The U.S. government has introduced a new fee requiring Nvidia and AMD to pay 15% of their revenue from specific AI chip sales to China. This revenue-sharing arrangement, unprecedented in U.S. export policy, is tied to export licenses for select high-performance chips used in artificial intelligence applications.
This measure, confirmed by multiple sources including the Financial Times, applies specifically to Nvidia’s H20 chip and AMD’s MI308. Both chips play a critical role in powering AI systems and are central to both companies’ ambitions in the Chinese market.
Under this new framework, Nvidia and AMD must pay the U.S. government a fee equivalent to 15% of the revenue from sales of these chips to China as a condition for receiving the necessary export licenses. This policy represents a significant shift from previous export controls, where fees of this nature were not standard practice.
A U.S. government official and industry insiders revealed that this fee structure was agreed upon by the two companies the previous week. However, details on how the U.S. government plans to utilize the billions of dollars expected from these fees remain unclear.
Nvidia responded by stating that they comply with U.S. regulations governing global market participation and emphasized that they have not shipped the H20 chip to China for several months. The company expressed hope that export control rules will continue enabling American businesses to compete both in China and globally.
This fee imposition emerges amid ongoing geopolitical tensions surrounding technology and trade between the U.S. and China. Experts suggest this move not only aims to generate revenue but also seeks to maintain a strategic advantage in advanced AI chip technology, representing concerns over technology transfer and national security.
The response in China has so far included warnings from state media urging Chinese companies to avoid purchasing these U.S.-made AI chips. The rationale includes concerns about energy efficiency, security risks, and the push for developing indigenous chip technologies. Chinese firms are being encouraged to build secure, controllable hardware and software ecosystems rather than relying on foreign components subject to U.S. restrictions.
Industry analysts note that this new fee could affect Nvidia’s profits, especially considering the company’s recent efforts to expand AI chip sales to China. The long-term impact on the global semiconductor market and the competitive dynamics between U.S. and Chinese chipmakers will be important to monitor.
This development marks a novel and assertive step by the U.S. to control advanced technology exports to China through direct financial levies, reflecting the evolving complexities of international trade and technology diplomacy.