China’s $3.2 Billion AI Pipeline Still Calls for Nvidia Chips Washington Restricts

Chinese developers are planning and building billions of dollars worth of AI data centers around Nvidia processors restricted by Washington, according to a Reuters investigation published Friday.

After examining more than 150 procurement notices, investment plans and corporate records, the news agency identified projects worth over 23 billion yuan ($3.2 billion). Many specify Nvidia's H100, H200, or H800 chips, which provide the computing power used to train and run advanced AI models.

But $3.2 billion is the total value of the projects, including buildings, electricity and other infrastructure. The documents do not establish that all the processors were obtained, or that any export laws were broken.

What the documents reveal

One proposed computing center in Fujian province is valued at 6.2 billion yuan ($873 million). Its official investment listing specifies H100 servers but is still seeking financing partners.

In Shaanxi, a state-owned engineering group won a contract linked to a 3 billion yuan center naming H100 and H200 chips. A 2.26 billion yuan project in Jilin proposes combining Nvidia hardware with Chinese processors.

Inner Mongolia has another proposed development specifying 625 H200 servers. Meanwhile, authorities in Wuxi reported that an H800 computing cluster began operating in March 2025. Operating facilities and proposed investments are not evidence of the same level of chip access.

Nvidia's processors can carry out thousands of calculations at once, making them valuable for AI training. Its widely used software also makes switching to rival systems more complicated and potentially expensive.

The United States began restricting advanced AI chip sales to China in 2022, citing national security concerns. In January 2026, the Commerce Department moved H200 export applications to case-by-case review, although sales still require licenses. H100 shipments remain more tightly restricted.

That leaves Chinese developers with a dilemma: design facilities around familiar Nvidia equipment they may struggle to acquire, or invest in alternative processors that can require software changes and additional engineering.

China's alternative is growing

An IDC report cited by Reuters estimated Nvidia accounted for 34% of China's AI accelerator shipments in the first half of 2026, compared with 28% for Huawei. Those figures cover the broader chip market, not confirmed shipments of restricted models.

Chinese AI developer Z.AI has already released an image-generation model trained entirely on Huawei chips, demonstrating that competitive systems need not depend on American hardware.