Nvidia Ties Cloud Expansion to a 51% Non-China Rule for Future Contracts

Nvidia is asking data-centre partners around the world to lean less on Chinese customers, according to a Financial Times report, while simultaneously widening its cloud alliances and taking a financial stake in one of them. The two developments landed within a day of each other and paint a company pushing growth on several fronts at once — not all of them free of friction.

The reported requirement: at least 51% of services under new contracts should go to companies outside China, capping Chinese customers at no more than 49%. Citing five people familiar with the matter, the FT said Nvidia has communicated the guidance verbally over the past few months. It applies globally but only to new agreements; existing computing-rental arrangements are said to be untouched. Reuters reported on the guidance but could not independently verify it.

A sales-level constraint, not a retreat

Nvidia's sales and regional teams are expected to enforce the rule, which would hand cloud providers an extra commercial hurdle: they must not only drum up demand for computing power but also weigh where new customers are based. Providers in Singapore warned that finding enough non-Chinese buyers could prove difficult, meaning strict enforcement might dent their business. For Nvidia, the exposure therefore extends to partners outside China whose demand ultimately rests on Chinese users.

The backdrop is a tightening US export regime for advanced AI technology and a crackdown on smuggling networks. The Commerce Department issued related guidance in May, with Blackwell chips among the products covered. Nvidia had already cut its list of approved Asian customers by more than 50% in July. Its vetting includes data-centre visits, contract checks and conversations with end users — scrutiny that reaches past hardware shipments to how cloud providers' own customers use the computing power.

Meta stake and Oracle build-out

On the growth side, Nvidia said on 27 October that it would deepen its partnership with Meta and subscribe to $3 billion of the company's share offering. The Meta tie-up covers its Muse models and the planned additional deployment of Vera Rubin and Vera CPUs. The same day, Nvidia and Oracle announced they would scale Oracle Cloud Infrastructure superclusters to multi-gigawatt size using NVIDIA Vera Rubin NVL72 systems; those systems and Vera CPUs are generally available on OCI, according to the companies.

A separate and larger transaction is still pending. Reuters reported on 28 October that the US Department of Justice had sought information on Nvidia's planned $12.93 billion acquisition of Hugging Face, including possible effects on competition in artificial intelligence. Closing is expected in the first half of 2027. That is a planned step, not a completed one — a distinction that matters when weighing the expansion strategy, since announced partnerships are not the same as signed deals.

What investors can pin down

The picture that emerges is of a company advancing partnerships and investment plans while competition questions and possible customer rules shape how those plans play out. No revenue impact from the cloud guidance can be quantified yet. One date is fixed: Nvidia has scheduled its third-quarter fiscal 2027 earnings call for 17 November. The quarter ended on 25 October.