To obtain approval from the United States, NVIDIA Corporation (NVDA.US) and AMD (AMD.US) will hand over 15% of their revenue from Chinese AI chips.

A U.S. official revealed on Sunday that NVIDIA Corporation (NVDA.US) and AMD (AMD.US) have agreed to pay 15% of their revenue from selling advanced computer chips to China to the U.S. government. The agreement involves advanced chips used for artificial intelligence (AI) applications, including NVIDIA Corporation's H20 and AMD MI308. The government led by U.S. President Donald Trump suspended the sales of H20 chips to China in April, but NVIDIA Corporation announced last month that the U.S. has approved the company to resume sales and they hope to start shipments soon. Another U.S. official said last Friday that the U.S. Department of Commerce has started issuing licenses for the sales of H20 chips to China. When asked if they agreed to pay 15% of their revenue from specific chip sales in China to the U.S., a spokesperson for NVIDIA Corporation stated in a declaration, "We abide by the rules set by the U.S. government for our participation in the global market." The spokesperson added, "While we haven't shipped H20 to China for several months, we hope export control rules can enable the U.S. to compete in China and globally." AMD has not responded to media requests for comment. Reportedly, these chip manufacturers have agreed to this arrangement as a condition to obtain export licenses for their semiconductor products, including AMD's MI308 chip. Reports suggest that the Trump administration has not decided how to use this money. U.S. Commerce Secretary Howard Lutnick said last month that the plan to resume sales of AI chips is part of negotiations between the U.S. and China on the rare earth issue. He stated in an interview that H20 is the "fourth best chip" of NVIDIA Corporation. Lutnick said that allowing Chinese companies to use U.S. technology is in the interest of the U.S., even if the most advanced technology is prohibited from export, these companies still continue to use the U.S. "technology stack."