Reuters identified more than 23 billion yuan ($3.2 billion) in Chinese AI data-center projects specifying Nvidia chips subject to U.S. export controls. That figure is total project investment, not confirmed chip purchases or shipments. The documents do not establish that all the chips were obtained or that any export laws were broken.
The projects are at different stages: a proposed Fujian center specifying H100 servers is still seeking financing partners, while authorities in Wuxi reported that an H800 cluster began operating in March 2025. H200 exports to China require licenses under a case-by-case review policy introduced in January 2026; H100 shipments remain more tightly restricted.
