Nvidia's data center revenue is roughly split down the middle, with about half generated by customers outside the hyperscaler category, the company's chief financial officer said, according to a Fortune report. The figure counters a common assumption that Nvidia's AI infrastructure sales depend almost entirely on a handful of the largest cloud computing operators.

The statement carries weight for anyone modeling Nvidia's revenue concentration risk. If hyperscalers account for only around half of the data center segment, the other half flows from a broader mix of buyers — and that mix matters for how analysts assess the durability of demand for Nvidia's GPU and networking products.

What does the CFO's figure change?

The disclosure reframes the debate over who is actually paying for AI compute. Public attention has focused on capital expenditure announcements from the largest cloud platforms, which purchase GPUs at scale to build out AI services. A customer base where non-hyperscalers represent approximately 50% of the data center business indicates that demand extends well past that group.

For investors, the number speaks directly to concentration risk. A revenue base dominated by four or five large buyers would expose Nvidia to sudden spending pauses at any one of them. A base where half the segment's revenue comes from other customer types spreads that exposure across a wider set of buyers.

For competitors and supply-chain partners, the figure sketches the addressable market beyond the major cloud operators. Enterprises, neocloud providers, sovereign and research buyers, and system integrators all fall into the non-hyperscaler half, though the CFO's reported statement did not provide a breakdown within that category.

How does this fit Nvidia's reporting?

The data center segment is Nvidia's largest revenue driver, and its customer composition has been a recurring question on earnings calls and in analyst notes. The CFO's figure — approximately half from customers beyond the hyperscalers — is the most direct quantification of that split to come from the company's leadership, as relayed by Fortune.

The statement does not specify a dollar amount for either half, a time period for the measurement, or the definitions used to classify a customer as a hyperscaler. Readers should treat the figure as a proportional characterization of the segment rather than a line-item disclosure from a financial filing.

Why does the split matter for the market?

Three implications follow from the reported figure:

• Demand breadth. AI infrastructure spending is not confined to the largest cloud platforms; other customer classes are committing material amounts, enough to match the hyperscaler share of the segment.
• Risk profile. Revenue tied to a wider customer base is less sensitive to procurement decisions at any single large account, a consideration in how the stock is valued.
• Competitive dynamics. Vendors selling alternatives to Nvidia's products now know that roughly half of the incumbent's data center business sits with customers whose buying criteria may differ from those of the biggest cloud operators.

The CFO's reported comment also serves a signaling function. By putting the split at about half, Nvidia's leadership is communicating that the company's growth story rests on more than the capital expenditure cycles of a few hyperscalers — a message aimed at investors who have flagged AI-related spending as concentrated among a small number of buyers.

What remains unknown?

The figure leaves open questions that the reported statement did not address. Fortune's report did not break down the non-hyperscaler half by customer type, geography, or product line. It also did not indicate whether the proportion has shifted over recent quarters or whether Nvidia expects the mix to change as additional compute capacity comes online.

Those details will likely surface in future earnings disclosures or executive commentary. For now, the headline number stands: about half of Nvidia's data center business comes from customers beyond the hyperscalers, according to the company's CFO.