Nvidia has told its data center and cloud partners around the world to keep Chinese customers below half of their new business, according to a Financial Times report published on Tuesday, November 3. Citing five people with knowledge of the matter, the report said partners must sell at least 51 percent of their services to non-Chinese companies, which caps the Chinese share at 49 percent.
The reported guidance applies globally, not only to facilities in Asia, and it applies only to future contracts. Existing agreements are not affected, the report said. It was communicated verbally through Nvidia's sales teams and regional offices over recent months, rather than through formal written instructions, and no such quota existed previously, according to the people cited.
Nvidia neither confirmed nor denied the report. In a statement reported by AFP, the company said that "data centers and clouds built on American chips are the foundation that American model and application makers need to compete worldwide." It added: "America cannot win the AI race unless America's AI infrastructure wins worldwide — if the world's clouds rely on foreign-made chips, the model and application layers will be foreign-built as well." A follow-up report by The Information, cited by other outlets, said the instructions were intended to encourage a broader customer mix and were not being enforced as a mandatory restriction. That distinction matters, because it is still unclear whether the 51 percent figure is a binding requirement, how compliance would be measured, or what would happen to a partner that missed it.
The target appears to be an indirect route to advanced computing power. Direct sales of Nvidia's most advanced AI chips to China are restricted under United States export controls, but Chinese companies have also been able to rent capacity from cloud providers based elsewhere. Requiring those providers to rebalance their customer mix toward non-Chinese firms would narrow that overseas leasing channel without changing the export rules themselves. The reported move also follows earlier tightening inside Nvidia's own sales process: reports in recent months said the company cut the number of Asian customers approved to buy its AI processors by more than half after introducing a stricter whitelist of vetted buyers.
For cloud operators, the practical effect would fall on new bookings. A provider whose growth has depended heavily on Chinese customers would need to win more business elsewhere before adding the same volume of Nvidia-based capacity, and some operators outside the United States have already begun courting non-Chinese accounts, according to reports following the Financial Times story. Hyperscalers with large, diversified global customer bases would find that mix easier to show than smaller, specialized providers built around a narrower set of tenants.
China pushed back in general terms. Asked about the report, foreign ministry spokeswoman Mao Ning said she was not aware of it, and said China consistently hopes all parties can provide a fair, transparent and nondiscriminatory business environment for companies from all countries, including Chinese companies.
For now, the story rests on attributed reporting rather than a confirmed policy, and investors and partners will be listening for clarification. Nvidia is scheduled to discuss its third-quarter fiscal 2027 results, for the quarter ended October 25, on November 17 — an obvious occasion for analysts to ask whether the guidance is real, whether it is mandatory, and whether it changes demand for new data center contracts.