In a highly unusual move that marks a new phase in the ongoing technology trade relations between the United States and China, tech giants Nvidia and AMD have reportedly agreed to a groundbreaking deal. Under this agreement, the companies will pay the U.S. government a 15% share of the revenue generated from the sale of high-end artificial intelligence chips to China. This arrangement is being made in exchange for receiving the necessary export licenses to sell these advanced semiconductors in the Chinese market.
The agreement comes after previous U.S. export controls, citing national security concerns, had halted the sale of certain advanced chips to China. Both Nvidia and AMD had subsequently developed less powerful chips specifically for the Chinese market, such as Nvidia’s H20 and AMD’s MI308, to comply with the restrictions. However, this new deal suggests a shift from outright prohibition to a revenue-sharing model, allowing the companies to re-enter a critical market while the U.S. government directly benefits financially.
This development has ignited significant debate among industry analysts and policymakers. Critics argue that this creates a “pay-to-play” precedent, effectively monetizing national security policy and setting a dangerous example. Some are concerned that the arrangement blurs the lines between trade tariffs and security measures. Conversely, supporters of the deal see it as a pragmatic solution that allows American companies to compete globally while still addressing some of the underlying security concerns. The deal is seen by some as a compromise that acknowledges the importance of the Chinese market for American tech companies.
The news has also brought renewed scrutiny to the strategic competition between the U.S. and China in the AI sector. While some believe this move could give China more access to advanced technology, others point out that the chips being sold are not the most cutting-edge models, and that the revenue-sharing model is a way to manage the flow of technology rather than completely block it.