Reuters reports that Northwestern Polytechnical University bought two Nvidia H200 accelerators in October through a domestic supplier and also contracted for computing capacity containing H200, A100, and A800 chips. The university is on the US Commerce Department’s Entity List. Reuters says it reviewed 26 procurement and corporate records involving Chinese universities and research institutions that bought or rented access to Nvidia processors subject to US restrictions.

The records do not answer every key question. Reuters could not determine the origin of all chips or the physical location of all servers, and the university, Nvidia, and the US Commerce Department did not respond to requests for comment in the report. A purchase record is evidence of a transaction, but it does not by itself establish how the goods were sourced, whether a specific export-control violation occurred, or which workloads used the hardware.

For AI developers, the story highlights how access to advanced compute depends on licensing, supply chains, and cloud capacity, not only model availability. It does not imply that ordinary developers will immediately lose access to a service. Watch official export rules, cloud provider disclosures, and regional capacity or pricing changes, and avoid relying on hardware that may be difficult to procure or legally operate.