Nvidia and AMD to Pay 15% of China AI Chip Sales to US Government

Aug 12, 2025
Nvidia and AMD have agreed to pay the US government 15% of their revenue from AI chip sales to China, following a new licensing agreement.

NVIDIA and Advanced Micro Devices, Inc. (AMD) have entered into an unusual agreement with the US government, agreeing to pay 15% of their revenue from AI chip sales to China. This arrangement follows a meeting between Nvidia's CEO Jensen Huang and President Trump at the White House, where the terms were discussed.

The agreement allows both companies to resume sales of certain AI chips to China, which had been previously restricted due to national security concerns. The US Commerce Department began granting licenses for these sales shortly after the meeting, effectively making the federal government a partner in the companies' Chinese business operations.

The deal is expected to generate over $2 billion for the US government, with Nvidia projected to sell more than $15 billion worth of its H20 chips and AMD expected to sell $800 million worth of its MI308 chips to China by the end of the year. This move marks a significant shift in US trade policy, as it involves a direct financial stake in the export activities of American semiconductor companies.

While the arrangement has been met with criticism from national security experts, who argue that it could incentivize China to seek similar deals for other technologies, the Trump administration has defended the decision as a strategic move to maintain US competitiveness in the global AI market.

We hope you enjoyed this article

Consider subscribing to one of our newsletters like AI Policy Brief, Silicon Brief or Daily AI Brief.

Free newsletter

AI Policy Brief

Weekly report on AI regulations, safety standards, government policies, and compliance requirements worldwide.

Whitepaper

Tensordyne Napier: What If One Rack Could Do the Work of Nine?

Tensordyne

This Tensordyne whitepaper presents Napier, an inference-focused AI processor and rack-scale system based on the company’s TDN Math logarithmic number system. It examines infrastructure requirements for large mixture-of-experts and agentic models, compares major inference architecture approaches, and details the TDN AIP processor, TDN72 pod, TDN Link fabric, and Napier Ultra configuration. The paper reports simulation-based performance, cost, and accuracy-validation results, including Tensordyne’s projected comparison of one Napier rack with a nine-rack Nvidia Rubin plus Groq deployment; the chip is reported as taped out and in fabrication.

Read more
Free, six days a week

Daily AI Brief: the AI news that matters, in your inbox.