Huang exit, US scrutiny imperil $3 billion in Chinese AI projects reliant on Nvidia

Chinese institutions and developers are backing more than $3 billion in artificial intelligence computing ventures designed around restricted Nvidia semiconductors, despite strict American export limits, according to an investigation by Reuters.

An examination of upwards of 150 official procurement announcements, building agreements, investment blueprints, and financial filings showed computing developments and supply contracts worth over 23 billion yuan, or around $3.2 billion. The initiatives span from prominent coastal commercial districts to inland areas such as Gansu province, involving university research facilities, local state-backed companies, and municipal computing hubs. The financial figures represent overall investments including facilities, electrical systems, and secondary equipment, rather than solely chip expenditures.

The evidence highlights the persistent Chinese appetite for high-performance American processors, including Nvidia's H100, H200, and H800 models, even after successive rounds of federal trade barriers began rolling out in 2022. While some computing centres are fully constructed or currently under development, others risk substantial delays, logistics breakdowns, or expensive retrofits using domestic chips if access to components is cut off.

The report arrives during heightened turmoil for Nvidia following the departure of its co-founder and chief executive, Jensen Huang. Huang stepped down following controversy concerning his ties to a Chinese businesswoman suspected by United States officials of having links to Beijing's intelligence apparatus. Nvidia stated its board opened an independent inquiry and emphasized it had found no proof that company secrets or American security were compromised. No regulatory body has accused Huang of breaking export laws, and the ventures detailed in the investigation have not been connected to the controversy surrounding his exit.

Nvidia previously indicated that trade restrictions essentially eliminated its presence in China's data-centre market, having dropped its direct market share from 95 percent to zero and removing domestic Chinese data-centre sales from company projections. Even so, research from International Data Corporation noted that the company made up roughly 34 percent of China's AI accelerator shipments across the opening half of 2026. The documents reviewed do not explain the exact supply chains through which the restricted hardware arrived in the country.

Analysts warn that Washington faces significant enforcement hurdles. Sunny Cheung, an analyst at the Jamestown Foundation, told Reuters that authorities should look past the nominal purchasers of hardware and investigate the end users and purposes of the resulting computing power.

Neither China’s commerce ministry nor Nvidia immediately provided comments on the report. With Huang no longer leading the company, industry observers suggest Nvidia could face diminished influence in resisting further clampdowns as American lawmakers push for even tougher trade restrictions.