Chinese university with PLA ties acquires Nvidia H200 chips despite U.S. blacklist

Chinese university with PLA ties acquires Nvidia H200 chips despite U.S. blacklist
Blacklisted chips acquired

Scrutiny of U.S. export controls on advanced AI hardware is intensifying after procurement records showed a defense-linked Chinese university obtained Nvidia H200 chips last month. The purchase adds to concerns that Chinese institutions are still accessing restricted computing power through domestic suppliers and rented server capacity.

Highlights

  • Northwestern Polytechnical University purchased two Nvidia H200 chips for 874,700 yuan ($123,000) on October 21, 2023, despite its U.S. blacklist status.
  • Procurement records show the university accessed additional H200 and other Nvidia AI processors through at least five contracts with Chinese companies for rented computing capacity in 2024.
  • The university's cybersecurity school awarded two contracts in January 2024 for seven-month server access containing eight H200 chips each, totaling 882,700 yuan for AI model training.

Procurement records expose access route

As reported by BusinessWorld Online, citing Reuters, Northwestern Polytechnical University bought two Nvidia H200 chips on October 21 for 874,700 yuan, about $123,000, according to a procurement record on the university's website. The chips were purchased for its management school from Shaanxi Bosuan Network Technology rather than directly from Nvidia.

Northwestern Polytechnical is part of China's Seven Sons of National Defense group of universities and is based in Xian, where it conducts research related to military aircraft, missiles and satellites. The H200 is among Nvidia's most capable chips for training and running advanced AI systems, and exports to blacklisted institutions such as Northwestern Polytechnical are normally subject to license denials under U.S. rules.

Records also show the university bought access this year to H200 chips and other Nvidia AI processors through five contracts for rented computing capacity from Chinese companies. Its cybersecurity school awarded two seven-month contracts in January for access to servers containing eight H200 chips each, for 441,700 yuan and 441,000 yuan, covering AI model fine-tuning and reinforcement-learning training.

The university also rented access to servers containing eight Nvidia A100 chips and eight A800 chips, while another institute in April bought computing services requiring at least 24 A100 chips in a 375,000 yuan contract for structural and fluid-dynamics simulations. The records provide little detail on where the processors originated or where the rented servers were operated.

As a reminder, we recently covered Nvidia’s push for deeper AI safety dialogue with Chinese companies amid U.S.-China regulatory scrutiny.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.