Nvidia crushed Wall Street’s fiscal second-quarter expectations, reporting $96.22 billion in revenue and $2.22 in adjusted EPS while guiding for a record $108 billion of sales next quarter. Yet Nvidia stock initially slipped about 1.2% in extended trading, exposing the extraordinary expectations now embedded in the world’s dominant AI-chip company. Revenue more than doubled from a year earlier, Data Center sales surged 117%, and guidance comfortably topped consensus — but for NVDA investors, a conventional earnings beat is no longer enough.
The results confirm that AI infrastructure demand is still accelerating. At the same time, the muted stock reaction shows that investors are increasingly focused on what happens beyond the headline numbers: the Vera Rubin ramp, gross margins, China, Nvidia’s massive AI-financing commitments and whether hyperscaler spending can remain strong enough to support another year of extraordinary growth.
Nvidia Earnings Blow Past Wall Street Estimates
Nvidia reported $96.22 billion of Q2 revenue, up 18% sequentially and an extraordinary 106% from $46.74 billion a year earlier. Wall Street had expected roughly $92.4 billion, meaning Nvidia beat consensus by almost $4 billion despite already operating at a quarterly revenue scale that would have seemed almost impossible only a few years ago.
Adjusted EPS reached $2.22, comfortably ahead of the approximately $2.09 consensus estimate and more than double the $1.01 earned in the year-earlier quarter on Nvidia’s current non-GAAP reporting basis. Non-GAAP operating income jumped 124% to nearly $64 billion, showing that the company continues converting its AI revenue explosion into enormous incremental profit.
Those numbers exceeded even Nvidia’s own expectations. Management had guided for Q2 revenue of $91 billion, plus or minus 2%, meaning the eventual result landed more than $5 billion above the midpoint. Yet the stock still struggled immediately after the release, reinforcing a pattern investors have seen repeatedly: Nvidia can beat Wall Street and still fail to beat the expectations embedded in its share price.
Related articles
Data Center Revenue Shows the AI Boom Is Still Accelerating
The strongest number in the report may be $89 billion of Data Center revenue, up 18% sequentially and 117% year over year. Wall Street had generally expected Data Center sales closer to the mid-$80 billion range, making this another meaningful beat.
Data Center now represents more than 92% of Nvidia’s quarterly revenue, demonstrating how completely the company’s financial profile has become tied to global AI infrastructure spending. Hyperscalers, AI laboratories, sovereign governments and newer AI-cloud providers are all racing to acquire compute capacity, while increasingly compute-intensive reasoning and agentic AI workloads require far more processing than earlier generations of generative AI.
CEO Jensen Huang said AI has reached an inflection point where compute itself is increasingly tied to revenue generation. Nvidia also said its Vera Rubin platform is now in full production, with systems running at partners including Google Cloud, Microsoft Azure, Oracle, CoreWeave and Nebius.
That helps answer one of Wall Street’s biggest concerns heading into earnings: demand does not appear to be collapsing as Nvidia transitions from Blackwell toward Rubin.
$108 Billion Q3 Guidance Beats Consensus
Nvidia expects fiscal third-quarter revenue of $108 billion, plus or minus 2%, marking the first time the company has guided to quarterly sales above $100 billion. FactSet consensus had been approximately $104.9 billion, so the midpoint represents more than $3 billion of upside.
The growth implied by that forecast remains staggering. Nvidia generated $57 billion in the comparable quarter last year, meaning $108 billion would represent roughly 90% year-over-year growth even as the revenue base approaches a scale usually associated with entire industries rather than individual semiconductor companies.
Yet the guide may explain part of the subdued stock reaction. Some bullish Wall Street analysts had gone into earnings expecting $110 billion or more, with UBS analyst Timothy Arcuri among those arguing that strong Blackwell demand and the beginning of Rubin contributions could push the October quarter above conventional consensus.
That creates Nvidia’s unusual earnings problem. A $108 billion forecast is objectively enormous and comfortably above published consensus, but investors had already begun pricing in unofficial whisper numbers even higher.
Why Nvidia Stock Can Fall After a Huge Beat
Nvidia shares were already coming off a difficult stretch before the report. The stock closed Wednesday at $209.66, down 1.6% during regular trading, following recent weakness that included a seven-session losing streak earlier in the week. Even before earnings, analysts warned that another ordinary „beat and raise“ might not be enough to generate a strong rally.
That is because Nvidia’s execution has become a prerequisite rather than a catalyst. Revenue beats, triple-digit Data Center growth and upward guidance revisions are now expected from a company whose market capitalization is measured in trillions of dollars.
Options traders had priced only about a 5.4% post-earnings move, the smallest implied Nvidia earnings reaction in years and below the company’s roughly 7.4% average move over the previous 12 quarters. That suggested traders increasingly viewed Nvidia’s results as predictable even before the numbers arrived.
The initial 1.2% after-hours decline therefore says less about weak earnings than about an exceptionally high bar. Investors did not find anything obviously wrong with the quarter. They simply did not immediately find a surprise large enough to force another dramatic valuation reset.
Gross Margins Remain Excellent – But Q3 Points Slightly Lower
Profitability remains extraordinary. Nvidia delivered 75.0% GAAP and non-GAAP gross margins in Q2, compared with 72.4% and 72.5%, respectively, a year earlier. The improvement shows that Nvidia retains immense pricing power despite the increasing complexity and cost of full rack-scale AI systems.
For Q3, however, management expects both GAAP and non-GAAP gross margins of 74.0%, plus or minus 50 basis points. That one-percentage-point sequential decline is hardly alarming, but margins have become increasingly important as investors worry about rising memory costs, expensive networking components and the transition toward increasingly sophisticated Rubin systems.
Before earnings, Wall Street had been watching for signs that high-bandwidth memory shortages and other component inflation could put pressure on profitability. Nvidia’s forecast suggests margins remain exceptionally healthy, although investors should not assume every dollar of incremental revenue will arrive with expanding gross profitability.
The next major question is whether margins can stabilize around the mid-70% range as Rubin scales.
Vera Rubin Could Determine Next Growth Leg
The Rubin update may ultimately prove more important than Q2 itself. Nvidia said Vera Rubin is now ramping into full production, with rack systems already operating across several major cloud partners.
A smooth transition matters because major architecture changes can create purchasing pauses if customers postpone current-generation orders while waiting for newer systems. Nvidia appears to be arguing that this is not happening: Blackwell demand remains powerful while Rubin enters production, allowing the company to move from one product cycle into the next without a significant revenue air pocket.
Rubin also arrives as AI models become increasingly computationally demanding. Reasoning systems, autonomous agents and multimodal models require more inference compute than conventional chatbot workloads, potentially expanding Nvidia’s addressable market even without equivalent growth in the number of AI users.
If Rubin ramps smoothly while Blackwell remains supply-constrained, Nvidia’s $108 billion Q3 forecast could once again prove conservative. If customers begin delaying deployments during the transition, however, the unusually optimistic 2027 earnings assumptions behind NVDA stock would face a tougher test.
China Is Still Missing From Nvidia’s Guidance
One potentially enormous source of upside remains deliberately absent. Nvidia said its Q3 outlook assumes no Data Center compute revenue from China, maintaining the conservative approach it adopted amid U.S. export restrictions.
China historically represented a major market for Nvidia’s AI accelerators, and analysts have estimated that restoring meaningful access could add billions of dollars to quarterly sales. Reports ahead of earnings suggested some Chinese technology companies could receive authorization to purchase Nvidia processors, but the regulatory situation remains fluid.
That creates asymmetric potential. Nvidia’s $108 billion forecast does not require a China recovery, so any meaningful reopening could provide incremental revenue. At the same time, investors should not include those sales in base-case forecasts until approvals translate into shipments and recognized revenue.
China therefore remains upside optionality rather than a dependable part of the current growth story.
Nvidia’s AI Financing Push Is the Bigger Long-Term Debate
The results also arrive as investors scrutinize Nvidia’s increasingly aggressive role in financing the broader AI ecosystem. The company recently announced partnerships with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR aimed at mobilizing more than $500 billion of third-party capital for AI infrastructure.
Supporters argue that Nvidia is helping solve the biggest bottleneck in AI adoption: financing the data centers, power infrastructure and computing systems required to deploy its products. Critics worry that Nvidia’s investments and financing relationships could create circular economics in which capital from the chip supplier indirectly helps customers purchase more Nvidia hardware.
Those concerns matter more as Nvidia’s revenue reaches unprecedented scale. Maintaining near-100% growth from a $100 billion quarterly base requires extraordinary amounts of global capital expenditure, and investors increasingly want proof that AI customers are generating sufficient economic returns to sustain that spending without constant financing support.
For now, the Q2 numbers suggest demand remains exceptionally strong. The longer-term return on the AI industry’s massive infrastructure investment remains harder to prove.
Outlook: Nvidia Beat Everything – And Wall Street Still Wanted More
The latest earnings leave little doubt about the underlying business. Revenue soared 106% to $96.22 billion, Data Center sales jumped 117% to $89 billion, adjusted EPS reached $2.22, gross margins held at 75% and management guided to a record $108 billion of Q3 revenue. Nvidia also returned about $26 billion to shareholders during the quarter and still had approximately $99 billion remaining under its repurchase authorization.
What the report does not solve is the expectations problem. NVDA investors now assume massive beats, sustained 70%-plus margins and uninterrupted AI infrastructure demand. Nvidia delivered all three, yet some bullish analysts had already moved their unofficial targets beyond the company’s guidance.
The next catalysts are clear: Rubin’s production ramp, Blackwell demand, gross margins, potential China revenue and evidence that hyperscalers and AI labs can keep financing increasingly expensive infrastructure.
Nvidia just delivered a quarter most companies could only dream of — and the stock still slipped. That may be the clearest sign yet that for NVDA, beating Wall Street is no longer the challenge. Beating what investors secretly expected is.
