Reuters: China's over 3.2 billion USD AI data center project still designated to use restricted NVIDIA chips

According to a Reuters survey, Chinese developers are still planning AI data centers around Nvidia processors that are subject to U.S. export controls, with a total project value exceeding 23 billion yuan (approximately 3.2 billion USD). The agency compiled over 150 procurement announcements, investment plans, and corporate records, many of which explicitly list H100, H200, or H800 chips. The 3.2 billion USD is the overall investment amount for the projects, covering infrastructure such as construction and power, and the documents did not prove that all related chips have been obtained, nor did they show any violations.

A proposed computing center in Fujian is valued at 6.2 billion yuan (approximately 873 million USD), with the official investment list specifying H100 servers, which are still seeking financing; a state-owned engineering group in Shaanxi has won a contract related to a 3 billion yuan project, with documents listing H100 and H200; a 2.26 billion yuan project in Jilin plans to combine Nvidia hardware with domestic processors; a proposed project in Inner Mongolia specifies 625 H200 servers, while Wuxi claims that an H800 cluster has been put into operation in March 2025.

Since 2022, the U.S. has restricted the export of advanced AI chips to China, and in January 2026, the Department of Commerce changed the export application for H200 to case-by-case approval, with sales still requiring a license, and restrictions on H100 being even stricter. Reuters cited IDC data stating that in the first half of 2026, Nvidia accounted for 34% of AI accelerator shipments in China, while Huawei accounted for 28%. Chinese developer Z.AI has released an image generation model fully trained on Huawei chips. These records show that Nvidia is still deeply embedded in China's AI infrastructure planning, putting pressure on U.S. officials facing congressional scrutiny.