Nvidia reportedly caps Chinese clients at 49% for data centers

Nvidia has reportedly introduced a new condition for the data center operators that buy its hardware: they must limit the share of their business that comes from Chinese companies. According to the claim, these customers will have to track how much of their total business is tied to the China-based firms they serve.

The alleged rule rests on a simple ratio. Business from Chinese companies could make up at most 49% of an operator's total, while the remaining 51% would have to come from customers outside China. The US company is said to be treating this as a new policy.

How the rule was communicated and who it covers

The new approach was reportedly not delivered through an official document or written announcement. Instead, Nvidia's sales and regional teams are said to have relayed it to customers directly, which means the policy does not appear to be laid out in any formal paper.

Another notable detail is that the restriction is not limited to data centers in Asia. It reportedly applies to operators worldwide. It would, however, affect only new and future contracts; existing agreements to rent Nvidia hardware fall outside its scope.

Nvidia's stated position

In its statement on the matter, Nvidia argued that American chips form the foundation that allows American model and application developers to compete worldwide. The company also said that current US policy encourages cloud providers to build on American infrastructure and to win business wherever they can.

Sanctions and the gray market

The move appears to be part of the ongoing tension between the US and China over AI capacity. Nvidia can sell its chips only to approved companies, and that has created a thriving gray market for the hardware. Earlier this year it emerged that even older-generation Nvidia servers had become so sought after that their value had tripled.

Whether the restrictions are working as intended is disputed. Nvidia hardware was recently found in the data centers of China Telecom, the state-owned Chinese telecom giant. The US government's reason for trying to limit China's access to Nvidia hardware is clear: it wants to hamper China's ability to develop its own AI technologies, since Nvidia chips remain the first choice worldwide for AI servers and data centers.

What it could mean for data center operators

The new requirement could add pressure on operators who are already competing for the same contracts. A narrower pool of acceptable customers could, over time, mean a smaller scale of operations for data centers. Having to monitor the proportion of Chinese clients would also be an added burden for service providers.