Nvidia has urged certain cloud customers to reduce their reliance on business from Chinese companies. According to reports citing sources familiar with the matter, the guidance aims to encourage cloud providers and Nvidia-backed neoclouds to cultivate a more diverse customer mix.
The non-mandatory directive applies exclusively to future contracts rather than existing agreements. Additionally, cloud providers are being asked to disclose the identities of end users accessing the hardware. These instructions were reportedly communicated verbally by sales representatives and regional teams.
This initiative addresses a regulatory gap in U.S. export controls, which restrict physical chip shipments but generally omit remote access via cloud services. Consequently, Chinese artificial intelligence developers have been able to utilize advanced Nvidia hardware by renting computing capacity at foreign data centers. U.S. authorities previously investigated similar remote access arrangements following allegations involving Moonshot AI and its Kimi chatbot.
This development matters because it highlights voluntary industry efforts to address regulatory loopholes regarding cross-border artificial intelligence compute access.