Jensen Huang told Bloomberg he lobbied Trump to allow Nvidia chip sales in China, and told CNBC he'd be honored to join the Beijing trip. That sentence should stop you for a moment.
Huang's chips are the primary determinant of which labs can train frontier models. The H100 export restrictions that exist right now exist because policymakers decided that compute differential was a national security lever worth pulling. Whether that differential holds or collapses is, in any serious analysis, one of the three or four most consequential technical decisions of the next decade. The delegation reportedly joining Trump in Beijing includes Nvidia, Apple, Qualcomm, Blackstone, Citigroup, Boeing CEO Kelly Ortberg (who described Trump as very helpful with international sales campaigns), and others. The framing is trade diplomacy. The primary visible agenda item is aircraft orders. Ortberg is there hoping China buys 737 MAXs.
The chip export rules, the AI hardware access questions, the compute gap between U.S. and Chinese frontier labs — these are getting negotiated in that same room, as part of that same trip, by people whose primary interest is market access. That's not a criticism of Huang specifically. It's a description of the institutional structure. Semiconductor policy and aircraft deals do not belong in the same meeting. They are there because the meeting is organized around who the president brings to Beijing, not around what decisions should be made by whom.