Dive Brief:
- Infrastructure giant Nvidia reported revenue of $81.6 billion for the first quarter of the company’s 2027 fiscal year, up 85% year over year, amid rising demand for AI infrastructure, executives said during the company’s earnings call Wednesday.
- Data center revenue reached $75.2 billion for the quarter, nearly doubling year over year due to sustained demand for Blackwell infrastructure as frontier model developers and hyperscalers deployed hundreds of thousands of Blackwell GPUs, CFO Colette Kress said during the call.
- Nvidia introduced a new reporting framework for its data center revenue, including two subsegments called Hyperscale and ACIE, which includes AI clouds, industrial and enterprise. Hyperscale revenue accounted for half of total data center revenue at $38 billion while ACIE revenue reached $37 billion, up 31% from the previous quarter.
Dive Insight:
Nvidia’s revenue reporting update shows earnings are not fully concentrated in hyperscalers, which the company is increasingly competing against for a bigger slice of the AI infrastructure layer.
Nvidia’s data center revenue was nearly equally divided between its Hyperscale and ACIE subsegments. The company is selling not just chips but integrated rack-scale infrastructure to customers spanning AI clouds, sovereign governments and the enterprise — a customer base that’s getting wider, said The Futurum Group CEO Daniel Newman in an email to CIO Dive.
“The introduction of ACIE as a permanent disclosure tells you the company sees the diversification away from hyperscaler concentration as a feature of the business, not a hedge against it,” Newman said.
The company is positioning itself as the operating layer for AI infrastructure, including across enterprise AI stacks, Scott Bickley, advisory fellow at Info-Tech Research Group, told CIO Dive in an email. Google, Microsoft and AWS are also competing for enterprise spend by investing across AI infrastructure layers — including chips, models and applications. Global AI spend is expected to reach $2.59 trillion in 2026, according to Gartner.
“That matters because the next phase of AI spending will not be driven by model training alone,” Bickley said. “As inference scales across enterprises, governments, and hyperscalers, Nvidia is looking at capturing more of the total AI infrastructure stack.”
Nvidia’s customer base is “diverse and growing,” Kress said during the earnings call. The company’s sovereign revenue grew more than 80% year over year, with Nvidia AI infrastructure deployed across nearly 40 countries, she said.
Part of the driving force behind accelerated AI infrastructure buildouts is the growing adoption of AI native products and services and the mainstream transition to agentic, Kress said, adding that industry adoption of AI is propelling revenue growth across energy, chips, infrastructure, models and applications.
On other fronts, Kress said the company is on track to start shipping its Vera Rubin platform starting in the second half of this year. The platform is expected to deliver up to 35 times higher inference throughput compared with Blackwell, she said.
Additionally, while the U.S. government has approved licenses for Nvidia to ship H200 GPUs to China, Kress said the company has yet to generate any revenue from the exports.
As Nvidia demonstrates a diversified customer base, moves through the Vera Rubin transition, monitors its China re-entry and keeps an eye on its ACIE revenue, the question becomes less about whether the hype cycle is real and more about durability heading into the company’s new fiscal year, Newman said.
“Nvidia is still the central nervous system of the AI economy,” Newman said. “That position looks more entrenched, not less.”
