Reuters Survey: Over 23 Billion Yuan in Chinese AI Projects Rely on Nvidia's Restricted Chips
nashnova research
A Reuters review of over 150 documents found Chinese AI projects tied to restricted Nvidia processors total more than RMB 23 billion ($3.2 bn); released the day after Jensen Huang stepped down as CEO, the investigation sharply raises the supply-chain stakes for every one of those projects.
What does RMB 23 billion in projects actually look like?
Reuters reviewed over 150 procurement notices, investment plans, construction contracts, and corporate disclosures covering planned investments, signed contracts, and completed facilities.
The projects stretch from eastern coastal tech hubs to remote western provinces. Participants include local-government-backed firms, universities, and regional computing centers — not just big tech.
The chips involved are Nvidia's H100, H200, and H800 — all subject to U.S. export controls.
This means → demand for restricted chips has not stayed confined to a few headline players; it has seeped into the capillaries of China's AI infrastructure.
How much of the RMB 23 billion is actual chip spending?
Reuters explicitly notes the RMB 23 billion includes buildings, power systems, and other infrastructure — it is not a pure chip-procurement figure.
Some facilities are already operational or under construction; others — such as a Gansu site expected to install thousands of Nvidia chips this year — have not confirmed whether the processors have actually arrived.
In plain terms = this number is the total price tag of "projects built around Nvidia chips," not the amount China spent buying chips directly.
Is Nvidia still in the Chinese market or not?
In an August 2026 regulatory filing Nvidia said it has been effectively shut out of China's data-center computing market.
Yet an IDC report shows Nvidia still held 34% of China's AI-accelerator shipments in H1 2026.
Reuters notes the two data sets use different methodologies and cannot, on their own, reveal how restricted hardware is actually flowing.
This means → "shut out" is Nvidia's official position, but market data shows restricted chips still circulating — the grey zone between the two is precisely where the controversy sits.
Why does Huang's departure make these projects more vulnerable?
Jensen Huang had been an active advocate for relaxing export controls; his exit is seen as weakening Nvidia's lobbying leverage in Washington.
Congress is pushing stricter export-control legislation. Huang had been the loudest voice resisting that push.
Jamestown Foundation analyst Sunny Cheung said: "Washington needs to stop focusing on who is buying chips on the surface and start examining who is ultimately using the compute — and for what purpose."
In plain terms = the person who could "take bullets" for Nvidia in Washington is gone; Congress now faces one fewer major obstacle to tightening controls.
What does the scandal add to the picture?
After Huang's resignation a scandal continued to build around his ties to a Chinese businesswoman suspected by U.S. authorities of intelligence links to Beijing.
The figure involved is Alice Huang (黄慧文), connected to Nvidia's Singapore cloud partner Megaspeed. Megaspeed denies illegally rerouting Nvidia chips to China.
Nvidia's board has launched an independent investigation.
This reflects a shift: the chip-control debate has spread from policy into corporate governance — Nvidia itself is now under scrutiny.
What does this mean for Chinese projects still betting on Nvidia chips?
For Chinese projects still seeking restricted Nvidia processors, the risks of supply disruption, delivery delays, or a forced pivot to domestic chips are rising.
Neither China's Ministry of Commerce nor Nvidia responded to Reuters' requests for comment.
This means → the RMB 23 billion project cluster faces more than a "can we still buy them" question — the entire supply logic may be rewritten.
市场有风险，内容仅供研究参考，不构成投资建议。