Nvidia’s Jensen Huang joins Trump’s Xi summit after last-minute Air Force One call

Nvidia chief executive Jensen Huang has been added to the U.S. business delegation accompanying President Donald Trump’s summit in Beijing with Chinese President Xi Jinping, a last-minute reversal of his earlier exclusion. Huang flew to Alaska to board Air Force One after a direct call from Trump overruled the original guest list. The stakes are immediate: Nvidia has spent eight months effectively shut out of selling its H200 artificial-intelligence chips into what was, before the U.S. export controls tightened, one of its largest data-center markets.

What happens around the summit table will shape costs and timelines well beyond Nvidia’s revenue line. The H200 is the central component in the AI infrastructure being built by Microsoft Azure, Google Cloud, AWS and Oracle Cloud, and the supply of that chip into China is the single most consequential lever U.S. policy has used in the current technology cold war. A loosening would re-route billions of dollars of compute capacity; a hardening would push Chinese cloud customers further toward Huawei’s Ascend silicon and the domestic alternative China has been quietly scaling.

Huang is not travelling alone. Tesla chief Elon Musk and Apple chairman-elect Tim Cook are also among the executives in the delegation, according to coverage from Euronews and Bloomberg of the manifest as it shifted in the hours before takeoff. The presence of three of the highest-profile U.S. corporate names with the most exposed China operations gives the trip the shape of an economic mission rather than a purely diplomatic one. Trump framed the request publicly in plain language: he intends to ask Xi to “open up” China to U.S. businesses, with chip exports the implicit headline.

The reason Huang was originally absent matters. Reporting from Semafor and Bloomberg indicated that Nvidia’s chief had been kept off the manifest to avoid “awkward conversations” about export controls that the U.S. government itself imposes. The reversal does not erase that architecture. The Bureau of Industry and Security at the Commerce Department, not the president alone, controls H200 licensing, and the export-control regime around advanced AI accelerators is bipartisan policy with congressional fingerprints. A summit handshake is not a license.

The skepticism layer is the part most market-day coverage is skipping. China has used the eight-month restriction to accelerate Huawei’s Ascend roadmap and to push Chinese hyperscalers — Alibaba Cloud, Tencent Cloud, ByteDance’s Volcano Engine — onto domestic silicon. Some of that compute will not migrate back to H200 even if licensing opens. Beijing also has its own price for any concession, and a chip channel can be re-opened with quotas, end-use audits or geographic restrictions tight enough that the headline win shrinks fast in practice. Nvidia’s own guidance for the quarter has already assumed continued China impairment; an unfreeze would be an upside surprise, not a baseline.

For the workers and consumers MCM’s readers ask about, the consequence is concrete. AI compute pricing is set globally, and the H200 supply line into China is the largest swing factor in 2026 capacity planning. If chips flow, hyperscaler capex stays at the elevated $300-billion run rate that has been driving both data-center construction jobs and the white-collar layoffs landing across Meta, Microsoft and Cloudflare. If chips don’t flow, the same capex re-routes into more U.S. domestic AI factories — fewer overseas customers, more domestic ones.

The summit runs Thursday and Friday. The broader tariff-reduction truce extended in November 2025 is set to expire on November 10, 2026, which is the deadline behind every conversation on Air Force One. Nvidia reports first-quarter fiscal 2027 results on May 20, the same week Meta begins its long-announced 8,000-job AI restructuring; both numbers will land into whatever the Beijing meetings decide about the H200.