DeepSeek and Moonshot passed certain customer requests to Claude to train and improve their own models and, in some processes, exposed sensitive information.

Beijing has repeatedly rejected accusations of distillation. DeepSeek and Moonshot did not respond to CNN’s request for comment.

“Given that the United States has a very leaky export control regime, including a gaping hole that allows the Chinese to access U.S. AI calculations through overseas data centers, it is very difficult to determine the extent to which Chinese AI developers rely on distillation to produce these incredibly advanced models,” Goujon said. “The question is, what is the US administration actually doing about this?

Washington has raised possible sanctions and other restrictions on open Chinese models. But Nvidia, Meta, Microsoft and dozens of other companies warned of “radical restrictions” on open models in an open letter in July – revealing a divide between US tech companies and Washington.

Despite growing calls to prevent AI development from spiraling out of control, the Trump-Xi summit produced little in the way of content, as many experts hoped.

Due to low levels of trust, cooperation between the two superpowers remains limited, said George Chen, president of digital practice at consultancy The Asia Group.

“Beijing continues to believe that Washington seeks to contain China’s rise in AI and other emerging technologies, a perception that will shape the pace and scope of the two countries’ future engagement on AI,” he said.