Nvidia tells cloud customers to take on less China business
Nvidia has told some of its cloud customers to take on less business from Chinese companies. The Information reports this, citing three people familiar with the matter. According to the report, the guidance is meant to push cloud providers, including Nvidia-backed neoclouds, toward a more diverse customer mix.
The directive is not mandatory, and it applies to future contracts rather than existing deals. Cloud customers are also being asked to disclose the identity of end users accessing the chips. The Financial Times reports that the instruction was communicated verbally, through sales reps and regional teams.
The Decoder, which published the account, says the move targets a gap in U.S. export controls. The rules restrict chip shipments but generally do not cover remote access through cloud services. That lets Chinese AI developers tap advanced Nvidia chips by renting compute capacity at data centers abroad, without ever importing the hardware.
U.S. authorities had been looking into Chinese AI companies accessing such chips through foreign data centers, after allegations surfaced involving Moonshot AI, the developer behind the Kimi chatbot.
Key facts
- Nvidia has told some cloud customers, including Nvidia-backed neoclouds, to take on less business from Chinese companies, per The Information, which cites three people familiar with the matter.
- The directive is not mandatory and applies to future contracts, not existing deals.
- Cloud customers are also being asked to disclose the identity of end users accessing the chips; the Financial Times says the message went out verbally via sales reps and regional teams.
- The target is a gap in U.S. export controls: the rules restrict chip shipments but generally do not cover remote access through cloud services.
- U.S. authorities had been looking into Chinese AI companies using such chips through foreign data centers after allegations involving Moonshot AI, the Kimi developer.
Why it matters
U.S. export controls limit shipments of advanced chips to China but generally do not cover renting them through cloud services. Chinese AI developers can therefore use advanced Nvidia chips by renting compute at data centers abroad, without importing hardware. The reported guidance shows Nvidia itself leaning on its cloud customers over that route, rather than the rules doing it. U.S. authorities had already been looking at the practice after allegations involving Moonshot AI.
Who it affects
Cloud providers that buy Nvidia chips are the direct audience, including Nvidia-backed neoclouds. Chinese AI companies that rent compute abroad could find fewer new contracts available. End users of the chips may also have to be identified, since cloud customers are being asked to disclose who is accessing the hardware.
How to use it
This is a report about a business instruction, not a product or tool, so there is nothing to adopt. For cloud providers, the practical points reported are these: the guidance concerns future contracts, not existing deals, and it comes with a request to disclose the identity of end users. Anyone tracking chip access rules can read it as a sign of where enforcement pressure is moving: toward cloud rental.
How solid is it
The core claim rests on The Information, which cites three anonymous people familiar with the matter. The Financial Times separately reports how the instruction was communicated. The source gives no date for the guidance and does not name which customers were told. No Nvidia statement is quoted, so the company's response is not given. This is reporting, not an official announcement.
Risks and caveats
The guidance is not mandatory, and the source describes no penalty or enforcement mechanism for ignoring it. No target or cap on Chinese business is stated. The allegations involving Moonshot AI are described only as having surfaced; the source does not say they were proven or what the U.S. inquiry found. It also does not say which Chinese companies are affected beyond that mention.