About TrendForce News

TrendForce News operates independently from our research team, curating key semiconductor and tech updates to support timely, informed decisions.

[News] China’s AI Projects Reportedly Worth Over US$3B Specify Restricted NVIDIA GPUs: A Closer Look


2026-11-18 Semiconductors editor

Following Jensen Huang’s departure as NVIDIA CEO, U.S. scrutiny of the chipmaker has intensified, with Washington introducing new restrictions on China’s access to advanced AI computing through overseas data centers. But how much demand for restricted NVIDIA hardware remains in China? According to Reuters, Chinese AI computing projects reportedly worth more than US$3 billion specify restricted NVIDIA processors, with some facilities already operational.

Major AI Projects Specify Restricted NVIDIA GPUs

The projects span several provinces, with some involving state-owned enterprises. In Fujian, a proposed RMB 6.2 billion (US$873 million) AI computing center in Xianyou County is seeking commercial partners. According to Fujian’s official investment promotion platform, the first phase targets approximately 4,000 PFLOPS of sparse computing capacity, with all servers specified to use NVIDIA H100 GPUs.

In Shaanxi, a subsidiary of state-owned engineering group Sinomach secured a contract linked to an RMB 3 billion computing center specifying NVIDIA H100 and H200 processors, Reuters reports. Meanwhile, a government prospectus in Jilin outlines an RMB 2.26 billion project combining NVIDIA H100 GPUs with domestic chips, with bank loans expected to finance about 80% of the investment.

Some projects have moved into construction, although GPU deliveries remain unconfirmed. In Inner Mongolia, an RMB 1.5 billion project plans to deploy 625 NVIDIA H200 servers, with Reuters reporting that construction of two buildings began in June as part of the first phase.

Meanwhile, in Gansu province, a local company awarded an RMB 1.16 billion construction contract in July for an AI computing center planning to deploy approximately 3,072 NVIDIA H100/H200 GPUs, according to Reuters, citing tender documents. The contract went to a consortium led by a subsidiary of state-owned Shaanxi Construction Engineering, although the documents do not confirm whether the GPUs have been delivered.

Some Chinese facilities have already deployed NVIDIA processors subject to U.S. export restrictions. In Jiangsu, the Wuxi municipal government announced in March 2025 that an NVIDIA H800 computing cluster had entered operation, Reuters notes. Meanwhile, Shanghai-listed Damingcheng disclosed in June that a computing venture in Fujian had built an H800 cluster and signed a computing services agreement initially worth RMB 759 million, although the contract was later terminated early.

Demand for restricted NVIDIA GPUs also extends to academic institutions. According to Reuters report, Shenzhen University awarded a contract for three H100 GPUs in September, while U.S.-blacklisted Northwestern Polytechnical University purchased two H200 chips the following month.

The continued demand for NVIDIA GPUs, from large-scale computing centers to university research, highlights the challenges China faces in shifting to domestic alternatives. According to Tom’s Hardware, citing the Financial Times, China’s domestic AI chip production is expected to reach 3.5 million units in 2026, falling short of demand exceeding 5 million.

Read more

(Photo credit: FREEPIK)

Please note that this article cites information from Reuters, Fujian’s official investment promotion platform, Tom’s Hardware,  and Financial Times.



Get in touch with us