Beijing is moving to gatekeep the country's most advanced AI models, threatening to upend a market where US startups have come to rely on Chinese technology that costs 60% to 90% less than leading American alternatives.
Beijing is moving to gatekeep the country's most advanced AI models, threatening to upend a market where US startups have come to rely on Chinese technology that costs 60% to 90% less than leading American alternatives.

Beijing is moving to gatekeep the country's most advanced AI models, threatening to upend a market where US startups have come to rely on Chinese technology that costs 60% to 90% less than leading American alternatives.
Chinese authorities have held meetings over the past month with Alibaba, ByteDance and Z.ai about potentially restricting overseas access to the country's top AI models, including those yet to be released, according to three people familiar with the discussions. The talks, led by the Ministry of Commerce, mark the latest step in Beijing's campaign to treat cutting-edge artificial intelligence as a critical national asset requiring export controls.
"The risk is that some firms get locked into the Chinese AI ecosystem — once you're hooked on the technology, there's a certain amount of inertia that gets created," said Joseph Spisak, vice president of product and head of open source at Reflection, an AI startup developing open-source foundational models. "It becomes harder and harder to move away."
The potential restrictions come as US companies increasingly adopt Chinese models. OpenRouter, a startup that lets American businesses route queries across different AI models, said Chinese models now handle about 30% of its traffic. Some early-stage startups have shifted all their traffic to models developed by DeepSeek, whose R1 model sparked a global price war last year. Alibaba's Qwen, ByteDance's Doubao and Z.ai's GLM-5.2 — which approaches the capabilities of leading US offerings at a fraction of the cost — are among the most widely used.
At the meetings, officials discussed making any leak or theft of proprietary AI technology an offense under China's national security law, one source said. They also raised the possibility of new measures restricting who can fund domestic AI startups. The scope of any restrictions remains under discussion and may apply only to future models, two sources said, adding that it was unclear when or if they would take effect.
The US-China AI Arms Race Intensifies
The talks mirror actions by the Trump administration, which in June ordered that foreign nationals not have access to Anthropic's most advanced Fable and Mythos models. Export controls for Fable were later lifted after new safeguards were put in place, but Mythos — designed for cybersecurity professionals — remains available only to select US organizations. Chinese authorities are deeply concerned about Mythos's potential to exploit software vulnerabilities and that Washington might deploy it against Chinese interests, two sources said.
Beijing has already taken several steps this year to protect homegrown AI. In April, China's state planner ordered Meta to unwind its $2 billion acquisition of Chinese-founded AI startup Manus. Tencent is now in talks to become Manus' largest shareholder, the Financial Times reported Thursday. In June, authorities issued sweeping new rules tightening control of overseas deals involving Chinese investors, technology and national security. China also launched investigations into Manus and other local AI startups that moved abroad.
A May roundtable of Chinese legal experts on open-source AI regulation proposed a tiered system: basic tools subject to simple filing, more advanced technologies facing security reviews, and the most sensitive frontier models barred from public release or restricted to domestic use, according to a summary published in an official Supreme People's Court journal.
What's at Stake for US Companies
If China cuts off access, the impact would ripple across the AI market. US companies that have built applications on Chinese models or grown accustomed to their cost advantages would face higher expenses and potential disruption. DeepSeek is also developing its own chip to reduce dependence on Nvidia and Huawei, according to Reuters, while Alibaba and ByteDance have their own chip initiatives.
The restrictions could provide tailwinds for American open-source alternatives. Reflection, founded in 2024 by Google DeepMind researchers and backed by 1789 Capital — where Donald Trump Jr. is a partner — has not yet released a model to the public but is valued at $25 billion. Nvidia shares, which have benefited from the AI infrastructure buildout, face an uncertain outlook if Chinese alternatives are walled off and US companies must pay more for American models.
This article is for informational purposes only and does not constitute investment advice.