Estimated reading time: 6 minutes
Key Takeaways
- Nvidia and AMD will remit 15% of all China AI-chip revenue to the U.S. Treasury as part of a newly struck accord.
- The deal unlocks export licences for modified processors such as Nvidia H20, keeping both companies in a market worth an estimated $25 billion.
- Washington shifts from blanket bans to a revenue-based control that it can dial up or down as tensions evolve.
- Margins will tighten, but executives expect volume growth to soften the blow.
- Policy watchers say the move could become a template for other sensitive technologies amid the wider U.S.–China rivalry.
Table of contents
Introduction
In a surprise compromise first revealed by a Financial Times report, Nvidia and AMD have agreed to channel 15 per cent of the revenue from China-bound artificial-intelligence chips to Washington. The pact lets the two U.S. giants resume shipments previously threatened by export controls, striking a delicate balance between national security and commercial imperatives.
“We needed a solution that safeguards critical technology yet recognises the realities of global supply chains,” a senior U.S. official told reporters.
Why the Levy Matters
AI accelerators have become the fulcrum of the technological rivalry between Washington and Beijing. By tying sales to a revenue tithe, U.S. policymakers retain visibility into chip flows while avoiding a total cutoff that could push Chinese clients toward fully domestic solutions.
- Security lens: The U.S. Commerce Department gains real-time insight into quantities, specs and end-users.
- Commercial lens: Nvidia and AMD keep a foothold in the world’s fastest-growing AI market without breaching sanctions.
- Policy lens: The model could spill over into quantum, biotech and other sensitive arenas.
Inside the 15% Revenue-Share Deal
Under the arrangement, every licence issued for an eligible AI processor will carry a payment schedule: 15% of the gross invoice value remitted quarterly to the U.S. Treasury. In exchange, export paperwork is fast-tracked for chips that conform to predefined performance ceilings.
- Applies to tailored parts such as Nvidia H20 and future iterations of AMD MI300.
- Ceilings restrict interconnect bandwidth to prevent easy clustering into supercomputers.
- Audit rights allow U.S. inspectors to trace shipments down to individual serial numbers.
Reactions on Both Sides of the Pacific
Beijing’s foreign-ministry spokesperson criticised the move as “economic coercion,” yet industry insiders in Shenzhen privately welcomed the clarity. Meanwhile, analysts at the Semiconductor Industry Association called the policy “a pragmatic middle road.”
“It’s not ideal for either side, but it avoids an outright tech divorce,” noted a researcher at the Center for Strategic and International Studies.
Financial Impact on Nvidia & AMD
Equity analysts estimate a margin hit of 150–250 basis points once the levy is in full swing. Yet pent-up demand from Chinese hyperscalers could lift shipment volumes enough to neutralise, or even outweigh, the lost margin dollars.
- Nvidia: Each percentage point of China revenue equals roughly $800 million in annual sales.
- AMD: China accounts for about 22 per cent of its data-center segment.
- Both firms reiterated full-year guidance but flagged “policy fluidity” as a risk factor.
Implications for the Global Chip Race
The levy signals a tactical evolution in U.S. export control strategy. Rather than erecting absolute barriers, Washington now aims to shape incentives—letting American firms compete while ensuring the state captures economic rent and maintains oversight.
For rivals in South Korea and Europe, the message is clear: compliance pathways exist, but they come with strings attached. Some observers predict similar revenue-share models for advanced lithography, sensor arrays and battery chemistries.
Outlook
The next six months will reveal whether the compromise truly cools bilateral tensions or merely pauses them. Nvidia and AMD must prove they can satisfy regulators while delivering for shareholders—a tightrope walk that could define the next era of semiconductor geopolitics.
Bottom line: Expect continued redesigns of high-end chips, iterative policy tweaks and a persistent tug-of-war between security objectives and market realities.
FAQs
Why did the U.S. choose a revenue-sharing model instead of an outright ban?
Officials believe tying sales to a levy offers visibility and leverage without forcing U.S. companies out of a lucrative market—reducing incentives for Chinese firms to accelerate full decoupling.
Does the 15% payment apply to all Nvidia and AMD products?
No. It covers only AI accelerators that fall below specific performance thresholds and receive an export licence. Consumer GPUs and older enterprise parts remain unaffected.
Could the percentage change in the future?
Yes. The Commerce Department purposely framed the levy as adjustable, giving policymakers room to tighten or loosen the terms as geopolitical dynamics evolve.
How will the money be used by Washington?
According to officials, funds will flow into a technology-security trust earmarked for AI research, export-control enforcement and workforce development initiatives.
What happens if Nvidia or AMD miss a payment?
Missing or under-remitting triggers automatic licence suspension and potential civil penalties, effectively shutting the door on further China exports until compliance is restored.
