Nvidia's artificial intelligence accelerator business serving non-hyperscaler customers is running at approximately $160 billion annually and growing faster than its hyperscale division, according to the company's second-quarter fiscal 2027 earnings guidance.
The ACIE (AI Computing for Enterprise) business unit generated $40.3 billion in quarterly revenue with 25 percent sequential growth, outpacing the hyperscale segment's 13 percent sequential increase, CFO Colette Kress said in earnings commentary. Kress projected that non-hyperscaler revenue will eventually represent roughly half of Nvidia's total data center business, implying a second demand engine approaching the scale of its current hyperscale fortress.
Nvidia's data center revenue derives from two distinct customer bases: the four hyperscale operators, Meta, Amazon, Google, and Microsoft, that have driven the company's explosive growth since 2023, and a broadening set of enterprise, cloud, and regional customers deploying AI infrastructure at scale.
Hyperscale customers remain by far Nvidia's largest segment and continue to represent the bulk of capital deployment in large language model training and inference. The 25 percent quarterly expansion of ACIE outpaces hyperscale's 13 percent pace. Kress's projection that non-hyperscaler revenue could reach parity with hyperscale would represent a structural shift: it would mean Nvidia is no longer dependent on concentration among four customers for sustained growth.
The commentary carries management's own projections and does not specify a timeline for when non-hyperscaler revenue might reach 50 percent of data center. Nvidia does not separately report ACIE revenue in its quarterly filings; the figures cited derive from CFO guidance distributed during earnings.
The disparity in sequential growth rates is 12 percentage points. Both segments are growing double digits quarter-over-quarter, and the company has not provided guidance of a deceleration in either category.
