Nvidia Splits Data Center Reporting Into Hyperscalers and a New ACIE Bucket to Calm Investor Nerves

Nvidia will start breaking out its data center revenue into two separate lines from this quarter on, a structural reporting change CEO Jensen Huang framed as housekeeping but Wall Street is reading as something more pointed: an attempt to insulate the chipmaker's narrative from a possible hyperscaler spending hangover.

The two new buckets are Hyperscalers — capturing sales to Meta, Amazon, Google, Microsoft, and Oracle — and a new category called ACIE, short for AI Clouds, Industrial, and Enterprise, which sweeps in every other data center customer from neoclouds like CoreWeave to sovereign-AI buyers and Fortune 500 IT departments. Last quarter, hyperscalers were exactly half of Nvidia's data center revenue.

"It's really about the fact that our business has now evolved and grown to such a large scale, it's helpful to segment it," Huang told analysts. The optics, however, line up neatly with a brewing investor anxiety. The same four hyperscalers Nvidia is now isolating have collectively committed about $725 billion to AI capex in 2026 — nearly double last year — while the revenue those buildouts are generating remains thin and concentrated. If the market decides the buildout has run ahead of demand, Nvidia would rather not be the proxy stock that takes the hit.

The numbers also give Huang a real story to tell. In the most recent quarter, hyperscaler revenue grew 12% sequentially while ACIE grew 31%, meaning the slower-growing half of the business is the part everyone is worried about, and the faster-growing half is the part nobody was paying enough attention to. Diversification, in other words, is already underway — Nvidia just wants investors to be able to see it on the income statement.

The change lands at a sensitive moment. Tom's Hardware reported this week that component prices are eating into hyperscaler budgets, with Microsoft alone attributing $25 billion of its AI spend to memory and chip cost inflation. At the same time, every one of Nvidia's biggest customers is now shipping a custom inference accelerator — Amazon's Trainium3, Meta's MTIA, Microsoft's Maia 200 — explicitly designed to peel workloads off Nvidia's roadmap. Splitting the disclosure now lets Nvidia tell two parallel stories: a maturing hyperscaler franchise, and an explosive enterprise and sovereign-AI tier that is only just getting started.