Nvidia's 'Compute Is Revenue' Surge and Risks: Breaking Down the Latest Earnings

On Wednesday, Aug. 26, Nvidia reported the strongest quarterly results in semiconductor industry history. Revenue was $96.2 billion, data center revenue was $89 billion, adjusted earnings per share was $2.22, beating expectations by over 6%, and Q3 revenue guidance was $108 billion, far above analyst forecasts. Subsequently, the stock rose about 4% in after-hours trading. This report will explain what these numbers actually mean, why gross margin is the core issue investors are most focused on, and what to watch next.

Key data: FY2027 Q2 revenue $96.22 billion, up 106% year-over-year · Data center revenue $89.02 billion, up 117% year-over-year · Adjusted EPS $2.22, above consensus estimate of $2.09 · Q3 revenue guidance $108 billion, above consensus estimate of about $104 billion · Q2 gross margin 75%, Q3 guidance 74%, Q4 expected to further decline to 71%-72% · Supply commitments surged from $119 billion to $279 billion · After-hours stock price rose about 4% to $218 · Returned $26 billion to shareholders this quarter

Section 1 — The Actual Numbers

Let's first explain in plain language what Nvidia actually reported.

For the quarter ended July 26, 2026, revenue was $96.22 billion, up 106% year-over-year, up 18% sequentially from Q1's $81.61 billion, and about 4.5% above the analyst consensus estimate of around $92.1 billion. A single-quarter revenue of $96.2 billion is higher than the annual GDP of many countries and exceeds Nvidia's entire fiscal 2023 annual revenue.

Data center revenue was $89.02 billion, up 117% year-over-year and up 18% sequentially. Data center now accounts for about 93% of Nvidia's total revenue. Gaming, automotive, professional visualization, and edge computing combined were only $7.2 billion, almost negligible compared to the $89 billion data center business.

Adjusted earnings per share was $2.22, beating the consensus estimate of $2.09 by about 6.2%. GAAP earnings per share was $2.46, which included a $7.8 billion equity investment gain (holdings include Intel and SpaceX). GAAP net income reached $59.7 billion, up 126% year-over-year. Gross margin remained at 75% on both GAAP and non-GAAP bases, flat from the previous quarter and up 250 basis points from 72.5% a year ago.

Q3 revenue guidance was $108 billion (plus or minus 2%), while analyst consensus was around $104 billion, exceeding by about $4 billion—this forward-looking number is what the market is most focused on.

Nvidia founder and CEO Jensen Huang declared in the press release: "AI has reached an inflection point. It is completing valuable work. Every token it produces is productive and profitable. Now, compute is revenue."

Educational note: Nvidia's fiscal year runs ahead of the calendar year. When the company reports "FY2027 Q2" results in August 2026, it refers to the three months ended July 26, 2026. This timing convention is worth remembering: financial media often call it "August 2026 earnings," while Nvidia itself labels it fiscal 2027, and both refer to the same set of numbers.

Section 2 — Where the Growth Comes From

The Nvidia story in 2026 ultimately boils down to a single driver: all major global tech companies are racing to build AI infrastructure, and Nvidia makes the chips they cannot do without.

The four major hyperscalers—Alphabet, Amazon, Meta, and Microsoft—had combined capital expenditures of $166 billion in the quarter ended June 2026, up 87% year-over-year and up 27% sequentially. Over the ten quarters since early 2024, the four companies' combined capital expenditures have grown by 272% cumulatively. CFO Colette Kress disclosed that the top five hyperscalers' capital expenditures are expected to grow from about $800 billion in 2026 to about $1.3 trillion in 2027. Much of this spending flows to Nvidia's data center business.

Within data center, hyperscalers contributed $48.7 billion in revenue, up 102% year-over-year. But the more striking growth came from elsewhere—AI cloud, industrial, and enterprise customers (ACIE) contributed $40.3 billion, up 138% year-over-year and up 25% sequentially. The customer base is genuinely diversifying: AWS plans to deploy an additional 2 million Nvidia GPUs, SpaceX said it will build its AI infrastructure entirely on Nvidia, and government-led sovereign AI projects have become a non-negligible revenue source.

One of the most noteworthy numbers in the entire earnings report, however, is not a revenue item. Supply commitments surged from $119 billion at the end of Q1 to $279 billion at the end of Q2—more than doubling in a single quarter—mainly related to memory procurement for the Vera Rubin platform. This locks in a large amount of future revenue that has not yet shipped and is the clearest proof to date: the demand Nvidia sees is not a forecast, but a committed order book.

The next-generation platform Vera Rubin began volume production and shipment on schedule in Q3, expected to account for about 20% of Q3 data center revenue. The CFO guided fiscal 2028 revenue growth of about 70%, explicitly stating this is a result of supply constraints. Jensen Huang said: "Although demand is far more than 70%, our supply gives us confidence to achieve 70%."

Section 3 — Gross Margin: Why It's the Most Watched Number

After hours, when the full picture—revenue beat, EPS beat, $108 billion guidance, $279 billion supply commitments—was presented to investors, the stock rose about 4% to around $218. However, amid this positive reaction, the most debated topic was not top-line revenue, but the gross margin trajectory.

Nvidia guided Q3 gross margin to 74%, down 1 percentage point from the just-achieved 75%. At a revenue scale of $108 billion, a 1 percentage point decline in gross margin means a reduction of over $1 billion in quarterly gross profit. The CFO further guided Q4 gross margin to 71% to 72%.

The reason behind this is memory costs. Management explicitly stated that memory cost increases have exceeded previous expectations and are expected to be "even higher" next year. Of the $279 billion in supply commitments, most are related to memory procurement—the price of high-bandwidth memory (HBM) embedded in every Nvidia GPU continues to climb. Nvidia earlier this month notified customers that it will raise AI server prices by more than 15% to cope with soaring memory costs, but cost increases have outpaced the price pass-through to customers, putting large-scale pressure on gross margin.

This directly echoes what we documented in the memory supercycle report and the SK Hynix report. SK Hynix achieved a 72% operating margin in Q1 2026, partly because Nvidia and its customers have no choice but to buy HBM at ever-rising prices. One company's cost pressure is another company's revenue tailwind.

Why does gross margin matter so much when revenue is still growing strongly? A simple example illustrates. If Nvidia's revenue grows from $96 billion to $108 billion, but gross margin falls from 75% to 74%, gross profit grows from about $72 billion to about $79.9 billion—still growing, but investors who modeled $108 billion revenue at 75% gross margin originally expected about $81 billion in gross profit. A gap of over $1 billion per quarter, compounding over future quarters, is not negligible at Nvidia's scale, which explains why analysts asked many questions about gross margin on the earnings call, even though revenue guidance beat expectations by about $4 billion.

Educational note: Gross margin is the ratio of revenue minus direct product costs, expressed as a percentage of revenue. For every $1 of revenue Nvidia collected in Q2, it retained $0.75 after paying for components, manufacturing, and assembly. When this ratio drops to $0.74 in Q3 and possibly to $0.71 to $0.72 in Q4, a few cents per dollar translates to billions of dollars at Nvidia's revenue scale. Hardware companies with 75% gross margins are extremely rare—Apple reaches 49%, and most chip companies are in the 50% range. The key question investors now have is: where is the bottom for this number.

Section 4 — What the Broader Numbers Tell You

Stepping back from the quarterly numbers, several structural facts about Nvidia's business become clear.

Business is highly concentrated in AI infrastructure. Data center accounts for about 93% of total revenue, making Nvidia effectively a single-product company at the earnings driver level. If AI infrastructure spending slows for any reason, the revenue impact would be immediate and significant.

Customer concentration is real but improving. Hyperscalers remain the largest buyers, contributing $48.7 billion; but ACIE is catching up with $40.3 billion and faster growth, a positive sign of structural improvement. A more diversified customer base will reduce the impact of any single buyer's decisions on Nvidia's performance.

Supply, not demand, is the constraint on growth. Jensen Huang's wording was clear: demand without supply constraints is "far more" than the 70% growth guidance. Nvidia's orders exceed capacity, and any supply chain disruption will immediately affect revenue realization.

The $279 billion commitment is the most underappreciated number. Supply commitments doubled from $119 billion to $279 billion in a single quarter, mainly related to memory procurement, conveying confidence in future demand more directly than any verbal guidance. This is real money committed to real supply partners.

Shareholder returns are substantial. Nvidia returned $26 billion to shareholders this quarter through buybacks and dividends, with $99 billion remaining in buyback authorization, while also issuing $25 billion in senior unsecured notes. With quarterly free cash flow of $21.1 billion, even under gross margin compression, Nvidia's single-quarter free cash flow exceeds the annual level of most S&P 500 companies.

Section 5 — Jensen Huang's Most Important Statement

"AI has reached an inflection point. It is completing valuable work. Every token it produces is productive and profitable. Now, compute is revenue."

The phrase "compute is revenue" is new and deliberate. It is the first explicit statement that companies are not just investing in AI infrastructure; that infrastructure is already generating valuable, quantifiable economic returns. This is the shift the entire investment community has been waiting for—the leap from AI as an investment thesis to AI as a profit thesis.

If Jensen Huang's judgment is accurate, this is the strongest structural argument for sustained demand. If it is premature—if the AI infrastructure built between 2024 and 2026 takes longer than expected to generate economic returns—the capital expenditure cycle driving Nvidia's extraordinary growth could reach an inflection point. The answer to this question will determine whether the current AI infrastructure buildout represents a lasting structural shift or an investment cycle that will eventually moderate. 

Section 6 — How This Connects to Everything We've Reported

Nvidia's FY2027 Q2 results are deeply connected to nearly every major theme in this series of reports.

The $166 billion in quarterly hyperscaler capital expenditure is the same force driving AI-related stock strength, fueling the memory supercycle, powering SK Hynix's record Q1 results, and simultaneously compressing Nvidia's gross margin through rising HBM prices.

The absence of the China market directly echoes the CXMT report—export controls have reduced China data center compute business to less than 1% of Nvidia's data center revenue, but also created market space for CXMT to fill domestic AI infrastructure demand.

The $279 billion memory supply commitment is directly linked to the SK Hynix listing report and the DRAM ETF report—Nvidia's memory procurement at this scale is one of the main demand drivers giving SK Hynix, Samsung, and Micron such strong pricing power.

The gross margin trajectory aligns with the logic in our Fed and yield reports. In an environment with risk-free rates of 3.5% to 3.75%, investors need high and stable profit margins to support premium valuations. When margins contract—even from 75% to 74%—the valuation math changes accordingly. 

Section 7 — What to Watch Next

Q3 earnings in late November 2026. The most critical metric is not revenue, but whether gross margin stabilizes at 74% or continues to decline along the guided path to 71% to 72%. The gross margin trajectory will determine whether Nvidia's current valuation level is justified.

Vera Rubin production ramp. Vera Rubin is expected to account for about 20% of Q3 data center revenue, and as volume continues to increase from Q4 through fiscal 2028, a faster-than-expected ramp could offset gross margin pressure, while delays would compound it.

Memory cost trends. HBM pricing from SK Hynix, Samsung, and Micron directly affects Nvidia's product costs. Any softening in HBM prices would directly benefit Nvidia's gross margin—while negatively impacting memory companies currently benefiting from high pricing.

China policy direction. Any change in U.S. export controls on AI chips to China would have immediate material impact. Current guidance assumes zero contribution from that market.

AI monetization data. Jensen Huang said "compute is revenue," and the investment community will closely track evidence that AI infrastructure is generating quantifiable economic returns.

2027 hyperscaler capital expenditure expectation of $1.3 trillion. CFO Kress disclosed that the top five hyperscalers' capital expenditures are expected to grow from about $800 billion in 2026 to about $1.3 trillion in 2027. If this materializes, Nvidia's supply-constrained growth trajectory will extend into fiscal 2028 and beyond. 

Conclusion

Nvidia's FY2027 Q2 results are extremely strong by any objective measure: revenue more than doubled, data center grew 117%, guidance beat expectations by about $4 billion, supply commitments surged from $119 billion to $279 billion, and the stock rose about 4% after hours.

However, amid this positive market reaction, what investors are most focused on is the gross margin trajectory over the next two quarters: from 75% in Q2, to 74% guided for Q3, to an expected 71% to 72% in Q4. This step-down reflects memory cost pressure that management explicitly said will "continue to rise."

Nvidia remains the single most important company in the AI infrastructure story. Its revenue scale, software moat built through CUDA, platform leadership from Blackwell to Vera Rubin, and locked-in future demand represented by $279 billion in supply commitments are all indisputable.

What the market is currently pricing in is the gross margin trajectory—and whether strong revenue growth can offset the contraction in profitability per dollar of revenue forcefully enough.

The answer will be revealed in late November.

Data as of Aug. 27, 2026. Sources: Nvidia official press release and SEC Form 8-K (Aug. 26, 2026); Nvidia official press release via GlobeNewswire; AlphaStreet Q2 FY27 earnings coverage; CNBC earnings live updates; 24/7 Wall St.; Tom's Hardware; BigGo Finance; FinanceFeeds; Benzinga; Kiplinger; RexShares; Robinhood market data; CNN Markets; NextPlatform; Shacknews.

Disclaimer: This article is for market information sharing and investor education only, and does not constitute any investment, trading, financial, legal, or other professional advice, nor a recommendation, offer, or solicitation for any securities, financial products, or investment strategies. The views, data, and forecasts herein are based on public information and the author's judgment, and may contain errors, omissions, or biases, and may be adjusted as market and company information changes. Past performance does not indicate future results, and investors should make independent judgments based on their own circumstances and bear corresponding risks. Part of this article is provided or co-written by external authors and does not represent BIT's official investment views, research conclusions, or endorsement of related views. BIT does not guarantee the accuracy, completeness, or timeliness of external authors' views. Information involving Nvidia and other securities is for general market discussion only. Investors should verify relevant information themselves and make independent judgments based on their own risk tolerance before making any investment decisions.