NVIDIA’s $96.2B Q2 proves the AI factory boom is still accelerating

The headline number: $96.2 billion, but not a simple sequential doubling

NVIDIA’s latest results redraw the scale of the AI infrastructure cycle. For the fiscal second quarter ended July 26, 2026, the company reported revenue of $96.221 billion, up 106% from the same quarter a year earlier and up 18% from the immediately preceding quarter [1]. That distinction matters: NVIDIA did not double revenue sequentially; it more than doubled revenue year over year while still adding $14.606 billion in quarterly sales in just three months [1].

The profitability attached to that growth remains striking. GAAP gross margin was 75.0%, GAAP operating income reached $63.734 billion, and GAAP net income was $59.688 billion, or $2.46 per diluted share [1]. On a non-GAAP basis, diluted EPS was $2.22, up 120% year over year and 19% sequentially [1]. For a company already operating at extreme scale, those margins show that demand is not being bought through price concessions; the mix, especially in data center infrastructure, is still supporting premium economics.

The quarter also beat external expectations. Reuters, citing LSEG data, reported that second-quarter revenue of $96.22 billion exceeded analysts’ average estimate of $92.17 billion [3]. AP put the consensus comparison in similar territory, saying revenue surpassed analysts’ average forecast of $92.27 billion [4]. In other words, the market expected exceptional growth and NVIDIA still cleared that bar by roughly $4 billion.

Data center is now the company

The center of gravity is unmistakable. Data Center revenue was $89.023 billion, up 117% from a year earlier and 18% from the prior quarter [2]. That one line represented roughly 92.5% of NVIDIA’s total quarterly revenue, making the company less a broad semiconductor supplier than the operating system and hardware supplier for the global AI buildout.

The more revealing detail is the split inside Data Center. NVIDIA said Hyperscale revenue was $48.710 billion, up 102% year over year and 13% sequentially, while AI Clouds, Industrial, and Enterprise revenue was $40.313 billion, up 138% year over year and 25% sequentially [2]. The hyperscalers remain the largest bucket, but the faster growth came from the second category: AI-native clouds, enterprises, sovereign customers, industrial buyers and hyperscalers using external AI clouds [2].

That mix supports Jensen Huang’s broader message that the AI infrastructure boom is no longer concentrated in one laboratory, one customer type or one cloud cycle. NVIDIA’s release quoted the CEO saying demand is accelerating as multiple frontier labs, startups, open-model developers and physical-AI use cases scale in parallel [1]. The numbers give that claim weight: ACIE’s 25% sequential growth is faster than the hyperscale line, suggesting demand is broadening rather than merely repeating the same megacap cloud purchasing pattern [2].

China remains the caveat. NVIDIA said shipments of Data Center Hopper products to China were less than 1% of Data Center revenue in the quarter [2]. The company also said its third-quarter outlook assumes no Data Center compute revenue from China [1]. That means the $108.0 billion guidance is not dependent on a near-term reopening of that channel, but it also leaves upside or policy risk outside the base case.

Blackwell Ultra now drives the engine

Management attributed the record Data Center result to the ramp of Blackwell Ultra infrastructure [2]. The CFO commentary said Data Center growth was “driven by the ramp” of Blackwell Ultra, and that hyperscale revenue more than doubled from a year earlier on the strength of the same platform [2]. Edge Computing, the smaller segment, also grew to $7.198 billion, up 27% year over year and 13% sequentially, helped by Blackwell workstation sales but partly offset by slower consumer PC demand and elevated memory and systems prices [2].

The product transition is therefore doing two things at once. In data centers, it is expanding the revenue pool through full-stack infrastructure sales rather than only accelerator cards. At the edge, it is helping sustain growth in workstations, local AI machines, robotics and automotive platforms, even as consumer PC conditions remain less robust [2].

NVIDIA’s release also placed Vera Rubin in the near future of that cycle, saying the platform was ramping into full production with racks running at partners including CoreWeave, Google Cloud, Microsoft Azure, Oracle Cloud Infrastructure and Nebius [1]. The company said inventory rose to $31.6 billion from $25.8 billion sequentially as it prepared for the introduction of Vera Rubin in the third quarter [2]. That inventory build is a signal of confidence, but it also raises the stakes: NVIDIA is committing balance-sheet capacity before the next architecture reaches full commercial run-rate.

The outlook: $108 billion next quarter, but a slightly lower margin guide

NVIDIA guided fiscal third-quarter revenue to $108.0 billion, plus or minus 2%, again with no assumed Data Center compute revenue from China [1]. Reuters reported that analysts’ average estimate for the quarter was $104.19 billion, meaning the guide came in comfortably above consensus [3]. The implied sequential increase at the midpoint is about $11.8 billion, or roughly 12.2%, from Q2’s $96.221 billion.

The margin guide is more nuanced. NVIDIA expects GAAP and non-GAAP gross margins of 74.0%, plus or minus 50 basis points, compared with the 75.0% reported in Q2 [1]. Operating expenses are expected to rise to about $9.2 billion on a GAAP basis and $9.0 billion on a non-GAAP basis [1]. The guidance still implies immense operating leverage, but it also suggests the next leg of growth will require heavier spending and may absorb some supply-chain cost pressure.

Investors reacted in the way investors often react to NVIDIA: first by testing the ceiling, then by debating whether the ceiling was high enough. AP reported that NVIDIA shares rose 4.1% in after-hours trading following the earnings call, while also noting that investors remain concerned about a possible comedown after a multiyear surge that lifted NVIDIA’s market value to roughly $5.2 trillion [4]. Reuters described the report as strong but said the outlook did little to end concerns about NVIDIA’s long-term dominance in AI chips [3].

The balance sheet is becoming part of the product

The most important part of the report may not be the income statement. NVIDIA is increasingly using its balance sheet to secure supply, finance infrastructure and support customers whose demand is growing faster than their own credit profiles.

The CFO commentary said supply and capacity commitments rose from $119 billion last quarter to $279 billion, primarily for memory procurement [2]. Total disclosed future commitments were $366 billion, including $279 billion of supply and capacity, $29 billion of cloud service agreements, $25 billion of data center leases not yet commenced, $25 billion of equity investments and $8 billion of capital expenditures [2]. These are not ordinary semiconductor purchase orders; they are the financial architecture behind the AI factory cycle.

NVIDIA also disclosed additional AI cloud agreements and third-party data center lease commitments totaling $56 billion [2]. More dramatically, it reported maximum gross guarantee exposure of $108.5 billion, including $105.0 billion tied to SB Energy Corp. guarantees for land, power and shell buildout at the PORTS-Pike Technology Campus in Ohio [2]. The same commentary said the site is expected to host NVIDIA infrastructure under 20-year leases to OpenAI, subject to limited exceptions, and that the guarantees become effective in phases beginning with data centers expected to be ready for service in fiscal 2029 [2].

This is where NVIDIA’s model becomes more complex. It is no longer only selling accelerators into an arms race; it is helping assemble the racecourse. The company said AI clouds and model makers are seeing extraordinary demand but are sometimes growing faster than their balance sheets and long-term credit profiles can support [2]. NVIDIA’s answer is to help selected customers secure land, power and data center capacity, while earning upfront revenue from infrastructure sales and, where criteria are met, participating in revenue share from third-party customers of those AI clouds [2].

The opportunity is obvious: if customers cannot finance capacity fast enough, NVIDIA can reduce the bottleneck and pull forward demand. The risk is also obvious: supplier commitments, customer financing, guarantees, leases and equity investments make the AI cycle more circular and more capital intensive. Accounts receivable rose to $63.1 billion, and days sales outstanding increased to 60 from 45 sequentially because of extended payment terms on large, multi-quarter agreements with certain investment-grade customers [2]. NVIDIA also issued $25.0 billion of senior unsecured notes in the quarter for general corporate purposes [2].

Shareholders still get paid

Even while expanding commitments, NVIDIA returned nearly $26.0 billion to shareholders through repurchases and dividends in the quarter [1]. The company ended Q2 with about $99.0 billion remaining under its share repurchase authorization and declared a $0.25 quarterly cash dividend payable on October 1, 2026, to holders of record on September 10, 2026 [1].

That combination is rare: a company funding massive supply commitments, customer-enabling infrastructure arrangements and large buybacks while still holding $56.6 billion in cash, cash equivalents and marketable debt securities [2]. It reflects the cash-generating power of a 75% gross-margin business growing at triple-digit year-over-year rates. It also leaves investors with a central question: how much of NVIDIA’s future demand is organic end-user pull, and how much is now being accelerated by NVIDIA’s own financing ecosystem?

The takeaway

The Q2 FY27 report confirms that NVIDIA remains the indispensable vendor in the current AI infrastructure buildout. Revenue more than doubled year over year, Data Center reached $89.0 billion, and guidance points to another record quarter at $108.0 billion [1]. But the story has moved beyond chip demand alone.

NVIDIA is turning into the central allocator of AI infrastructure capacity: securing memory, backing power and land, supporting AI clouds, preparing Vera Rubin and helping customers absorb the cost of the next wave [2]. That strategy could extend its lead if AI demand keeps compounding. It could also increase scrutiny if the market begins to question whether infrastructure growth is being financed too aggressively.

For now, the quarter says the boom is still real, still broadening and still profitable. The next test is whether NVIDIA can convert a record backlog of commitments into another generation of high-margin revenue without turning the AI supply chain into a balance-sheet risk of its own.

Sources from the last 72 hours

1. [1]NVIDIA Announces Financial Results for Second Quarter Fiscal 2027Aug 26, 2026, 8:00 PM UTC
2. [2]Nvidia forecasts quarterly revenue above estimatesAug 26, 2026, 8:24 PM UTC
3. [3]Strong AI chip demand fuels Nvidia's Q2 results well beyond Wall Street's expectationsAug 26, 2026, 8:40 PM UTC
4. [4]CFO Commentary on Second Quarter Fiscal 2027 ResultsAug 26, 2026, 8:22 PM UTC

AI-generated article based on recent web research, then preserved as a dated editorial snapshot.