Huang Exit Puts $3.2B of China AI Chip Projects Under Scrutiny

Computing projects worth more than 23 billion yuan ($3.2 billion) in China are built around Nvidia processors the company says it can no longer legally sell there, according to a Reuters review of more than 150 procurement notices, investment plans, construction contracts and corporate disclosures published November 13, 2026. The findings landed one day after Jensen Huang resigned as Nvidia's chief executive.

The gap is stark. Nvidia has said U.S. export controls shut it out of China's AI data-center market — Huang put it in October 2025: "We are 100% out of China." Yet local governments, state-owned engineering groups and developers kept writing H100 and H200 chips into specifications.

One caution up front. The reporting establishes exposure and heightened scrutiny, not outcomes. No specific project is confirmed cancelled, frozen or blocked, and nothing shows Huang's departure changed a single contract.

What did the review actually count?

Reuters reviewed more than 150 public documents — government procurement notices, corporate investment plans, construction contracts and disclosures from listed companies. The combined value exceeds 23 billion yuan; a Reuters factbox rounds it to "more than $3 billion."

The method has limits. It captures only published plans, so the $3.2 billion is a floor for documented intent, not a count of installed chips. A procurement notice describes what a developer wants to buy. It doesn't prove delivery, and several projects are still seeking partners or financing.

The three named projects

• Xianyou county, Fujian: a proposed 6.2 billion yuan ($873 million) AI computing center specifying H100 chips, still seeking commercial partners. A partner must price three risks at once: sourcing, enforcement attention, and tightening scrutiny.
• Shaanxi: a 3 billion yuan computing center specifying H100 and H200, where state-owned engineering group Sinomach said a subsidiary won a related contract. The H200 reference signals planning around the newer, more capable part. Public disclosure of an Nvidia-dependent plan makes it easy to cite in congressional hearings.
• Jilin: a 2.26 billion yuan project described in an April government prospectus combining H100 chips with Chinese-developed chips. The developer reportedly expects bank loans to fund about 80% of the investment — a single-source detail. Hybrid design suggests planners knew Nvidia supply was uncertain months before Huang's exit.

The three projects sum to roughly 11.5 billion yuan, about half the total — this publication's arithmetic, not a Reuters figure.

What does the Huang resignation change?

Reuters reported Huang resigned Thursday, November 12, 2026, with CFO Colette Kress becoming interim CEO. The company said: "The NVIDIA Board is initiating an independent investigation in light of recent reporting that founder and CEO Jensen Huang has been the target of a foreign intelligence effort."

Huang was Nvidia's most visible advocate for Chinese market access. Reuters framed his loss as a potential blow to Nvidia's influence over export policy — a reasonable concern, not an established fact. An interim CEO, a board probe and a skeptical Congress make a rapid policy shift toward Nvidia less likely.

A second thread: Reuters identified Alice Huang, a former executive at Singapore-based Nvidia cloud partner Megaspeed, whom U.S. authorities reportedly suspect of intelligence ties to Beijing. That's a reported suspicion, not a charge or finding.

Chinese firms look overseas

Chinese AI companies are seeking to rent Nvidia-powered capacity at data centers outside China, Reuters separately reported, fearing tighter U.S. limits on remote access. A major Chinese cloud provider has discussed securing overseas capacity; others have contacted operators to reserve access. The hedge has a flaw: reservations only work if the rules permitting them survive.

U.S. lawmakers from both parties are pushing for tighter China chip restrictions. The project list gives them concrete material — provinces, chip models, yuan figures — likely to surface in letters to the Commerce Department.

Markets moved fast. Chinese AI developer and U.S.-linked hardware supplier shares fell Friday, November 13, on fears of a deeper technological divide.

Huang's own words

In May 2024: "If we completely leave the market, someone else will definitely step in." In May 2025: "If we leave a market altogether, there's no question somebody else would step in." And in a Fortune interview: "I think the opportunity in China is large and growing." Beside "100% out of China," the statements describe the squeeze — a large market Nvidia says it can't serve, while Chinese developers keep specifying its chips.

The sequence compressed fast: an April prospectus, a November 11 board meeting, the November 12 resignation, the November 13 review and sell-off, and an earnings call scheduled November 17. A governance story became a supply-chain story in under 72 hours.

The takeaway: $3.2 billion measures documented ambition. The risk is policy and financing — and coming filings, hearings and earnings commentary will show how much materializes.