A White House official has accused Chinese artificial intelligence company Moonshot of accessing Nvidia's advanced GB300 chips despite U.S. export controls that prohibit such transfers to China [1]. Moonshot, which unveiled its Kimi K3 model last week, reportedly acquired GB300-equipped servers and accessed these chips in Thailand, likely for training its AI models, according to Michael Kratsios, director of the White House Office of Science and Technology Policy [1]. Nvidia's GB300 chips, while one generation behind the company's latest Vera Rubin systems, are still considered cutting edge and are subject to U.S. export restrictions, though some less advanced semiconductors are permitted for shipment to China [1].
The Kimi K3 model, released on Friday, is the largest open source AI model to date with 2.8 trillion parameters, and it reportedly closes the performance gap with leading frontier models such as Anthropic's Fable 5 and OpenAI's GPT 5.6 Sol, even surpassing them on some benchmarks [1]. The release of Kimi K3 comes amid intensifying competition between the U.S. and China for AI leadership, with American lawmakers considering further measures to restrict Chinese access to advanced AI hardware and to curb the adoption of Chinese AI models by Western companies [1].
The U.S. is seeking to close loopholes in export controls that allow Chinese firms to access advanced AI chips, including through remote cloud-based access. The proposed Remote Access Security Act (RASA), which passed the House of Representatives in January and is awaiting Senate action, aims to expand these controls [1]. Keegan McBride of the Tony Blair Institute for Global Change noted that American computing power remains foundational to the global AI race, with China leveraging U.S. technology both domestically and abroad [1].
While the U.S. cleared 10 Chinese firms to purchase Nvidia's H200 chips in May, a top U.S. trade official stated earlier this month that 'very few' had actually been shipped to China and Hong Kong, leading Chinese tech companies to increasingly pursue domestic alternatives [1]. Additionally, Kratsios alleged that Moonshot AI used distillation techniques to extract knowledge from Anthropic's Fable model for the development of Kimi K3, echoing Anthropic's earlier claims of 'industrial-scale' information extraction campaigns targeting its models [1].
CONCLUSION
The White House's allegations against Moonshot AI highlight ongoing tensions in the U.S.-China AI competition and the challenges of enforcing export controls on advanced technology. The situation underscores the critical role of American computing hardware in the global AI landscape and may prompt further tightening of U.S. export regulations.