Procurement documents posted online indicate that Northwestern Polytechnical University - an institution with deep links to China's defence establishment - acquired two Nvidia H200 AI chips last month and arranged multiple contracts this year to rent access to servers containing powerful Nvidia processors. The transactions appear to have been executed through Chinese suppliers rather than directly from the chipmaker, raising questions about enforcement of U.S. export controls on advanced AI hardware.
According to a procurement entry dated October 21 on the university's website, the management school purchased two H200 processors for 874,700 yuan. The record lists Shaanxi Bosuan Network Technology as the supplier rather than direct purchase from Nvidia. The H200 is among Nvidia's most capable processors for training and running sophisticated AI systems and is subject to U.S. export licence approvals that are ordinarily denied for entities on the U.S. trade blacklist.
Northwestern Polytechnical, located in the northwestern city of Xian, is identified in procurement records as having multiple contracts this year for rented computing capacity that granted access to H200 chips and other high-end Nvidia processors. Those computing contracts and the outright chip purchases are disclosed in the same set of procurement documents made public by the university.
Blacklisting and licensing background
The university was added to the U.S. Entity List in 2001 over alleged involvement in missile proliferation activities and military research. The restrictions were tightened in 2019, subjecting exports of controlled U.S. technology to the university to a policy of denial in most circumstances. Despite those measures, the procurement records demonstrate that Chinese institutions with military affiliations continue to obtain access to U.S.-engineered AI computing power through domestic suppliers and rented server capacity.
Separately, policy changes earlier this year introduced case-by-case licensing for some H200 exports to China. Those rules require safeguards intended to prevent remotely enabling blacklisted entities and military end-users to access chips exported under the policy. However, the separate Entity List restrictions for organisations such as Northwestern Polytechnical were not lifted by that policy change.
Details of purchases and rented access
Beyond the management school's purchase of two H200 processors, procurement entries show the university's cybersecurity school awarded two seven-month contracts in January for access to servers containing eight H200 chips each. The two contracts were valued at 441,700 yuan and 441,000 yuan respectively, and were described as supporting fine-tuning of AI models and reinforcement-learning training.
Records further indicate the university rented access to servers containing eight Nvidia A100 chips and eight A800 chips. The A100 and A800 processors are subject to more stringent restrictions and are banned from export to all Chinese entities. In April, another institute within the university secured computing services that required at least 24 A100 chips under a 375,000 yuan contract cited for structural and fluid-dynamics simulations.
The procurement awards were made to several relatively obscure Chinese technology suppliers. The public announcements provided limited detail about the origin of the processors or the physical locations where the leased servers operate. The procurement notices do reflect a legal requirement under Chinese public procurement rules that agencies verify a supplier's ability to fulfil orders prior to announcing winning bidders and signing contracts.
A broader pattern across Chinese universities
These entry-level and rented-access transactions form part of a wider pattern in the procurement records reviewed. The documents include purchases and rental contracts involving other Chinese universities and research institutions obtaining U.S.-restricted Nvidia chips or paying for access to Nvidia-based computing services via domestic vendors.
- Shenzhen University recorded a purchase of three H100 processors for 769,800 yuan in September.
- Wuhan University of Technology bought a server containing an H100 chip for 355,000 yuan in June and separately acquired four A100 chips for 344,000 yuan.
- Beihang University, another institution noted for defence-related research, purchased four A800 processors for 489,000 yuan in April.
- Procurement announcements in July and September named suppliers selected to provide computing services using H800 and A800 chips to the University of Chinese Academy of Sciences, with one listing H800 access priced at 10 yuan per chip-hour.
- A January procurement award identified a supplier providing six A100 chips worth 484,800 yuan for a Shenzhen-based public institution.
Those entries reflect both outright purchases of restricted chips and procurement of computing time on servers that host such processors, suggesting a market for both physical units and hosted compute services inside China.
Responses and gaps in public information
Requests for comment sent to Northwestern Polytechnical University, Nvidia and the U.S. Commerce Department's Bureau of Industry and Security were not answered. The procurement records provide limited details about suppliers or the supply chains for the processors, offering little clarity on the chips' points of origin or the jurisdictions in which servers are located.
Publicly available procurement announcements typically provide minimal vendor information and do not disclose comprehensive supply-chain provenance. The vendors named in the awards tend to be smaller, lesser-known Chinese firms, and the records do not explain how the processors were initially obtained.
Previous reporting and market dynamics
Separate inquiries and reporting have documented instances in which Chinese institutions procured restricted Nvidia processors embedded in servers manufactured by established server vendors and acquired through domestic resellers. Those lines of inquiry have also described an underground market in China for restricted Nvidia chips sourced from overseas inventories and intermediaries operating through third countries.
More recently, small quantities of H200 chips reportedly arrived in China through indirect routes while broader approvals for large-scale imports remained withheld by Chinese authorities. The procurement records published by the universities add to this picture by showing legitimate-looking domestic purchases and service contracts that provide access to the same categories of restricted Nvidia processors.
Implications for export controls and industrial players
The procurement entries reveal how advanced AI hardware subject to U.S. export controls can be acquired within China through a combination of direct purchases from domestic vendors and rented access to hosted servers. For semiconductor firms and cloud and hosting providers, the records underscore the complexity of policing end-user restrictions once hardware has entered domestic markets and been reallocated through local suppliers.
For educational and research institutions, the records show a demand for leading-edge AI compute to support activities from model fine-tuning to reinforcement learning and scientific simulations. When such demand is tied to organisations placed on restrictive trade lists, it creates a compliance challenge for governments and firms that must balance licensing policies, safeguards and enforcement.
Summary of transactions, procurement values and restrictions are drawn from university procurement records made publicly available on institutional websites and from related procurement announcements.