Nvidia’s China Cloud Rules Leave Existing Contracts Intact, but Future Demand Is Unclear

Nvidia’s forecast for the third quarter of fiscal 2027 offers no estimate of data-center computing revenue in China. That leaves a separate question unanswered: how reported restrictions on cloud partners might affect their future orders. Nvidia projects revenue of $108 billion, with a tolerance of ±2%, but has not quantified any effect on partners’ bookings.

The Financial Times, citing five people familiar with the matter, reported that Nvidia is asking cloud customers to direct at least 51% of services under new contracts to companies outside China. That would cap the share going to Chinese companies at 49%. The guidance is said to apply worldwide, but only to future agreements; existing data-center and cloud rental contracts would remain unchanged. Nvidia has not confirmed either the reported quota or how it would be enforced, and Reuters said it could not independently verify the report initially.

The distinction between new and existing business is central. The reported instruction has been communicated orally over recent months through Nvidia’s sales and regional teams, rather than described as a change to active agreements. It could nonetheless constrain the customer mix cloud providers can accept in future deals. Providers in Singapore have warned, according to media reports, that finding enough customers outside China may be difficult and that strict implementation could put pressure on their businesses.

The reported policy would target an indirect route to Nvidia computing power: Chinese companies renting capacity from cloud providers based elsewhere. US restrictions on chip sales to China have tightened since at least 2022. In May, the US Commerce Department issued guidance related to advanced AI technology and efforts against smuggling networks; products covered include Blackwell chips. The reported cloud rules would extend scrutiny beyond hardware shipments to how customers use computing capacity.

Nvidia has also tightened checks along that customer chain. Reported measures include visits to data centers, contract reviews and discussions with end users. In July, the company cut its list of approved Asian customers by more than 50%. The extent to which the new quota is binding remains unclear: it has not been established whether partners face contractual requirements or nonbinding guidance, or how Nvidia would measure compliance.

Nvidia is due to report its next quarterly results on November 17. Investors will be looking for clarification on whether the company confirms the reported rules and how broadly they apply. Until then, the reported protection for existing contracts limits what can be inferred about near-term disruption, while the impact on future partner demand remains unquantified.