NVIDIA earnings offered a concise but consequential reading on the scale of AI infrastructure spending this week. The company reported $96.221 billion in revenue for its second quarter of fiscal 2027, which ended July 26, 2026. That was up 18% from the prior quarter and 106% from a year earlier. Its Data Center business generated $89.0 billion, up 117% year over year.
The results, released after the market closed on August 26, do not settle the debate over how durable the AI buildout will be. They do show that spending on the computing systems behind large-scale AI remains unusually concentrated in one company’s data-center line. For readers trying to separate broad claims about artificial intelligence from measurable business activity, that is the central fact of this quarter.
What NVIDIA earnings say about Data Center demand
NVIDIA’s Data Center revenue was the dominant part of its $96.221 billion quarter. The division is where the company records sales of the accelerated-computing systems that cloud providers, AI labs and enterprises use to train and run AI models. The 117% year-over-year increase is therefore more revealing than a general statement about AI enthusiasm: it is a disclosed measure of hardware revenue already recognized in the period.
The company reported GAAP diluted earnings per share of $2.46 and non-GAAP diluted earnings per share of $2.22. Those are different accounting measures and should not be treated as interchangeable. NVIDIA says its non-GAAP presentation is meant to supplement, rather than replace, its GAAP financial statements.
Independent coverage from Investopedia reported that Visible Alpha’s pre-release analyst estimates were $92.37 billion in revenue and $2.10 in adjusted earnings per share. The company’s reported revenue and its non-GAAP per-share figure were above those figures. Those estimates provide market context; the company’s own release remains the source for the reported results.
Guidance is forward-looking—and excludes China data-center compute revenue
For the third quarter of fiscal 2027, NVIDIA forecast revenue of $108.0 billion, plus or minus 2%. The full range matters: it gives a span of possible outcomes rather than a single midpoint. NVIDIA also said that the outlook assumes no Data Center compute revenue from China.
That qualification makes the next quarter’s forecast more specific than a simple demand prediction. It describes the outlook that management is prepared to give without assuming a contribution from that market. It does not, by itself, establish how much revenue NVIDIA might otherwise have earned in China, nor whether conditions will change. But it makes geography and policy part of the operating context readers should keep in view.
Margins are another test. NVIDIA reported both GAAP and non-GAAP gross margin of 75.0% in the second quarter, then guided to 74.0%, plus or minus 50 basis points, for the third. A modest projected change in a gross-margin percentage can still be meaningful at this scale, particularly when the company is navigating product transitions and a supply chain under pressure.
Why this quarter matters beyond one stock
NVIDIA’s results are relevant to a wider question: are the large budgets allocated to AI translating into deployed infrastructure, or are they mainly commitments and narratives? Revenue is not a complete answer. It cannot show whether customers are earning a return on their AI investments, and it says little about the distribution of gains across software companies, cloud providers, developers or workers.
Still, the company’s Data Center result is a useful checkpoint because it reflects systems sold rather than a survey of intention. The next set of results will need to show whether that demand carries through a changing product mix and tight component availability. In a post-results update, Investopedia reported that Chief Financial Officer Colette Kress said memory supply was expected to remain a bottleneck through the end of fiscal 2028. That is management commentary reported by a secondary source, not an independently verified forecast.
The quarter also raises the bar for what comes next. NVIDIA reported that Vera Rubin had entered full production and linked the platform to the current buildout. As the transition progresses, readers should watch for whether revenue guidance is achieved within its stated range, whether gross margins track the company’s forecast, how supply constraints affect deliveries, and whether the China assumption remains in place.
For now, the cautious conclusion is not that every AI investment is justified. It is that a key supplier’s disclosed revenue continues to show extraordinary demand for the physical infrastructure underpinning the AI economy. The durability, economics and geographic reach of that demand remain the more important questions.
