Nvidia and AMD have reached a unique agreement with the U.S. government: in exchange for export licenses to sell AI chips to China, both companies will pay 15% of their China chip sales revenue to the U.S. Treasury. The arrangement was concluded after Nvidia CEO Jensen Huang met with President Trump last week and lifts months of restrictions that had limited access to the Chinese market.
The deal represents a significant change in U.S. export policy. Historically, export controls were rooted in national security considerations and sanctions, not structured as revenue-sharing arrangements. By tying export permission to a percentage of sales, the administration has created a commercial mechanism that allows companies to regain market access while providing the government with a direct financial benefit.
For Nvidia, the commercial consequences of prior restrictions were substantial: company filings indicated that being barred from parts of the Chinese market contributed to a revenue shortfall—estimated at about $4.5 billion in a single quarter. Restoring the ability to sell advanced AI chips in China can therefore have immediate and meaningful effects on revenue and product deployment for both firms.
Supporters of the arrangement note that it provides a pragmatic path to balance economic interests and strategic concerns. It lets U.S. chipmakers continue to compete globally while the government retains leverage over sensitive technologies. Enabling sales under a licensing framework can also give regulators continued oversight of exports and supply chains.
Critics, however, warn about the broader implications. Turning export decisions into a revenue stream may set a precedent that changes how national security and trade policy are handled. Observers worry that monetizing access could weaken nonproliferation and security objectives over time, and that it could encourage other governments to demand similar payments for market access to strategic sectors. Analysts also note potential legal and ethical questions around charging companies for permission to conduct normal commerce.
Another concern is the message this sends in the context of U.S.-China competition in artificial intelligence. Allowing a steady flow of advanced chips into China, even under license, could accelerate capabilities there. Policymakers must weigh the immediate economic benefits to U.S. firms against long-term strategic risks and consider whether licensing terms include robust safeguards to prevent misuse.
The mechanics of the agreement, as described publicly, require the companies to report sales in China and remit 15% of those revenues to the U.S. government. Details about compliance measures, auditing, and enforcement have not been fully disclosed, and industry watchers expect further clarification on how the program will be administered. Questions remain about which specific chip models and customers qualify for licenses, and how the policy will be adjusted as technology and geopolitical conditions evolve.
For shareholders and customers, the deal is likely to be interpreted in pragmatic terms: it reopens major revenue opportunities and reduces uncertainty around China sales. For regulators and strategists, it introduces a new tool in the export-control toolkit—one that mixes commercial incentives with security oversight. Whether this becomes a one-off solution or a model for future export policy will depend on its outcomes, including the effectiveness of safeguards and the reaction from international partners.
Ultimately, the agreement between Nvidia, AMD, and the U.S. government marks a notable shift in the intersection of commerce and national policy. It underscores the difficult trade-offs governments face when deciding how to protect sensitive technologies while supporting domestic industry. As the policy is implemented, observers across business and government will be watching closely for its economic impact and strategic consequences.