Q1 results: Hyperscalers now 50% of Nvidia's data-center sales as demand diversifies

Nvidia’s first-quarter results gave one of the clearest breakdowns yet of who is buying AI infrastructure beyond the biggest cloud companies. On May 20, the company said hyperscalers accounted for about 50% of data-center revenue in the quarter, while the other half came from what Nvidia now groups as AI cloud, industrial and enterprise customers. The figures arrived with record quarterly revenue of $81.6 billion and data-center revenue of $75.2 billion. They also came with a second message from management: demand is spreading across sovereign projects, enterprise installations and regional AI cloud operators. ### Why does the 50% figure matter? Colette Kress, Nvidia’s chief financial officer, told analysts that hyperscale revenue was $38 billion, or about half of data-center revenue, up 12% sequentially. In the same call, she said the company’s ACIE segment — AI cloud, industrial and enterprise — generated $37 billion, up 31% sequentially, with AI cloud revenue more than tripling from a year earlier. (investor.nvidia.com) That split matters because Nvidia’s recent surge has often been framed around a handful of giant cloud buyers. The quarter showed the rest of the market is now large enough to match hyperscaler demand in aggregate, at least by Nvidia’s new internal grouping. Nvidia also said sovereign revenue grew more than 80% from a year earlier and that its infrastructure is now deployed in nearly 40 countries. (fool.com) ### Who is included in the non-hyperscaler bucket? Nvidia described the second half of demand as coming from AI cloud providers, industrial users and enterprise customers. Kress said partner data centers above 10 megawatts nearly doubled in a year to more than 80 sites, a sign that compute capacity is being built outside the largest public-cloud platforms. (fool.com) The company’s framing also points to more on-premise and nationally backed deployments. In the earnings-call figures, sovereign customers were listed alongside enterprise and industrial buyers rather than as an edge case, suggesting governments and regional operators are becoming a more regular part of Nvidia’s sales mix. That interpretation is drawn from Nvidia’s own segment breakdown and management commentary on the call. (fool.com) ### What did Nvidia say about Singapore? CNBC reported on May 20 that Nvidia plans to launch a research hub in Singapore focused on AI and physical testbeds as the city-state expands its AI push. The report said the site would support work tied to “physical AI,” an area Nvidia has used to describe systems that connect AI models to robotics, simulation and real-world environments. (fool.com) Singapore fits with the company’s broader geographic message. The quarter’s sovereign growth figures and the Singapore plan both point to Nvidia spending time on markets where governments, regional cloud operators and industrial groups are building local AI capacity. That is an inference based on Nvidia’s earnings-call data and CNBC’s report on the research hub. (cnbc.com) ### Is this only a cloud story anymore? Nvidia’s numbers still show hyperscalers remain the single biggest customer bloc. The company said $38 billion of quarterly data-center revenue came from that group alone. But the same quarter also showed $37 billion from ACIE customers, plus edge-computing revenue of $6.4 billion and trailing-12-month physical AI revenue above $9 billion. (fool.com) Jen-Hsun Huang, Nvidia’s chief executive, and Kress have been arguing for several quarters that AI spending is moving into more deployment models. The latest results gave that argument harder numbers: sovereign builds, enterprise systems and AI-native clouds are no longer marginal to Nvidia’s data-center business. ### What comes next from here? (fool.com) Nvidia said production shipments of Vera Rubin are set to begin in the third quarter, with the ramp continuing in following quarters. The company also said total supply commitments, including inventory purchase commitments and prepaids, rose to $145 billion, underscoring how much hardware it expects customers to absorb. (fool.com) The next test will be whether the same customer mix holds as new systems ship. Nvidia’s investor materials for the quarter were posted on May 20, and the company’s next quarterly update will show whether hyperscalers remain at roughly half of data-center revenue or whether sovereign, enterprise and AI cloud buyers take a larger share. (investor.nvidia.com) (fool.com)