When the U.S. government quietly announced that Nvidia and AMD must surrender 15% of their revenue from AI chip sales to China, the policy was framed as a strategic guardrail. A way to slow China’s access to cutting-edge computing without entirely severing trade. On the surface, it sounded like a compromise between national security and economic reality. Dig one layer deeper, and the deal looks less like policy and more like a handshake in a locked room.
Missing Details
It was announced that Nvidia and AMD agreed to pay the US government 15 % of their sales of advanced artificial-intelligence chips to China. But why on earth? Specifically, that percentage applies to chips like Nvidia’s H20 and AMD’s MI308, as a condition for getting export licenses under the Trump administration. An unprecedented move in export-control history.
The strangest part isn’t the number. It’s that no one outside a few offices in Washington and the corporate boardrooms of Nvidia and AMD knows exactly how it works. When is the 15 % due? Is it on delivery, quarterly revenue, or after audited sales? Is it gross revenue (before expenses) or net (after deductions)?
Then there’s the billion-dollar question: what does the U.S. government do with this money? Is it funneled into general revenue, defense R&D, AI safety research or just swallowed by the sprawling federal budget? If the payment is truly meant to offset national-security risks, there should be a visible, legislated link to national-security initiatives. Without it, the move risks looking like a revenue grab wrapped in security rhetoric.
Consumers in the Crosshairs
If Nvidia and AMD decide to eat the cost, their margins take a hit. This can affect jobs, R&D spending, and competitiveness. If they pass the cost along, the price bump lands on customers. That doesn’t just mean buyers in China; global chip pricing could shift, affecting U.S. gamers, researchers, and AI startups. Right now, the deal includes no consumer protections to prevent cost pass-through at home. An odd omission for a policy claimed to serve the national interest.
Accountability: Zero Point Zero
This arrangement was hashed out behind closed doors, with no public debate, no congressional oversight, and no requirement for public reporting on how funds are collected or spent. As a precedent, it’s alarming. If this model stands, future administrations could apply similar “security fees” to any industry they choose, bypassing the legislative process entirely. That’s not export control. That’s a quiet new revenue stream, shielded from scrutiny.
China’s Strategic Position
We relied on materials from China, and perhaps other places, to make these chips. China has smartly positioned itself as a key supplier of those materials. Now, companies get slapped with tariffs on inputs and must surrender 15 % of sales to the U.S. That’s a double whammy.
Bigger Than One Deal
The lack of clarity here isn’t just a bookkeeping problem. It reflects a broader trend in U.S.–China tech competition: policy by improvisation. Instead of a clear, long-term export-control framework, the U.S. is layering ad hoc measures on top of one another. Each negotiated in private, each harder for the public to track. The danger isn’t just that these policies might fail. It’s that they might succeed in reshaping the global tech economy without anyone realizing what happened until it’s too late.
In an era when AI chips are as strategically important as oil, the public deserves more than vague assurances and withheld details. National security doesn’t have to mean national secrecy. If the U.S. is going to claim 15% of a major industry’s foreign sales in the name of security, it owes the public a clear answer to the simplest questions: how much, when, for what purpose and who’s watching.
Because if no one’s watching, history suggests someone, somewhere, is cashing in.
