Go-Ahead for AI Chip Sales to 10 Chinese Firms Raise Alarms

Artificial Intelligence & Machine Learning , Next-Generation Technologies & Secure Development , Standards, Regulations & Compliance Go-Ahead for AI Chip Sales to 10 Chinese Firms Raise Alarms Reports: Trump Administration Approval of Nvidia H200 Sales Poses Frontier AI Risks Chris Riotta (@chrisriotta) • May 14, 2026     Credit Eligible Get Permission Trump administration discussions on AI governance with China are colliding with reports that Washington may permit expanded Nvidia H200 chip sales to Chinese firms. (Image: Shutterstock) The Trump administration's apparent willingness to permit expanded access to Nvidia's H200 chips for China while opening new artificial intelligence talks with Beijing is fueling concerns in Washington that the move could accelerate China's frontier AI ambitions and pose security risks. See Also: Securing AI Workloads With Ubuntu Pro AI was a key topic at President Donald Trump's summit this week with Chinese President Xi Jinping, with promises of future AI governance discussions between the world leaders. But on the sidelines of the event, the administration's apparent clearance of advanced AI hardware to China prompted debate over semiconductor access, export controls and strategic competition. Reports surfaced indicating that the U.S. has cleared roughly 10 Chinese firms to buy Nvidia's H200 chips, though no deliveries have been made as the administration weighs competing national security and economic priorities surrounding AI competition with Beijing. Nvidia CEO Jensen Huang joined Trump's trip to Beijing after receiving a personal invitation from the president, with analysts noting how the move reflected the increasingly critical role advanced semiconductors and frontier AI infrastructure now plays in U.S.-China relations (see: Nvidia CEO Huang Warns Export Bans Empower Chinese AI Firms). Reports from the summit focused heavily on trade tensions, economic cooperation and regional security issues, with little direct mention of cybersecurity or AI. Some Chinese state-linked summaries appeared to place greater emphasis on technology coordination and AI governance discussions than the White House's public statements. Despite the differing readouts, Treasury Secretary Scott Bessent said on the sidelines of the summit that the world's two leading AI powers would begin talks around AI governance and "best practices" aimed at preventing advanced models from falling into the hands of non-state actors. "We're going to be discussing AI guardrails with the Chinese," Bessent said during an interview with CNBC. "The two AI superpowers are going to start talking. We're going to set up a protocol in terms of how do we go forward with best practices for AI to make sure non-state actors don't get a hold of these models." Bessent also said he expects a major "step-function jump" in upcoming large language model releases from Google’s Gemini and OpenAI. The Treasury secretary separately said he had no knowledge of the H200 approvals, and the White House did not immediately respond to a request for comment. Decisions around chip export licenses typically fall under the Commerce Department. Experts told ISMG the split between the AI guardrails discussion and the H200 approvals highlights the central tension now facing Washington: whether the U.S. can pursue limited AI risk-reduction talks with Beijing while still restricting the compute infrastructure that could help Chinese firms narrow the frontier model gap. "The U.S. shouldn't trade export controls for dialogue," Leah Siskind, an AI research fellow at the Foundation for Defense of Democracies, told ISMG. "Export controls are the single most effective constraint on PRC AI development - Chinese labs and officials say so themselves." The warning comes as China-focused policy analysts increasingly adopt the view that AI diplomacy with Beijing may only be useful if it doesn't weaken the ongoing pressure campaign around chips and model security. Siskind noted how U.S. frontier labs operate under more developed safety and security regimes than Chinese labs, including third-party red teaming, responsible scaling policies and government engagement. "Chinese labs aren't replicating it," she added. China's frontier AI ecosystem continues advancing rapidly through state-backed industrial policy, aggressive talent recruitment and alleged large-scale distillation of American AI models. Analysts say Beijing has shown little indication it is prepared to meaningfully address concerns surrounding model replication, intellectual property extraction and state-linked AI development practices. The talent competition has also sharpened concerns over how quickly Chinese firms can absorb expertise from leading American AI labs. Recent examples include Wu Yonghui, a former Google vice president of research who helped develop Gemini and now leads research for ByteDance's AI arm, and Yao Shunyu, a former OpenAI researcher who was named Tencent Holdings' chief AI scientist. Those moves have added to concerns in Washington that Beijing's commercial AI gains could feed directly into military and intelligence priorities under China's military-civil fusion strategy. Security analysts and former federal cyber officials told ISMG the AI dialogue surrounding Trump and Xi's summit highlights interest in "constructive competition" between the United States and China. But sooner or later, the approach could run into the unsettled question of Taiwan, according to experts. "Safeguarding peace and stability across the Taiwan Strait is the biggest common denominator between China and the U.S.," Xi reportedly told Trump. A White House statement about the discussions did not mention Taiwan. The Commerce Department did not respond to a request for comment.