Nvidia, AMD to Pay 15% on China AI Chip Sales in US Deal

In an unprecedented move that could reshape the landscape of international technology commerce, Nvidia Corp. and Advanced Micro Devices Inc. have agreed to remit 15% of their revenues from artificial intelligence (AI) chip sales in China directly to the U.S. government. This highly unusual arrangement is part of a deal to secure critical export licenses, a development that is likely to send ripples of concern through both American corporations and the highest echelons of Beijing.

The agreement specifically targets Nvidia’s H20 AI accelerator and AMD’s MI308 chips, two advanced components crucial for developing sophisticated AI systems. According to sources familiar with the internal deliberations, who spoke on condition of anonymity, both semiconductor giants will surrender an identical 15% share of their earnings from these specific product lines sold within the Chinese market. This effectively means a direct slice of their commercial success in a key foreign market will now flow into U.S. coffers as a prerequisite for trade.

Such a revenue-sharing model is virtually unheard of in the realm of export controls, which typically involve outright bans, licensing requirements, or strict conditions on technology transfer, but rarely a direct financial levy on sales. This innovative approach by the U.S. government suggests a deepening and evolving strategy in its efforts to manage the flow of advanced technology to China, particularly in critical sectors like AI.

For Nvidia and AMD, two of the world’s leading designers of high-performance chips, this arrangement presents a complex challenge. While securing export licenses allows them to maintain a presence in the lucrative Chinese market—a vital source of revenue and growth—the 15% cut will undoubtedly impact their profitability and financial projections. It also sets a potentially uncomfortable precedent, raising questions about whether similar demands could be levied on other U.S. technology companies operating in sensitive sectors or markets in the future. The agreement could also be perceived as a form of “tax” on their international sales, adding an unforeseen layer of complexity to their global business strategies.

From Beijing’s perspective, this deal is likely to be met with apprehension. It could be interpreted as a direct assertion of U.S. control over economic activities within China’s borders, even for products sold by American companies. Such a move might intensify China’s drive for technological self-sufficiency, accelerating its efforts to develop domestic alternatives to U.S.-made chips and reduce reliance on foreign technology. The broader context is the ongoing technological rivalry between the U.S. and China, where Washington has increasingly employed export controls to limit Beijing’s access to advanced semiconductors and equipment critical for its military modernization and AI ambitions. This new revenue-sharing model adds an unexpected dimension to that strategic competition, signaling a more assertive and financially integrated approach to tech governance.

The long-term implications of this agreement remain to be seen, but it undeniably marks a significant shift in how the U.S. government is wielding its authority over technology exports. It transforms a regulatory barrier into a direct revenue stream, potentially setting a new standard for how nations manage the economic and strategic dimensions of global technology trade.