Key Takeaways:
Top US AI executives are warning that China's Moonshot has matched American frontier models at a fraction of the cost, deepening a White House divide over how to respond.
The White House is divided on whether to restrict Chinese open-source AI models after top US executives warned that Moonshot's Kimi K3, a 2.8-trillion-parameter model matching frontier performance at one-third the cost, threatens American technological dominance.
"Leading closed labs want the government to eliminate their open-source competition," David Sacks, outside White House AI adviser, said on X, warning that regulatory capture would benefit OpenAI and Anthropic at the expense of broader innovation.
The Commerce Department last year considered adding multiple Chinese AI labs to its Entity List, effectively cutting off US access without a license, a source close to the administration said. The NSA and White House cyber office also weighed an advisory discouraging US companies from using Chinese models. Moonshot's Kimi K3 costs $0.95 per intelligence task versus $1.04 for OpenAI's most advanced model, according to Artificial Analysis.
A crackdown could disrupt supply chains for US companies reliant on cheaper Chinese alternatives while potentially benefiting domestic champions. Harmonic Security estimates one in 12 US and UK workers already use some form of Chinese AI, and 29 countries including Russia, Brazil and Pakistan have signed up to a China-led AI cooperation organization.
The Regulatory Arsenal
The administration has multiple tools short of an outright ban. Procurement rules, Entity List threats and public pressure campaigns aimed at US companies using Chinese models represent "what's actually happening — slower and more durable," one source familiar with government discussions said. Commerce last summer circulated draft rules leveraging supply-chain authorities to target Chinese open-source models, though administration officials keen on keeping regulation from stifling innovation killed those efforts.
Pro-competition voices have since lost key allies. Former White House adviser Sriram Krishnan has left the administration, and national security hawks have grown louder. With that personnel shift — combined with the rise of more powerful Chinese technology and fresh cybersecurity fears — ban momentum is picking up again, sources said.
China's Open-Source Gambit
Moonshot's rise reflects a deliberate strategy by Beijing. Xi Jinping said Friday that China would help developing countries scale up their AI capabilities, one day after 29 nations signed up to the China-led World Artificial Intelligence Cooperation Organisation. The approach mirrors China's successful takeover of the solar panel industry and its emerging dominance in electric vehicles, according to Richard Windsor, an independent technology analyst.
"An open weights model will be available to anyone to use and adapt, which combined with its lower cost will make it very attractive worldwide outside the US's direct sphere of influence," Ciaran Martin, former head of the UK's National Cyber Security Centre, said. "It puts pressure on the US frontier labs' business models."
The last time Chinese AI models triggered a comparable US policy response was in early 2025, when DeepSeek's R1 model matched OpenAI's performance at a reported 10 percent of the training cost. That episode led to a 17 percent single-day drop in Nvidia shares and accelerated export controls on advanced chips to China.
For US tech investors, the stakes are high. Nvidia, Microsoft and Alphabet face potential disruption to their AI supply chains and pricing power if Chinese models continue to erode the cost advantage of domestic alternatives. The uncertainty around White House policy direction adds to market nervousness, with the potential for sharp moves in AI-related stocks as the regulatory debate unfolds.
This article is for informational purposes only and does not constitute investment advice.
