In an arrangement with the U.S. government, semiconductor leaders Nvidia and Advanced Micro Devices (AMD) have agreed to provide 15% of their revenues from certain chip sales in China in exchange for new export licenses. The deal grants the companies permission to sell their advanced processors—Nvidia’s H20 and AMD’s MI308—into a market that has faced strict U.S. trade restrictions due to national security concerns.
This agreement represents a notable departure from traditional export control policy. Historically, U.S. measures on sensitive technology have focused on blocking or limiting sales outright, rather than allowing them under revenue-sharing terms. By collecting a portion of the earnings from these sales, the government is introducing a new model that blends oversight with economic participation.
The chips involved in the deal, although modified to comply with U.S. export rules, remain highly capable in artificial intelligence, machine learning, and advanced data processing. These capabilities make them attractive to Chinese technology companies, which continue to expand their demand for high-performance computing hardware. The products were previously restricted under measures intended to limit the potential use of such technology in military or surveillance programs.
Observers say the arrangement appears to strike a strategic balance—allowing U.S. companies to maintain a foothold in a valuable market while still imposing conditions designed to protect national interests. China remains one of the most lucrative destinations for advanced processors, and the new licenses position Nvidia and AMD to compete more directly in a market they were previously unable to access.
Not everyone sees the deal as purely positive. Some policy analysts warn that tying export permissions to government revenue could create a precedent where national security restrictions are viewed more as negotiable financial terms than as safeguards. Questions remain about whether such agreements might dilute the original intent of technology controls if revenue generation becomes a larger priority than security enforcement.
Nvidia has stated in the past that it complies fully with all U.S. export laws and licensing requirements, while AMD has not publicly commented on the specifics of this arrangement. The U.S. government has also not released a detailed public statement outlining the full scope of the revenue-sharing program or how the funds will be allocated domestically.
The deal comes at a time when tensions over technology access between the U.S. and China remain high. In recent years, Washington has enacted a series of rules aimed at curbing China’s ability to develop advanced AI and supercomputing systems. This agreement signals a shift toward a more flexible, though still tightly managed, export strategy that could influence future trade policies for sensitive technologies.
Whether similar revenue-sharing arrangements will be applied to other sectors of the tech industry remains to be seen. For now, the partnership between the U.S. government, Nvidia, and AMD marks a significant evolution in how economic opportunity and national security are being balanced in the global technology market.