Nvidia Revenue Hits $96.2B—Data Centers Supply $89B of the Total

On August 26, 2026, Nvidia’s fiscal Q2 2027 release put revenue for the quarter ended July 26 at $96.2 billion, with Data Center supplying $89 billion, and set fiscal Q3 guidance at $108 billion, plus or minus 2%, without assuming any China Data Center compute revenue. Total revenue increased 106% year over year, while Data Center revenue rose 117%.

An Associated Press earnings comparison shows that revenue exceeded the $92.27 billion analyst consensus, adjusted earnings of $2.22 per share beat the $2.09 forecast, and operating expenses increased 55% to $8.41 billion; it also records management’s warning that demand exceeds available supply. Those figures separate the completed quarter from both Wall Street’s prior expectations and Nvidia’s projections for the next one.

Data Center supplied more than nine-tenths of revenue

Data Center accounted for approximately 92.5% of Nvidia’s quarterly sales, calculated from the two headline revenue figures. The company’s other activities therefore supplied roughly $7.2 billion, or 7.5% of the total.

The concentration is more pronounced than the record total alone suggests. Nvidia’s year-over-year expansion now depends overwhelmingly on infrastructure purchased for large-scale AI computing, so changes in data-center demand, production capacity or delivery timing can have an outsized effect on company-wide growth.

The segment is not limited to the largest cloud operators. Fortune’s Data Center breakdown assigns $48.7 billion to hyperscale customers and $40.3 billion to AI clouds, industrial and enterprise customers. The first category contributed about 54.7% of segment revenue, leaving approximately 45.3% for the broader customer group.

This produces two different concentration patterns. At company level, Data Center dominates Nvidia’s revenue base; within the segment, sales are divided comparatively closely between hyperscalers and other AI infrastructure customers. That internal split reduces dependence on one customer category, but it does not change the company’s reliance on Data Center as a whole.

The earnings beat came with faster spending

Nvidia exceeded the cited revenue consensus by approximately $3.95 billion and the adjusted earnings forecast by $0.13 per share. These are differences between actual results and analysts’ pre-release estimates, not additional revenue or profit categories.

Operating expenses rose sharply, though their 55% increase remained well below the 106% rate of revenue growth. The comparison indicates that spending expanded without matching the pace of sales growth during the quarter. It does not establish whether that relationship will persist as Nvidia funds new products, staffing and the operational requirements of a larger business.

The earnings comparison also requires attention to accounting definitions. Adjusted earnings exclude specified items and therefore differ from the company’s GAAP earnings measure; the relevant comparison is the current adjusted result against the consensus prepared on that basis. Mixing the two measures would overstate or understate the size of the earnings beat.

The $108 billion figure is guidance, not booked revenue

Nvidia’s fiscal Q3 midpoint is a management projection for the August-to-October period. Applying the stated tolerance to $108 billion produces an indicative range of approximately $105.84 billion to $110.16 billion, but neither endpoint represents completed sales or guaranteed orders.

The midpoint is $3.14 billion above the $104.86 billion analyst forecast cited in the independent comparison. That is a comparison between two forward-looking estimates, unlike the completed-quarter beat, which measures actual results against expectations set before the release.

The China exclusion is another boundary on the outlook. Nvidia has placed no China Data Center compute revenue in its baseline, so potential sales in that market are not required to reach the guided figure. The assumption does not guarantee that shipments will remain at zero, nor does it settle the future availability of particular products or export permissions.

Guidance therefore combines expected demand with assumptions about regulation, production and delivery. Customer plans cannot become recognized revenue until systems can legally ship and reach the stage at which Nvidia records the sale.

Component supply limits how quickly demand becomes sales

Memory availability is one concrete constraint on the outlook. Axios’s post-earnings account says memory prices were rising faster than Nvidia expected and putting pressure on margins, while management linked the scarcity to the same AI infrastructure expansion driving demand.

The constraint can affect both volume and cost. Nvidia and its manufacturing partners need sufficient memory, packaging and other components to assemble complete systems within a reporting period; a customer’s willingness to buy does not ensure that every requested system can be delivered on schedule.

As of August 28, the completed quarter shows a business in which Data Center contributes about 92.5 cents of every revenue dollar, while the next-quarter target remains a projection bounded by supply and a no-China baseline. The next earnings release will show whether deliveries reached the guided range, how component costs affected margins and whether realized Data Center sales remained consistent with those assumptions.

Also read:

• AMD’s Data-Center Revenue Doubles—and Now Supplies 58% of Sales
• AMD Data-Center Revenue Jumped 107%—Gaming Became the Smallest Segment
• Nvidia Is Bringing a Data Center to Every Home