“算力即营收”的狂飙与隐患：一文拆解英伟达最新财报

On Wednesday, August 26, NVIDIA reported the strongest quarterly results in the history of the semiconductor industry. Revenue came in at $96.2 billion, data center revenue reached $89 billion, and adjusted earnings per share hit $2.22, beating expectations by more than 6%. Q3 revenue guidance of $108 billion came in well above analyst forecasts. Subsequently, shares rose roughly 4% in after-hours trading. This report will explain what these numbers actually mean, why gross margin is the core issue investors are focusing on most, and what to watch next.

Key data: FY2027 Q2 revenue of $96.22 billion, up 106% year-over-year · Data center revenue of $89.02 billion, up 117% year-over-year · Adjusted EPS of $2.22, above the consensus estimate of $2.09 · Q3 revenue guidance of $108 billion, above the consensus estimate of approximately $104 billion · Q2 gross margin of 75%, Q3 guidance of 74%, Q4 expected to decline further to 71%-72% · Supply commitments surged from $119 billion to $279 billion · After-hours share price rose approximately 4% to $218 · Returned $26 billion to shareholders during the 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 and 18% sequentially from Q1's $81.61 billion, surpassing the analyst consensus of approximately $92.1 billion by 4.5%. A single quarter's revenue of $96.2 billion is higher than the annual GDP of many countries, and exceeds NVIDIA's entire revenue for fiscal year 2023.

Data center revenue was $89.02 billion, up 117% year-over-year and 18% sequentially. Data center now accounts for approximately 93% of NVIDIA's total revenue. All other businesses combined—gaming, automotive, professional visualization, and edge computing—amounted to just $7.2 billion, almost negligible compared to the $89 billion data center business.

Adjusted earnings per share were $2.22, beating the consensus estimate of $2.09 by approximately 6.2%. GAAP earnings per share were $2.46, which includes 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 a GAAP and non-GAAP basis, flat sequentially and up 250 basis points from 72.5% a year ago.

Q3 revenue guidance is $108 billion (plus or minus 2%), versus the analyst consensus of approximately $104 billion—an upside of roughly $4 billion. This forward-looking figure is what the market is watching most closely.

NVIDIA founder and CEO Jensen Huang declared in the press release: "AI has reached the inflection point. It is doing valuable work. Every one of its tokens 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 typically call it the "August 2026 earnings report," while NVIDIA itself labels it as fiscal 2027—both refer to the same set of numbers.

Section 2 — Where the Growth Comes From

NVIDIA's 2026 story ultimately comes down to a single driving force: all major technology companies globally are racing to build AI infrastructure, and NVIDIA produces the chips they cannot do without.

The four major hyperscalers—Alphabet, Amazon, Meta, and Microsoft—posted combined capital expenditures of $166 billion in the quarter ended June 2026, up 87% year-over-year and 27% sequentially. Over the ten quarters since the start of 2024, their combined capex has grown 272% cumulatively. CFO Colette Kress disclosed that capital expenditures from the top five hyperscalers are expected to grow from approximately $800 billion in 2026 to approximately $1.3 trillion in 2027. A significant portion of this spending flows directly 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 25% sequentially. The customer base is genuinely diversifying: AWS plans to deploy an additional 2 million NVIDIA GPUs, SpaceX has stated it will build its AI infrastructure entirely on NVIDIA, and government-led sovereign AI projects have also become a revenue source that cannot be ignored.

One of the most noteworthy numbers in the entire earnings report isn't a revenue line item at all. Supply commitments jumped from $119 billion at the end of Q1 to $279 billion at the end of Q2—more than doubling in a single quarter—largely related to memory procurement for the Vera Rubin platform. This locks in substantial future revenue that has not yet shipped, and it is the clearest evidence to date: the demand NVIDIA sees is not a forecast, but a committed order book.

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

Section 3 — Gross Margin: Why This Is the Most Watched Number

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

NVIDIA guided Q3 gross margin to 74%, a 1 percentage point decline from the 75% just achieved. At a revenue scale of $108 billion, a 1 percentage point decline in gross margin means more than $1 billion less in gross profit for a single quarter. The CFO went further, guiding Q4 gross margin to 71%-72%.

The reason behind this is memory costs. Management explicitly stated that memory cost increases have exceeded prior expectations and are expected to be "even higher" next year. Of the $279 billion in supply commitments, the majority is related to memory procurement—the price of high-bandwidth memory (HBM) embedded in every NVIDIA GPU continues to climb. Earlier this month, NVIDIA notified customers it would raise AI server prices by more than 15% to address surging memory costs, but cost increases have outpaced the price passed through to customers, putting significant 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, in part because NVIDIA and its customers have no choice but to purchase HBM at ever-rising prices. One company's cost pressure is precisely another company's revenue tailwind.

Why is gross margin so important when revenue is still growing strongly? A simple example illustrates this. If NVIDIA's revenue grows from $96 billion to $108 billion, but gross margin declines from 75% to 74%, gross profit increases from approximately $72 billion to approximately $79.9 billion—still growing in absolute terms. But investors who modeled $108 billion in revenue at a 75% gross margin were expecting approximately $81 billion in gross profit. The gap of over $1 billion per quarter, compounding over future quarters, is not negligible at NVIDIA's scale. This explains why analysts pressed heavily on gross margin during the earnings call, even though revenue guidance beat expectations by approximately $4 billion.

Educational note: Gross margin is revenue minus the direct cost of products, 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 costs. When that ratio falls to $0.74 in Q3 and possibly $0.71 to $0.72 in Q4, the difference of a few cents per dollar translates to billions of dollars at NVIDIA's revenue scale. A hardware company with a 75% gross margin is extremely rare—Apple reaches 49%, and most chip companies sit in the 50% range. The question investors care about most right now: where is the bottom of this number?

Section 4 — What the Bigger Picture Tells You

Stepping back from the quarterly numbers, several structural facts about NVIDIA's business come into clear focus.

The business is highly concentrated in AI infrastructure. Data center accounts for approximately 93% of total revenue, making NVIDIA effectively a single-product company at the earnings-driver level. If AI infrastructure spending were to slow 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 results.

Growth is constrained by supply, not demand. Jensen Huang was clear in his language: demand without supply constraints is "far more" than the 70% growth guidance. NVIDIA has more orders than capacity, and any supply chain disruption would immediately impact its ability to convert orders into revenue.

The $279 billion in commitments is the most underappreciated number. Supply commitments doubling from $119 billion to $279 billion within a single quarter, largely tied to memory procurement, conveys 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 through buybacks and dividends in the quarter, with $99 billion remaining in buyback authorization, while also issuing $25 billion in senior unsecured notes. With $21.1 billion in quarterly free cash flow, NVIDIA generates more free cash flow in a single quarter—even amid gross margin compression—than most S&P 500 companies generate in a full year.

Section 5 — Jensen Huang's Most Important Statement

"AI has reached the inflection point. It is doing valuable work. Every one of its tokens is productive and profitable. Now, compute is revenue."

The phrase "compute is revenue" is new, and it is deliberate. This is the first time it has been clearly stated that companies are not just spending money on AI infrastructure—that infrastructure is already generating valuable, quantifiable economic returns. This is precisely the shift the entire investment community has been waiting for: the transition from AI as an investment thesis to AI as a profitability thesis.

If Jensen Huang's assessment is accurate, it is the most powerful structural argument for sustained demand. If it is premature—if AI infrastructure built between 2024 and 2026 takes longer than expected to generate economic returns—then the capex cycle driving NVIDIA's extraordinary growth could face an inflection point. The answer to this question will determine whether the current wave of AI infrastructure buildout represents a durable 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 covered in this series.

The $166 billion in quarterly hyperscaler capex is the same force driving AI-related stocks higher, powering the memory supercycle, fueling 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's data center computing business to below 1% of NVIDIA's data center revenue, while also creating market space for CXMT to fill domestic AI infrastructure demand.

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

The gross margin trajectory follows the same logic as our Federal Reserve and yields report. In an environment with risk-free rates at 3.5% to 3.75%, investors need high and stable margins to justify premium valuations. When margins begin to contract—even just 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 will not be revenue, but whether gross margin stabilizes at 74% or continues along the guided trajectory toward 71% to 72%. The gross margin trajectory will determine whether NVIDIA's current valuation is justified.

Vera Rubin production ramp. Vera Rubin is expected to account for approximately 20% of Q3 data center revenue. As volumes continue to scale through Q4 and into 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 impacts NVIDIA's product costs. Any softening in HBM prices would directly benefit NVIDIA's gross margin—while negatively affecting memory companies currently benefiting from high pricing.

China policy direction. Any changes to US export controls on AI chips to China would have immediate and material impact. Current guidance assumes zero contribution from this market.

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

The hyperscaler capex expectation of $1.3 trillion in 2027. CFO Kress disclosed that top-five hyperscaler capex is expected to grow from approximately $800 billion in 2026 to approximately $1.3 trillion in 2027. If this materializes, NVIDIA's supply-constrained growth trajectory will extend through fiscal 2028 and beyond. 

Conclusion

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

Yet within this positive market reaction, what investors are focusing on most is the gross margin trajectory over the next two quarters: from 75% in Q2, to 74% guided for Q3, to an expected 71%-72% in Q4. This stepwise decline reflects memory cost pressures that management explicitly stated will "continue to rise."

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

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

The answer will be revealed in late November.

Data as of August 27, 2026. Sources: NVIDIA official press release and SEC Form 8-K (August 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 purposes only and does not constitute any investment, trading, financial, legal, or other professional advice, nor does it constitute a recommendation, offer, or solicitation of any security, financial product, or investment strategy. The views, data, and forecasts contained herein are based on public information and the author's judgment and may contain errors, omissions, or biases. Relevant data and views may be subject to change as market and company information evolves. Past performance does not indicate future results. Investors should make independent judgments based on their own circumstances and bear the corresponding risks. Some content in this article was provided or co-written by external authors and does not represent BIT's official investment views, research conclusions, or endorsement of the views expressed. 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 independently verify relevant information and exercise independent judgment based on their own risk tolerance before making any investment decisions.