Synopsis
Chinese universities have previously acquired restricted chips in servers made by Super Micro and other manufacturers, Reuters reporting from 2024 shows. But the continued practice, particularly by institutions with links to the PLA, is likely to stoke concerns of some U.S. lawmakers.It was not clear how the servers were sourced.
Concerned about the potential for artificial intelligence chips to enhance China's military capabilities, the U.S. from 2022 began banning the sale of some Nvidia chips such as the A100 to China.
San Jose, California-headquartered Super Micro hit the headlines last week when three people associated with the company, including its co-founder, were charged with helping smuggle at least $2.5 billion of U.S. AI technology to China.
The company was not named in the indictment and says it was the victim of an elaborate scheme by the individuals.
Chinese universities have previously acquired restricted chips in servers made by Super Micro and other manufacturers, Reuters reporting from 2024 shows. But the continued practice, particularly by institutions with links to the PLA, is likely to stoke concerns of some U.S. lawmakers.
On Monday, two U.S. senators, citing the indictments of the three people linked to Super Micro, urged U.S. Commerce Secretary Howard Lutnick to consider pausing all export licenses allowing advanced Nvidia AI chips and server systems to be sent to China or intermediaries in Southeast Asia.
In addition to the universities that bought the Super Micro servers, two others - including one linked to the military - sought to make similar purchases, Reuters checks of publicly available tender documents for 2025 and early this year show. In those two cases, it was not clear if there was a successful purchase.
Super Micro declined to comment on the information in the documents. Nvidia said it continues to work closely with customers and the U.S. government on compliance as export regulations have expanded.
China's commerce ministry and the U.S. Commerce Department did not respond to a request for comment.
A hot-button issue
Much remains unclear about how many sales of U.S. AI chips to China will be acceptable to Washington and Beijing.
The Trump administration last year greenlighted the sale of Nvidia's H200 chips - more powerful than currently restricted products - albeit with a number of conditions that could limit amounts sold.
For its part, Beijing, keen to bolster Chinese chipmakers, has discouraged companies from purchasing U.S. AI chips. But sources said this month that Nvidia has won Beijing's approval to sell the H200. As yet, there have been no confirmed sales.
Critics of such sales worry that they will accelerate research.
That "may in turn help improve China's weapons design and testing, military planning and logistics, autonomous weapons systems, or surveillance," said Jacob Feldgoise, senior data research analyst at Georgetown University's Center for Security and Emerging Technology.
Examples of procurement of restricted AI chips include a document from Beihang University, one of China's "Seven Sons of National Defense" institutions that plays a key role in aerospace and defence research and is linked to the PLA.
It said in a March 16 notice this year it had procured a machine-learning workstation built on a Super Micro system, configured with four Nvidia A100 chips.
A July notice from Harbin Institute of Technology (HIT), another one of China's "Seven Sons" that has worked on missile, satellite, and robotics technologies, shows it procured a Super Micro system with eight Nvidia A100s.
Beihang University, located in Beijing, and HIT, based in the northeastern city of Harbin, did not respond to Reuters requests for comment. Both are on a U.S. export blacklist, which makes it hard for U.S. companies to secure licenses to supply them.
