NVIDIA, the undisputed leader of the global AI supply chain, delivered accelerating revenue growth last quarter, setting yet another quarterly record. The company now guides for its first-ever $100 billion-plus revenue quarter ahead, and while the gross margin outlook came in slightly soft, its next-fiscal-year growth projection crushed consensus estimates, temporarily steadying market sentiment.

On Wednesday after the bell, NVIDIA reported fiscal Q2 2027 (period ending July 26, 2026) revenue of $962.21 billion, up 106% year-over-year and more than 4% above analyst forecasts. Adjusted EPS on a non-GAAP basis came in at $2.22, up 120% year-over-year and nearly 6% ahead of expectations.

AI data centers remained the undisputed growth engine. Data center revenue surged 117% year-over-year to $890 billion in Q2, roughly 4% above analyst estimates and representing over 90% of total company revenue. Compared with the prior quarter's growth rates, both total revenue and data center revenue accelerated, with actual Q2 results landing more than $50 billion above the midpoint of the company's own guidance.

Looking ahead to fiscal Q3, NVIDIA guides for revenue of $108 billion at the midpoint, up nearly 90% year-over-year and above the $105.15 billion analysts had penciled in. This implies the company will almost certainly formally enter the era of $100 billion-plus quarterly revenue. At the same time, management guided non-GAAP gross margin to 74.0% at the midpoint, below the 75% consensus, with non-GAAP operating expenses rising to roughly $9 billion.

Following the print, NVIDIA shares initially ticked up in after-hours trading before quickly reversing lower, at one point falling 4%. Analysts noted the post-earnings dip does not signal a weak report or a reversal in the AI demand narrative. The Q3 gross margin guide came in slightly light, revenue guidance did not exceed the most bullish estimates on the Street, and the outlook excludes China data center compute revenue, making the guide more conservative and preventing a potential China recovery from serving as a definitive upside driver in the near term.

During the subsequent earnings call, management stated that demand is still accelerating even at current scale, projecting fiscal 2028 revenue growth of roughly 70%, well above the approximately 45% analysts had modeled. Amazon also committed to significantly expanding its adoption of NVIDIA products, adding an extra 2 million GPUs. NVIDIA shares turned higher after hours, at one point gaining nearly 5%.

Heading into the print, NVIDIA shares had already climbed more than 10% this month, reflecting substantial optimism. For a core AI asset like this, the market trades on whether results can meaningfully exceed buy-side expectations, not simply clear sell-side consensus estimates.

Revenue and profit beat across the board; Q2 booked nearly $7.8 billion in net equity securities gains

NVIDIA's Q2 revenue of $962.21 billion doubled year-over-year at 106%, the fastest annual growth rate since fiscal Q2 2025 (quarter ending July 2024), and rose 18% sequentially, underscoring that AI infrastructure demand remains in a rapid expansion phase. The company had guided Q2 revenue to $910 billion at the midpoint in its prior quarter's report, meaning actual results came in roughly $52 billion higher.

Profitability was equally robust. Non-GAAP adjusted net income reached $539.54 billion in Q2, up 118% year-over-year and 18% sequentially, with EPS of $2.22 versus the $2.09 consensus. Non-GAAP operating profit hit $639.56 billion, up 124% year-over-year and above the $611.9 billion analysts expected, corresponding to an operating margin of roughly 66%, still at exceptionally elevated levels.

On margins, both GAAP and non-GAAP gross margin came in at 75.0% in Q2, essentially in line with expectations and up about 2.5 to 2.6 percentage points from a year ago. For a company generating close to $100 billion in quarterly revenue, sustaining roughly 75% gross margin speaks to NVIDIA's continued pricing power in the AI accelerated computing market.

Additionally, adjusted operating expenses of $82.32 billion came in below the $83.2 billion consensus.

These figures indicate NVIDIA is not only beating on the top line but also managing costs better than expected.

Notably, Q2 "other income" included approximately $77.7 billion in net gains from equity securities, which are excluded from non-GAAP profit. GAAP EPS of $2.46 exceeded the non-GAAP figure of $2.22 primarily because GAAP incorporates this nearly $7.8 billion equity securities gain. Stripping out this investment income, non-GAAP net income of $539.54 billion better reflects core operating performance.

Data center revenue accelerated 117% year-over-year, representing over 90% of total revenue

Q2 data center revenue reached $890 billion, up 117% year-over-year and 18% sequentially, above the roughly $85.8 billion to $85.9 billion analysts had projected. This segment now contributes approximately 92.5% of total revenue, cementing NVIDIA's deep alignment with the global AI infrastructure buildout cycle.

CEO Jensen Huang emphasized in the earnings release: "AI has reached the inflection point. It is doing useful work, and tokens are becoming productive and profitable. Compute is now revenue."

He also noted that a year ago, primarily one AI lab was driving the buildout; now multiple frontier labs are expanding in parallel, the open-source model ecosystem is thriving, and physical AI is coming online.

The report disclosed that the Vera Rubin platform is in full production ramp, with racks already running at partners including CoreWeave, Google Cloud, Microsoft Azure, Oracle Cloud Infrastructure, and Nebius. NVIDIA's Spectrum-6 switching system, Vera CPU, and DSX platform are also central to the "AI factory" narrative, signaling a shift from selling individual GPUs to system-level delivery encompassing full racks, networking, CPUs, software, and deployment solutions.

Within data center sub-segments, hyperscale revenue, the most closely watched metric, came in at $487.1 billion, above the $435.5 billion estimate, confirming that large cloud providers and hyperscale customers remain the primary drivers of current AI infrastructure investment. Compute & Networking revenue reached $883 billion, also beating the $846.9 billion expectation.

However, not every data center sub-segment exceeded. The ACIE business, covering AI cloud, industrial applications, and enterprise deployments, generated $403.1 billion, below the $419.6 billion market consensus. This suggests that while overall data center demand remains robust, this quarter's beat was primarily fueled by hyperscalers, while enterprise, industrial, and some AI cloud-related demand expanded at a slightly slower pace than markets had anticipated.

By contrast, edge computing revenue reached $72 billion in Q2, up 27% year-over-year and 13% sequentially, above the $66.1 billion estimate.

This segment spans RTX, on-device Windows AI, robotics, and autonomous driving. Though far smaller in scale than data centers, it sits at the core of NVIDIA's "physical AI" narrative. While still growing, it is unlikely to alter the company's heavy reliance on data center revenue in the near term.

Q3 revenue guidance shows deceleration; did not exceed most bullish expectations

NVIDIA guided Q3 revenue to $108 billion, plus or minus 2%, implying a range of roughly $105.8 billion to $110.2 billion, equivalent to year-over-year growth of 85.6% to 93.3%, with the midpoint up 89.5%. Even at the low end of the range, the company would enter the $100 billion quarterly revenue club.

This guidance came in above the consensus forecast. Various firms had Q3 estimates clustered around $104 billion to $105 billion, meaning the $108 billion midpoint still represents a roughly 3% to 4% upward revision.

But this was also a key reason shares came under pressure after hours: the market's "true bar" for NVIDIA may have been set higher than the consensus.

Before the report, the most optimistic estimate in LSEG's compilation reached $112 billion. Against those loftier buy-side expectations, the $108 billion guide, while strong, did not open up additional upside imagination.

Furthermore, the Q3 midpoint implies roughly 12% sequential growth, below Q2's 18% quarter-over-quarter increase. As the base rapidly expands, slower growth is hardly surprising, but for a high-multiple stock, marginal changes in sequential momentum get amplified by the market.

Gross margin stable, but Q3 guide ticks slightly lower

The 75.0% gross margin in Q2, flat with Q1, was a solid result, especially given the company's near-$100 billion revenue scale and the highly strained data center supply chain. However, Q3 gross margin guidance of 74.0%, plus or minus 50 basis points, represents a roughly one-percentage-point step-down.

This is not a severe deterioration. Margins around 74% remain at the extreme high end of the global semiconductor industry and demonstrate NVIDIA's continued pricing leverage. But for investors, the question is whether gross margin is approaching a cyclical peak.

Several factors could influence future margins: the full production ramp of Vera Rubin may introduce early-stage cost and yield volatility; HBM, advanced packaging, and substrate supply chain costs continue to rise; and the rising share of full-rack system deliveries may carry a different product mix than standalone GPU sales.

Thus, the market's sensitive after-hours reaction to the Q3 margin guide does not mean profitability is deteriorating, but rather that NVIDIA's stock already embeds elevated expectations of "high growth plus sustained or improving gross margins."

On the expense side, the company expects Q3 GAAP operating expenses of approximately $9.2 billion and non-GAAP operating expenses of about $9.0 billion, both above Q2's $84.08 billion and $82.32 billion. As Vera Rubin ramps into production, networking products expand, and the AI software and "AI factory" ecosystem advances, R&D and operational investments continue to accelerate.

Rubin enters full production: the next growth narrative has begun

The most important product signal in the report is that Vera Rubin is accelerating into full production. This indicates NVIDIA is shifting market attention from the Blackwell cycle toward the Rubin cycle.

Based on company disclosures, Rubin is not a single chip but an entire AI factory architecture encompassing the Vera CPU, Rubin GPU, Spectrum-6 networking, BlueField, security, storage, software toolchain, and the DSX platform. NVIDIA repeatedly uses the phrase "AI factory" to emphasize that it supplies not just chips but the complete system for building, operating, and scaling AI compute infrastructure.

This matters for the valuation narrative. Markets had worried whether NVIDIA could smoothly transition to the next platform after the Blackwell demand peak. If Rubin ramps smoothly as management suggests and deploys quickly across major cloud providers and AI infrastructure firms, it would help extend the growth cycle.

But product transitions also bring near-term uncertainty. New platform ramps typically involve changes in supply chain, delivery cadence, customer qualification, and cost structure. The lower Q3 gross margin guide may also steer some investors' attention toward margin impact during Rubin's early production phase.

China revenue excluded from guidance: reduces forecast risk but also caps near-term upside imagination

NVIDIA explicitly stated that its Q3 guidance "does not assume any data center compute revenue from China." This is a crucial detail.

From a positive standpoint, this means the $108 billion revenue guide does not include China data center compute contribution. If regulatory conditions, export licensing, or product arrangements improve, China could serve as an additional upside factor.

But from a trading perspective, it also reminds the market that China data center compute revenue remains uncertain in the near term. For a company deriving over 90% of revenue from data centers, any regional restriction gets priced into risk by investors.

Therefore, excluding China revenue makes the guidance "cleaner" and more conservative, but it also prevents the market from baking in a potential China recovery as a certainty in Q3 growth.

Cash flow and capital returns: buybacks robust, but working capital absorption clearly rising

Q2 free cash flow came in at $213.41 billion, still a highly impressive figure, though down sharply from Q1's $485.54 billion. The primary reason is not deteriorating profitability but increased working capital absorption.

The cash flow statement shows accounts receivable up $223.46 billion, inventory up $57.84 billion, and prepayments and other assets up $54.97 billion in Q2. This reflects rapidly expanding business scale, customer deliveries, and supply chain stocking, but it also means the conversion of profits to cash did not scale proportionally this quarter.

At quarter-end, the company held $224.43 billion in cash and cash equivalents, $1 billion in short-term debt, and $323.66 billion in long-term debt. During Q2, NVIDIA returned approximately $26 billion to shareholders through buybacks and dividends, including roughly $19.7 billion in repurchases and about $6 billion in dividends. As of quarter-end, approximately $99 billion remained under its buyback authorization.

Capital returns were not weak by themselves, but some market participants had hoped NVIDIA would adopt an even more aggressive Apple-style buyback program. This report disclosed remaining authorization and routine dividends but did not offer a more aggressive capital return commitment beyond existing authorization, which may have disappointed some investors.

Ecosystem investments and financing arrangements address AI infrastructure bottlenecks; CFO responds to "circular financing" concerns

NVIDIA announced strategic partnerships with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR, aiming to mobilize more than $500 billion in third-party capital for AI infrastructure construction, subject to final agreements.

From an industry perspective, this is NVIDIA's attempt to resolve AI infrastructure bottlenecks: compute demand requires not just GPUs but also power, land, data center shells, networking, cooling, and long-term financing. The company also disclosed a collaboration with SB Energy at the PORTS-Pike Technology Campus in Ohio, securing land, power, and building shell capacity to host NVIDIA compute resources.

But this is also where the market diverges most. Investors will ask: how much of future AI infrastructure demand comes from customers' organic capital expenditure versus supplier, private capital, and credit support working in concert? Could NVIDIA take on potential credit support, residual value risk, or long-term obligations?

CFO Colette Kress defended the company's aggressive AI investment strategy on the earnings call. She stated that NVIDIA expects top frontier AI labs like OpenAI "will become the largest technology companies in history."

Kress noted these labs face bottlenecks in accessing all the compute they need to develop products and optimize models, which is precisely why they need NVIDIA's financial support. "We understand the scale of this support, and we understand that some may call it 'circular financing.' But we see it differently. The capital we deploy will generate excellent equity returns."

On the balance sheet, NVIDIA's investment assets are growing rapidly.

At the end of Q2, marketable equity securities stood at $427.83 billion, with non-marketable securities at $511.57 billion, totaling roughly $93.9 billion, up significantly from the start of the year. The company recorded approximately $77.7 billion in net equity securities gains in Q2 and roughly $237 billion cumulatively over the first six months. These investments contribute to GAAP profit but also heighten market scrutiny of the relationship between ecosystem investments and core revenue.