NeuroAI NEUROAINEUROAI.SITE
China AI, reported clearly
1,300+ verified stories on models, robotics, BCI and the business of AI.
Browse stories
ESC

Nvidia's New Cloud Business Grew 18%. Last Quarter It Grew 25%.

Nvidia's AI cloud, industrial and enterprise segment grew 18% last quarter, after 25% in the one before, and missed consensus by $1.4 billion.

2026-11-19 · 3927 words · NeuroAI
Nvidia's New Cloud Business Grew 18%. Last Quarter It Grew 25%.

Consensus had $48.9 billion for the segment Nvidia calls ACIE. It delivered $47.5 billion, and that gap is the most informative number in the report.

A $1.4 billion shortfall is not a catastrophe in a quarter that produced $115.8 billion of revenue. It is a signal, because the company began breaking out this split only this fiscal year, and the split now makes a pattern visible.

Eighteen, after twenty-five

ACIE covers AI cloud, industrial and enterprise customers, sovereign accounts among them. Revenue was $47.5 billion, up $7.2 billion from the prior quarter, a gain of 18%. In the quarter before, the same segment grew 25%. New cloud providers account for more than half of it.

The miss is the first hard reading on the half of Nvidia's business that is not hyperscale. The bullish case for the stock has rested on diffusion, the idea that AI capacity is spreading from a few buyers to many. This quarter, the many grew more slowly than the few.

The other half, and the arithmetic

Hyperscale revenue was $60.1 billion, up $11.4 billion sequentially, or 23%, against expectations of $53.3 billion. Add the two segments and you get $107.6 billion, the data center total, up 21% on the quarter.

Of the $18.6 billion sequential increase in data center revenue, hyperscale supplied $11.4 billion. That is about 61% of the growth, from a customer group that was already the larger half.

A concentration number

Sixty-one percent is the number to keep.

It says that the sequential acceleration in Nvidia's core business came mostly from the buyers who were already spending the most, at a time when the next tier down was slowing and missing estimates.

That is not a demand problem. It is a demand-shape problem, and it matters for how the next four quarters get valued.

The credit valve closes

On the same call, management described a change in how it backs customers. Nvidia will no longer provide data center guarantees during the construction phase, only once a facility is ready, and a guarantee lapses automatically if construction slips. The approach to new cloud providers is now described as selective and more prudent, with each deal reviewed on its own.

Kress put it plainly: "We will look very carefully at which companies deserve financing support."

The balance sheet numbers behind that sentence are large. Commitments tied to emerging cloud agreements total $32 billion. Total guarantees at quarter end were $118 billion. Purchase commitments, the supply Nvidia has promised to buy, reached $390 billion, more than three times quarterly revenue.

Tightening into a slowdown

Here is where the two facts collide. The segment that missed is the segment whose customers lean most on Nvidia's balance sheet, and Nvidia is now tightening the terms on which that support is available.

That can be read two ways, and the report does not settle which. Either the company is pulling credit from a cohort whose projects it no longer trusts, which is prudence arriving late. Or it is withdrawing the financing that made part of that demand possible, which would turn a deceleration into something steeper.

Both readings point the same direction for the next two quarters. The second tier has to fund itself.

Margins pay for the ramp

Gross margin was 74.9% on a non-GAAP basis, against expectations of about 74%. Guidance for the current quarter drops to 71% to 72%.

A ramp explains most of that. Vera Rubin shipped in volume for the first time and contributed close to $20 billion, about 18% of data center revenue, while Blackwell Ultra continued to ship. New platforms always cost margin before they earn it. Networking passed $21 billion, roughly a fifth of the segment, with co-packaged optics scale-out products in volume production.

Seventeen in a row

Revenue rose 103% year over year. It was the 17th consecutive quarter above consensus. The cloud backlog Nvidia tracks moved from $2 trillion to $2.5 trillion in three months, and management raised its fiscal 2028 growth expectation to about 75%.

Those are the numbers that will be quoted. The one that should be is 18%.

The test for the next print

If ACIE grows below 20% again while hyperscale stays above 20%, Nvidia will be a concentration story wearing a diffusion story's valuation. If ACIE reaccelerates without the old guarantee regime, the diffusion argument survives on its own evidence.

One quarter cannot decide that. One quarter can set the question, and this one did. The company's largest, most profitable business is still accelerating. The part that was supposed to become the next one is not.

*Disclaimer: This article is based on publicly available information and is for reference only. It does not constitute investment advice.*

Related coverage

Cite this article

NeuroAI (脑机网), "Nvidia's New Cloud Business Grew 18%. Last Quarter It Grew 25%," 19 November 2026, https://neuroai.site/a/na-chips-nvidia-acie-slowdown-1119b

Machine-readable: BibTeX · RIS · JSON · Markdown

More in “Chips & Compute” → · Back to home · Markdown version

China AI, reported clearly
1,300+ verified stories on models, robotics, BCI and the business of AI.
Browse stories