Nvidia says U.S. export controls have effectively shut it out of China's AI data-centre market. Reuters looked at the paperwork and found more than $3 billion of projects specifying the chipmaker's restricted processors anyway.

The review, published on November 13, catalogues plans stretching from wealthy coastal hubs to remote western provinces. In Fujian's Xianyou county, a proposed 6.2 billion yuan (about $873 million) AI computing centre is seeking commercial partners and lists Nvidia H100 chips. In Shaanxi, state-owned engineering group Sinomach said a subsidiary won a contract for a 3 billion yuan centre specifying H100 and H200 processors. A Jilin government prospectus from April describes a 2.26 billion yuan project pairing H100s with Chinese-made chips and roughly 80 percent financed by bank loans. In Inner Mongolia, a 1.5 billion yuan development proposes 625 H200 servers.

Some of it is further along. In Gansu, a consortium led by a subsidiary of state-owned Shaanxi Construction Engineering won a July construction contract worth 1.16 billion yuan for a centre planning about 3,072 Nvidia processors; tender documents specify H100 and H200 hardware but do not establish whether the chips were delivered. In Jiangsu, the Wuxi municipal government said in March 2025 that a cluster using H800 processors had entered operation, and Shanghai-listed Damingcheng disclosed an H800 cluster in June.

Demand reaches research institutions too: Shenzhen University awarded a contract for three H100 processors in September, and blacklisted Northwestern Polytechnical University bought two H200 chips the following month.

The context sharpens the risk. The findings appeared a day after Nvidia co-founder Jensen Huang resigned as chief executive over a scandal — a departure that Reuters notes could deepen Chinese buyers' exposure to delays and supply disruption as U.S. curbs tighten. Every project that names a restricted part is a plan that may have to be delayed, re-costed or redesigned around domestic silicon.