Nvidia and AMD have reached a deal to pay the United States government 15% of all revenue generated from high-end AI chip sales to China. This arrangement, reported by the Financial Times, grants the tech giants specific government licenses to bypass previous export restrictions and resume shipping advanced hardware to the Chinese market.
The Terms of the Export Agreement
Under this new regulatory framework, the revenue-sharing agreement applies directly to the sale of specialized hardware. Specifically, Nvidia is set to share profits from its H20 AI chips, while AMD will contribute a portion of the proceeds from its MI308 chip sales. The U.S. government has officially begun issuing the necessary licenses to facilitate these transactions.
Shifting Geopolitical Strategy
The policy landscape has been volatile. After the Trump administration restricted high-performance AI inference chips in April, the ban was briefly paused following Nvidia’s commitment to invest $500 billion in domestic U.S. data centers. By July, Nvidia moved to resume H20 chip sales—products specifically engineered to comply with earlier Biden-era export controls.
U.S. Commerce Secretary Howard Lutnick indicated that the shift in policy is tied to broader trade negotiations with Beijing, particularly regarding access to rare-earth elements essential for electric vehicle battery production and other critical components.
Industry Compliance and Growing Backlash
Nvidia has maintained a stance of strict regulatory adherence. “We follow rules the U.S. government sets for our participation in worldwide markets,” an Nvidia spokesperson stated. “While we haven’t shipped H20 to China for months, we hope export control rules will let America compete in China and worldwide.”
Despite the administration’s approval of these sales, the move faces significant opposition. Last month, a coalition of national security experts and former government officials sent a formal letter to Secretary Lutnick, urging the government to reconsider and reinstate stricter export prohibitions on high-end AI processors.