Reuters Maps $3 Billion in Nvidia-Powered AI Data Centre Projects in China

A Reuters review published Friday identified more than $3 billion in proposed and operating Chinese computing projects that specify Nvidia processors — in a market the US chipmaker says export controls have effectively closed to it. The factbox draws on government prospectuses, tender documents and corporate disclosures, and spans projects from coastal technology hubs to remote western provinces.

The figure shows how deeply Nvidia's technology remains embedded in China's AI infrastructure build-out. It also shows how fragile that supply chain has become.

Which projects made the list?

The largest single project is a proposed 6.2 billion yuan ($873 million) AI computing centre in Xianyou county, Fujian province. It seeks commercial partners and specifies Nvidia H100 chips for large-scale AI development.

The pipeline, as mapped by Reuters, includes:

• Shaanxi province: state-owned engineering group Sinomach said a subsidiary won a contract for a 3 billion yuan computing centre specifying H100 and H200 processors.
• Jilin province: an April government prospectus describes a 2.26 billion yuan project combining H100 processors with Chinese-developed chips. The developer expects bank loans to cover roughly 80% of the investment.
• Inner Mongolia: a 1.5 billion yuan development proposes 625 H200 servers. Chinese industry reporting in June said construction had begun on two buildings in its first phase.
• Gansu province: a local company awarded a 1.16 billion yuan construction contract in July for a centre planning approximately 3,072 Nvidia processors, spanning H100 and H200 models. The contract went to a consortium led by a subsidiary of state-owned Shaanxi Construction Engineering.

For scale: the Fujian project alone would cost more than the entire mapped pipeline of several smaller provinces combined, and the Gansu centre's 3,072 planned processors exceed the server counts of most regional facilities on the list.

Are any clusters already running?

Yes, and that distinction matters. The Wuxi municipal government in Jiangsu province said in March 2025 that a computing cluster using H800 processors had entered operation. In Fujian, Shanghai-listed Damingcheng disclosed in June that a computing venture had built an H800 cluster and signed a computing-services agreement originally worth 759 million yuan. That contract ended early.

Demand extends to research institutions. Shenzhen University awarded a contract for three H100 processors in September. Northwestern Polytechnical University — a school already on the US entity list — bought two H200 chips in October, Reuters reported separately.

In plain words: some of this is intent on paper, but operating clusters prove restricted silicon has already reached Chinese facilities.

What does the gap between policy and procurement mean?

The projects expose a gap between US policy and procurement reality. Nvidia has said US export controls have effectively shut it out of China's AI data-centre market. Yet Chinese developers continue to specify Nvidia's restricted processors in projects worth billions of yuan.

Reuters notes that tighter curbs risk construction delays, inflated costs, or forced redesigns around domestic alternatives. Timing sharpens the risk: the factbox appeared a day after Nvidia co-founder Jensen Huang resigned as chief executive over a scandal. Huang had been a vocal advocate of easing export restrictions. His departure could deepen vulnerabilities for Chinese companies exposed to potential supply disruptions as US curbs tighten.

What remains undocumented?

Plenty — and Reuters says so itself. The documents show intent and contract awards, not confirmed delivery of restricted chips in every case. The Gansu tender documents explicitly do not establish whether the processors have been delivered. For proposed projects, financing and commercial partnerships are not yet secured. How many will proceed as specified, and through what channels the chips would arrive, is absent from the public filings.

The takeaway: $3 billion in planned Nvidia capacity is a statement of demand, not a statement of delivery — and the gap between the two is exactly where US enforcement, Chinese procurement and Nvidia's future in the market will collide.