Reuters: Chinese AI projects worth over $3.2 billion built around restricted Nvidia chips

A Reuters investigation has identified more than $3.2 billion in Chinese computing projects built around Nvidia chips subject to US export restrictions. The review, published this Friday (13), examined more than 150 Chinese procurement notices, investment plans, construction contracts and corporate disclosures, totalling more than 23 billion yuan in projects and equipment agreements.

The projects stretch from coastal technology hubs to remote western provinces and show how deeply Nvidia technology remains at the centre of China's AI expansion, even after the company said export controls had effectively shut it out of China's AI data-centre market.

The projects identified

  • Fujian (Xianyou): a proposed 6.2 billion yuan ($873 million) AI computing centre seeking commercial partners and specifying Nvidia H100 chips.
  • Shaanxi: a subsidiary of state-owned engineering group Sinomach won a contract in a 3 billion yuan project specifying H100 and H200 processors.
  • Jilin: an April government prospectus describes a 2.26 billion yuan project combining H100 processors with Chinese-developed chips; about 80% of the investment is expected to come from bank loans.
  • Inner Mongolia: a 1.5 billion yuan project proposes installing 625 servers with H200 chips; construction of two buildings in its first phase has begun, according to local industry reporting.
  • Gansu: a 1.16 billion yuan construction contract signed in July for a centre planning to install about 3,072 Nvidia processors, according to tender documents specifying H100 and H200 hardware.
  • Jiangsu (Wuxi): a cluster using H800 processors entered operation in March 2025, according to the municipal government.
  • Universities: Shenzhen University awarded a contract for three H100 processors in September, while US-blacklisted Northwestern Polytechnical University bought two H200 chips the following month.

The context: Jensen Huang's resignation and pressure in Washington

The findings come a day after Nvidia co-founder Jensen Huang, a vocal advocate of easing export restrictions, resigned as chief executive. According to Reuters, his departure could deepen the vulnerability of Chinese companies exposed to delays and supply disruptions, by weakening Nvidia's lobbying leverage with US lawmakers. In the US Congress, consensus is growing for new restrictions on exports of advanced chips to China, according to another Reuters report.

Nvidia said in an August regulatory filing that it was effectively excluded from China's data-centre computing market, and Huang said last year that the company's market share in China had fallen from 95% to zero. Even so, an IDC report cited by Reuters in October estimated that Nvidia still accounted for 34% of Chinese AI-accelerator shipments in the first half of 2026.

What is still unproven

Reuters cautions that the figures include planned investments, signed contracts and completed installations, and cover buildings, power systems and other infrastructure — not chips alone. The numbers also do not establish how the restricted hardware reached Chinese users. In the Gansu case, the documents do not confirm whether the processors were delivered. The agency also states that the projects identified have not been linked to Megaspeed or the allegations involving Huang: the company has denied illegally diverting chips, and Nvidia's board launched an independent investigation.

Sunny Cheung, an analyst at the Washington-based Jamestown Foundation, told Reuters: “Washington must stop focusing solely on who buys the chips on the surface and start examining who ultimately uses the computing power — and for what purpose.”

Also on Geekiki: Malaysian PM says there is no indication AI chips were exported in the Nvidia-Megaspeed case and Jensen Huang resigns as Nvidia CEO; Colette Kress takes over as interim chief.

Sources: Reuters, via WIXX; Reuters, via WDSM; Reuters, via WBOW.

Image credit: 极客湾Geekerwan, via Wikimedia Commons (CC BY 3.0).

This article was produced with the assistance of artificial intelligence, cross-checked against public posts on X and internet sources, and editorially reviewed. Please check the original sources to confirm the facts.

By Muse