Skip to content 
Nvidia reported its fiscal first-quarter 2027 results on 20 May 2026. Revenue for the quarter was $81.6 billion, an 85% increase from the same period a year earlier and a 20% sequential increase from the previous quarter.
Non-GAAP diluted earnings per share were $1.87, compared with the Wall Street consensus estimate of approximately $1.76. Data center revenue reached $75.2 billion, growing 92% year-on-year. Operating income on a GAAP basis was $53.5 billion, 147% higher than the comparable quarter.
Free cash flow for the quarter was $48.6 billion, against capital expenditure of approximately $1.8 billion. The company guided second-quarter revenue at $91 billion plus or minus 2%, above the consensus estimate of approximately $86 billion, and explicitly excluded any data center compute revenue from China from that guidance.
By every standard measure of business performance, this was a record quarter. Despite the results, Nvidia’s shares declined in after-hours trading following the report. This is the fourth consecutive quarter in which Nvidia has beaten consensus estimates and provided above-trend forward guidance without producing a positive stock-price reaction.
The remainder of this analysis examines what the underlying disclosures show, how the company’s reporting structure has changed, and which metrics warrant attention beyond the headline numbers.
Revenue is accelerating while stock reactions are weakening
The pattern across Nvidia’s last four reporting periods is summarised below.
| Quarter | Reported | Revenue | YoY growth | Next-day stock reaction |
| Q2 FY26 | Aug 2025 | $46.7B | +56% | –0.8% |
| Q3 FY26 | Nov 2025 | $57.0B | +62% | –3% |
| Q4 FY26 | Feb 2026 | $68.1B | +73% | –5% |
| Q1 FY27 | May 2026 | $81.6B | +85% | Decline after-hours |
Two trends are visible in the same dataset. Revenue growth is accelerating in both absolute and percentage terms. At the same time, the stock-price reaction to each successive report has become more negative. This combination is unusual and reflects how the market currently prices the company.
The most likely explanation is that the published consensus estimate is no longer the relevant benchmark for the stock reaction. The published consensus is the average of formal forecasts submitted by sell-side analysts. Institutional investors increasingly trade on a higher unpublished expectation, often called the buy-side whisper.
Going into Q1 FY27, the published consensus for the Q2 guide was around $86 billion. The buy-side whisper was in the range of $89 to $90 billion. Nvidia guided $91 billion. This was meaningfully above published consensus but only marginally above the buy-side expectation. For a company with a market capitalisation of approximately $4 trillion, only material outperformance against the buy-side number tends to produce a positive stock-price move.
There is a broader dynamic underneath this. For most of 2023 and 2024, Nvidia’s share price rerated upward as analysts repeatedly revised their estimates of the total addressable market for accelerated computing. Each earnings report added new information about the scale of AI demand, producing upward revisions to long-term earnings models.
In 2025 and 2026, that revision cycle has slowed. The remaining question is not how large the market can be but whether the current revenue run-rate is sustainable. This is a harder question to resolve in any single quarterly disclosure and produces less stock movement when results meet expectations rather than expanding them.
Nvidia has changed how it reports revenue
A significant change appeared in the Q1 FY27 disclosure that has received limited coverage. Nvidia has restructured the segments it reports.
Until Q4 FY26, the company reported revenue across five end markets: Data Center, Gaming, Professional Visualization, Automotive, and OEM and Other. Beginning Q1 FY27, three of these segments (Gaming, Professional Visualization, and Automotive) have been combined into a new segment called Edge Computing, which also captures revenue from devices for agentic and physical AI, AI-RAN base stations, and robotics. Data Center, which remains by far the largest segment, has been split into two sub-segments. Hyperscale captures revenue from public-cloud providers and the largest consumer internet companies. ACIE, an acronym for AI Clouds, Industrial, and Enterprise, captures all other data-center revenue, including sovereign AI deployments, AI factories, model builders, second-tier cloud operators, and on-premise enterprise installations.
This is a material change in how the company communicates its business to investors. The previous segmentation reflected Nvidia’s history as a chip designer with multiple end-markets of comparable strategic importance. The new segmentation reflects the company’s position as a supplier of AI infrastructure, with everything outside data centers grouped into a single residual category. Edge Computing produced $6.4 billion of revenue in Q1 FY27, which is less than 8% of total company revenue.
The data inside the new segmentation is also informative. The historical figures have been recast under the new structure, which makes year-on-year comparisons possible. Hyperscale revenue grew from $17.6 billion in Q1 FY26 to $37.9 billion in Q1 FY27, an increase of approximately 115%. ACIE revenue grew from $21.5 billion to $37.4 billion in the same period, an increase of approximately 74%. Edge Computing revenue grew from $4.95 billion to $6.4 billion, an increase of approximately 29%.
The implication is that hyperscaler capital expenditure on Nvidia hardware is growing faster than every other category. This is contrary to a narrative that has been developing in financial media that hyperscaler AI spending is maturing while sovereign and enterprise demand is accelerating. The underlying data shows the opposite. Hyperscaler purchases are the part of Nvidia’s business expanding fastest.
There is one nuance worth noting in the trajectory of ACIE. The segment fell sequentially from $21.5 billion in Q1 FY26 to $17.2 billion in Q2 FY26, before recovering to $37.4 billion in Q1 FY27. The Q2 FY26 decline coincided with the H20 inventory write-down and the effective closure of the Chinese data-center market to US chip exports. The subsequent recovery suggests that demand from sovereign AI deployments and enterprise customers has, in revenue terms, replaced the China shortfall over four quarters.
Product roadmap and margin trajectory
Nvidia’s current architecture, Blackwell, including the GB300 Ultra variant, accounts for the large majority of data center compute revenue. The company stated on the call that the platform is adopted and deployed by every major hyperscaler, every cloud provider, and every model builder. Lead times for Blackwell systems remain extended, and the company has not signalled price concessions to its largest customers. The next architecture, Vera Rubin, is on track to begin shipping in fiscal Q3 2027 and to ramp in fiscal Q4 2027, with initial deployments at Amazon Web Services, Microsoft Azure, Google Cloud, Oracle Cloud Infrastructure, and CoreWeave.
Gross margins have recovered substantially. Non-GAAP gross margin for Q1 FY27 was 75.0%, compared with 60.8% in Q1 FY26. The 14.2 percentage point improvement is partly the absence of a $4.5 billion inventory write-down that depressed Q1 FY26, but it also reflects favourable pricing on Blackwell as the platform scales. Management has guided Q2 FY27 non-GAAP gross margin at 75.0% plus or minus 50 basis points. Maintaining gross margins in the mid-70s through the transition to Vera Rubin will be the more meaningful test, since architectural transitions historically involve some margin pressure as the new platform ramps before the older one fully exits.
Free cash flow has scaled at a rate that should be assessed in its own right. The $48.6 billion in free cash flow generated in Q1 FY27 was nearly double the $26.1 billion in the comparable quarter, against capital expenditure of only $1.8 billion. This produces a free cash flow margin of approximately 60% of revenue. There are no other businesses operating at this scale that convert revenue to cash at this ratio. The drivers are the company’s fabless model, which keeps capital expenditure low, its pricing power on Blackwell, which keeps gross margins elevated, and the operating leverage produced by relatively stable headcount-driven costs against fast-growing revenue.
China is excluded from forward guidance
The Q2 FY27 revenue guidance of $91 billion explicitly assumes no data center compute revenue from China. This is the second consecutive quarter in which the company has guided on this basis. The April 2025 export rule effectively eliminated H20 chip sales to Chinese customers, and the previously disclosed $4.5 billion inventory charge related to H20 stock has not been recovered through subsequent sales.
The US government has subsequently approved sales of the H200 chip to a list of ten Chinese firms including Alibaba, Tencent, and ByteDance. As of the Q1 FY27 report, no revenue from these approved sales has been recognised. Management has therefore taken the position that any contribution from the Chinese market will be treated as upside to the guidance rather than embedded in it. CEO Jensen Huang has publicly described the Chinese AI chip market, which he estimates at approximately $50 billion, as effectively closed to American industry under current conditions.
There are two competing readings of this. The first reading is that the China exclusion creates substantial unpriced upside in Nvidia’s forward estimates. If trade conditions improve and the approved H200 sales begin to recognise revenue, that revenue would flow directly to the company without offsetting investment requirements. The second reading is that prolonged exclusion accelerates the maturation of Chinese domestic alternatives, principally Huawei’s Ascend chip line and SMIC’s foundry capabilities, which would limit the size of the addressable market available to Nvidia even if the trade restrictions were eventually relaxed. The longer the restrictions remain in place, the more likely the second scenario becomes.
A historical reference point exists in the trajectory of Russian semiconductor capabilities under US export controls, where domestic alternatives developed faster than the size of the Russian market would have suggested. The strategic logic of export controls is that the restricted party cannot reproduce the controlled technology. That assumption is currently valid for the most advanced semiconductor nodes but the margin is narrowing with each year of restrictions. The honest assessment is that the future trajectory of the China line cannot be modelled with confidence, and management has appropriately declined to do so.
Capital allocation: returns to shareholders and strategic investments
Nvidia returned approximately $20 billion to shareholders during Q1 FY27, comprising share repurchases and dividend payments. The quarterly dividend was raised from $0.01 per share to $0.25 per share, a twenty-five-fold increase. While the absolute dividend remains small relative to the company’s free cash flow, the increase indicates a shift toward more substantial capital return. A new share repurchase authorization of $80 billion was announced in addition to approximately $39 billion remaining under the previous authorization. The combined authorized buyback capacity is therefore approximately $119 billion, which is approximately 3% of the company’s current market capitalisation.
In the same quarter, Nvidia deployed $18.6 billion into investments in private companies and infrastructure funds. The company disclosed in its filing that some of these investees may indirectly purchase or use Nvidia products through cloud services. This investment activity sits alongside previously announced commitments of up to $100 billion to OpenAI, of which approximately $30 billion has been deployed, and up to $10 billion to Anthropic, along with equity stakes in several second-tier cloud operators that resell GPU capacity. The disclosed equity investment book is currently above $40 billion and growing.
The fact that the company is simultaneously returning approximately $20 billion to shareholders and deploying approximately $19 billion into strategic investments illustrates the scale of cash generation. Both activities are being funded from a single quarter’s operating cash flow, with substantial reserves remaining.
The circular financing concern
The pattern of Nvidia investing in companies that subsequently use its products has drawn analytical attention from sell-side analysts including Stacy Rasgon at Bernstein, Matt Bryson at Wedbush, and Jay Goldberg at Seaport. The shorthand for this pattern in market commentary is circular financing. The historical reference most commonly drawn is Cisco Systems during the late 1990s, which extended substantial vendor financing to telecommunications customers whose network buildouts later proved to exceed real demand. When those customers failed, the vendor financing produced losses, and Cisco’s stock has not recovered its 2000 peak in inflation-adjusted terms even now.
There are several reasons the comparison is imperfect when applied to Nvidia. Nvidia’s investments are typically equity stakes rather than vendor loans. Customer revenue is paid in cash that originates outside the loop, principally from Microsoft for OpenAI and from sovereign wealth funds and venture capital for other AI companies. The company has stated that its OpenAI commitment is not earmarked for direct chip purchases. The largest customers of Nvidia, including Microsoft, Google, Amazon, and Meta, fund their AI infrastructure spending from their own free cash flow, which is independently generated and substantial. Nvidia’s reported operating cash flow of $50 billion in a single quarter cannot be characterised as a function of its own balance sheet.
There are also reasons the comparison should not be dismissed entirely. While the hyperscalers do not require Nvidia’s financial support, several second-tier customer categories, including frontier model laboratories and specialised cloud operators, are increasingly dependent on capital flows from Nvidia or from Nvidia’s hyperscaler customers. A portion of marginal AI demand, probably small but probably greater than zero, is therefore being structurally supported by capital deployed from the supplier side. The mark-to-fair-value gains being recognised on Nvidia’s investment portfolio depend on the continued strong performance of the AI investment cycle. If that cycle weakens, the investment portfolio will produce losses simultaneously with any slowdown in operating revenue, amplifying both directions of the cycle.
The analytical resolution is that the underlying cash flow profile of Nvidia is not comparable to Cisco’s situation at its peak. The structural pattern of supplier-side investment supporting customer-side purchases does carry similarities. Whether the outcome resembles Cisco depends on whether the marginal AI demand currently being supported by Nvidia’s capital becomes self-sustaining over the next several years.
Indicators to track in upcoming quarters
Several specific metrics will provide useful signals over the next four reporting periods.
The first is the ratio of Hyperscale revenue to ACIE revenue within data center disclosure. If Hyperscale continues to grow faster than ACIE, the AI capital expenditure cycle remains driven by the largest cloud providers and the sovereign and enterprise narrative is supporting rather than leading. If ACIE accelerates relative to Hyperscale, the customer base for AI infrastructure has genuinely diversified.
The second is gross margin direction through the Vera Rubin ramp. The maintenance of non-GAAP gross margin near 75% through Q4 FY27 and into FY28 would indicate that Nvidia’s pricing power survives the architectural transition. Margin compression in the same period would indicate increased competitive pressure or normalisation of pricing.
The third is the size and concentration of Nvidia’s private investment book. Quarter-on-quarter growth in disclosed equity investments is the cleanest available signal of the degree to which the AI ecosystem is being capitalised through circular flows. The disclosure has historically been made in the company’s 10-Q filings rather than emphasised in the investor presentation, so it will require attention to footnote-level reporting.
The fourth is the gap between management’s forward revenue guidance and the buy-side whisper number. Across the last four quarters, this gap has been the most reliable predictor of the stock’s next-day reaction. A material expansion of guidance above whisper expectations would indicate that the company is once again expanding rather than confirming the bull case. A continued pattern of matching or marginally exceeding whisper expectations will likely produce the same muted stock reaction observed over the past year.
What this quarter does not resolve is whether the next twelve months will represent a continuation of the current operating trajectory or the early signal of a slowdown. The product cadence and gross margin trajectory point toward continuation. The China line provides no information by design. The circular financing structure introduces a tail risk that cannot be assessed from quarterly disclosures alone. The market’s muted reaction to the report indicates that the bar for positive stock movement has risen to a level that even strong execution within the current framework is unlikely to clear without an expansion of the underlying thesis.
