China’s People’s Liberation Army has been aggressively courting Nvidia for its most advanced semiconductors as it builds out artificial intelligence-enabled weapons systems, according to a congressional advisory panel, which warned Wednesday that Beijing is working to sidestep American export controls to secure the technology.
The U.S.-China Economic and Security Review Commission, in its annual report to Congress, said domestic Chinese processors still fall short of the leading American designs, leaving the military dependent on foreign chips for its modernization push. The panel described Beijing’s objective as “intelligentized warfare,” the PLA’s term for embedding AI across weapons, command and control, kill chains and logistics.
“China is seeking to evade U.S. export controls to gain the most advanced semiconductors that underpin its intelligentization drive,” the commission wrote.
The report singled out Huawei’s Ascend line and Cambricon’s Siyuan family as the strongest homegrown alternatives, but concluded both remain behind products from Nvidia and Advanced Micro Devices. It also flagged Nvidia’s CUDA software ecosystem as a barrier Chinese firms have yet to overcome, alongside gaps in leading-edge logic fabrication, memory and chipmaking equipment.
“The resulting gap in compute is substantial, largely due to U.S. export controls on advanced chips,” the report said.
According to the panel, Chinese entities have been reaching Nvidia hardware indirectly by tapping data centers in third countries, including Malaysia. On Tuesday, the Commerce Department’s Bureau of Industry and Security issued a rule aimed at closing that route. The action followed criticism from Treasury Secretary Scott Bessent and lawmakers who argued the department had not moved aggressively enough against diversion of restricted chips.
The commission was established by Congress in 2000 to assess the national security consequences of the U.S.-China trade and economic relationship.
Wednesday’s findings intensify scrutiny of Nvidia’s China exposure at a delicate moment for the company. Former chief executive Jensen Huang stepped down last week over extramarital affairs, one of them with a former executive at Singapore-based Megaspeed, a business partner that U.S. authorities suspect of spying for China. Investigators are also examining whether Megaspeed helped divert Nvidia chips to Chinese buyers.
Nvidia has maintained that U.S. export restrictions have effectively locked it out of China’s AI data center market. Even so, a Reuters review identified more than $3 billion in Chinese computing projects running on its restricted hardware, a figure that undercuts the company’s position and gives lawmakers fresh material.
Washington moved again on Monday, announcing tighter limits on China’s access to advanced chips from Nvidia and other suppliers. Separately, the Trump administration is weighing a requirement that American companies publicly disclose their use of Chinese AI models, a step designed to discourage reliance on such technology without imposing a formal ban that could provoke Beijing, according to a Reuters exclusive published Wednesday.
The same outlet reported earlier in the week that Chinese organizations are deploying increasingly sophisticated techniques to train their own models on outputs from top American AI systems.
The policy backdrop has grown more crowded in recent days:
DateActionMondayWashington announces further restrictions on China’s access to advanced chips, including Nvidia’sTuesdayCommerce Department’s BIS issues rule targeting access to Nvidia chips through third-country data centersWednesdayCongressional commission releases annual report on China’s military AI ambitionsWednesdayReport emerges that the Trump administration may require disclosure of U.S. firms’ use of Chinese AI models
Note: Dates are as described in source reporting.
For Nvidia, the report lands as the company tries to defend a China business already squeezed by years of tightening controls. Each new allegation of diverted hardware raises the risk of broader licensing conditions or congressional action, and the leadership vacuum left by Huang’s exit removes a prominent voice from the debate over how far restrictions should go.
For chipmakers more broadly, the panel’s assessment that Chinese substitutes remain a step behind reinforces the argument that export limits still carry leverage, while the third-country loophole it describes suggests enforcement, rather than rule-writing, may be the harder problem. Chinese AI developers, meanwhile, face a compute shortfall the report expects to persist, pushing them toward workarounds that U.S. regulators are now explicitly targeting.
The commission’s findings are advisory and do not carry the force of law, but its annual reports have historically shaped legislation on Capitol Hill. With the Commerce Department’s new rule already in place and disclosure requirements under consideration, the report gives lawmakers a consolidated case for tightening the perimeter around advanced AI hardware.