The Trump-Xi summit on May 15 produced approval for Nvidia H200 sales to ten Chinese technology firms, with one unusual condition attached. The US Treasury will receive 25% of the revenue from the approved sales. Per coverage of the agreement, the structure was negotiated personally between President Trump and Nvidia CEO Jensen Huang, who attended the summit at a late invitation.
The specific legal mechanism has not been published. Industry counsel is treating the arrangement as a quasi-tax on individual transactions, with no sustained policy framework yet attached, but the framing is fluid.
I have watched a lot of administrations of both parties draw their own lines between government action and private commerce. I have not seen this particular line drawn quite this way before.
Set aside the politics. Set aside whether the structure is good or bad. The question I want to ask is: what would have to be true for this to become routine?
A 25% revenue cut on export-controlled chip sales is roughly the same percentage as US corporate income tax. It is more than the dividend yield of most large industrial firms. It is several multiples of the operating margin most chip designers run on individual sales. Applied to Nvidia's projected China-eligible revenue line, the structure is meaningful both as a federal revenue source and as a private cost.
If it stops at Nvidia, it is a one-off. If it spreads to other export-controlled categories, which it will because policy precedents do not stay narrow, then the United States is doing something that fifteen years ago its trade representatives would have filed WTO complaints over. The complaint would have been "your government is monetizing private export licenses as revenue." Other governments do this. China does this. Various European governments have done this in regulated industries. The United States has historically been the regime that argued against the practice and used trade leverage to discourage it elsewhere.
I have no insight into whether this structure will hold. The legal mechanism has not been disclosed and may turn out to be narrower than the press has suggested. Or it may be the start of a different commercial-industrial pattern for the US. Both possibilities are real.
What I do know is that the people running publicly-traded export-heavy businesses now have a new line item to model. And the people running commercial defense and energy export businesses are going to look at this structure and ask whether their categories are next.
A defensible reading is that 25% on China-specific chip sales is the appropriate price of access to a sensitive, dual-use market that was previously closed entirely. The chips would not have shipped at all under the prior policy. A 25% reduction in commercial value, in exchange for being allowed to ship, is still better economics than zero. Under that view, the structure is closer to a controlled-export licensing fee than a tax or revenue-share. Industry counsel can build that interpretation defensibly. I am noting the broader category-creation question; I am not assuming the worst case.
- The 25% structure was negotiated bilaterally between Trump and Huang during the summit, per CNBC
- Chinese buyers are reportedly delaying purchases due to domestic political pressure favoring Huawei
- The approved firms include Alibaba, Tencent, ByteDance, and seven others not yet publicly disclosed
For chip exporters: model the 25% scenario into your forward revenue assumptions even if it does not formalize. For policy watchers: this is the structural innovation worth paying attention to, more than the headline of "China sales approved." For everyone else: the line between private commerce and government revenue just moved, in a direction the US used to argue against.
— Hank