NVIDIA Q1 FY2027 Earnings: $81.6B Revenue, $75.2B Data Center, $80B Buyback, and Why the Stock Fell on a Beat

What You'll Learn

• How NVIDIA's $81.6B revenue beat landed against an $78B consensus and why the headline understates the data center story
• Why data center networking hit $14.8B (up 199% YoY) and what Spectrum-X means for the AI fabric wars
• The three forces (China H20, gross-margin guide, and the capex cycle) that pulled NVDA stock down on a beat
• How the $80B buyback and 25x dividend hike change NVIDIA's shareholder-return profile
• What Vera Rubin, Blackwell Ultra, and the $1T Blackwell revenue path mean for the next 18 months

The Headline Numbers: A Record $81.6 Billion Quarter

NVIDIA's fiscal first quarter of 2027, ended April 26, 2026, delivered a print that should have been impossible to argue with. Total revenue hit $81.615 billion, up 85.3% year over year and 19.8% sequentially. That is the third consecutive quarter of acceleration from the company whose growth curve was supposed to be flattening a year ago. Eight quarters ago, NVIDIA's quarterly revenue was $30.0 billion; today it is more than 2.7x that level, with no signs of compression.

The beat landed well above the $78.0 billion guidance NVIDIA itself had set three months earlier and well above the $78.8 billion Visible Alpha consensus. GAAP earnings per share came in at $2.39, while the non-GAAP figure (which, beginning in Q1 FY27, no longer excludes stock-based compensation expense) was $1.87. Both figures cleared Street estimates of $1.76 to $1.78. Net income reached $58.3 billion for the quarter. Operating expenses totaled $7.621 billion, up 12% year over year, a signal that the company is still investing aggressively even as margins expand.

But the most consequential line of the income statement is the new capital-return announcement: NVIDIA's board approved an additional $80 billion in share repurchase authorization, on top of the cash already returned during the quarter. The company also raised its quarterly dividend by 25x, a 2,400% increase that puts the dividend yield meaningfully above its prior sub-1% range. CFO Colette Kress disclosed on the call that NVIDIA returned a record $20.0 billion to shareholders in Q1 FY2027 through repurchases and dividends combined. The combination of buyback scale and dividend acceleration is, in dollar terms, the most aggressive capital-return quarter in U.S. semiconductor history.

Data Center: The $75.2 Billion Engine

Data Center is no longer a segment. It is the company. Q1 FY2027 data center revenue hit $75.2 billion, up 92% from a year ago and 21% from the prior quarter, accounting for roughly 92% of NVIDIA's total revenue. The line item has gone from $4.3 billion in the year-ago quarter to $75.2 billion in eight quarters, a 17x increase. There is no precedent in enterprise technology for a segment to compound at this scale for this long.

Inside the data center line, the disaggregation tells an even more important story. Computing revenue, which is largely the GPU systems (Blackwell rack-scale, HGX baseboards, and DGX SuperPOD configurations), came in at approximately $60 billion, up 77% year over year. Networking revenue, including Spectrum-X Ethernet switches, NVLink switch trays, BlueField DPUs, and InfiniBand, came in at approximately $15 billion, nearly tripling year over year. On the conference call, management noted that Spectrum-X is "now larger than all Ethernet network peers combined," a remarkable claim given that Arista Networks, Cisco, and Broadcom all sell tens of billions in datacenter networking annually.

The acceleration in networking is the most under-appreciated part of the quarter. NVIDIA's N1X at Computex 2026 showed that the company is no longer content to be a GPU vendor; it wants to own the entire AI factory stack, from the rack to the NIC. That is a meaningfully different competitive posture than the one NVIDIA held just two years ago, when networking was a footnote to GPU allocations.

For context, peer data center results from this quarter help frame the magnitude. Snowflake's Q1 FY2027 results with the $6 billion AWS deal and 36% growth were considered a strong print for cloud data infrastructure. NVIDIA's data center line alone is 12.5x larger than Snowflake's entire annualized revenue run rate. That is the gap between "AI infrastructure" and "AI compute substrate."

Networking, Edge, and Gaming: The Segments Outside Data Center

Outside data center, the rest of NVIDIA's business is small in dollar terms but still meaningful in product-cycle terms. Edge Computing revenue, which now includes the segment that was previously labeled "Gaming" (reflecting NVIDIA's decision to recategorize GeForce and RTX Pro under the Edge Computing umbrella), was $6.4 billion, up 10% from the prior quarter. The slowdown in gaming GPUs has been attributed to elevated memory prices, but the launch of the RTX 5060 and RTX 5060 Ti helped stabilize the consumer side. RTX Pro workstation revenue, which is part of the Edge Computing segment, surged 29% on Blackwell workstation demand from professional visualization and edge AI deployments.

NVIDIA also announced the RTX PRO 4500 Blackwell Server Edition GPU during the quarter, a new enterprise SKU targeting on-prem and private-cloud AI deployments. That product line is small today but is strategically important: it lets NVIDIA monetize the AI buildout in regulated industries (financial services, healthcare, defense) that cannot run training workloads on public hyperscaler infrastructure.

Automotive revenue was approximately $1.7 billion, up slightly from the prior quarter. The automotive line is still small but is the cleanest AI-robotics exposure in NVIDIA's portfolio, with Mercedes-Benz, Volvo, and several Chinese OEMs locked into the DRIVE Thor roadmap. Robotics and the upcoming Jetson Thor refresh remain the longest-dated bets in the mix and the most likely sources of upside surprise in 2027 if the embodied-AI market opens up faster than consensus models assume.

Margins, Cash Flow, and the New Capital Return Playbook

Gross margin came in at 74.9% on a GAAP basis and 75.0% on a non-GAAP basis, essentially flat sequentially. That matters because Wall Street had been modeling compression through the Blackwell ramp, on the theory that the more complex packaging and higher HBM3E content would pressure unit economics. It has not. The Blackwell Ultra refresh and the Vera Rubin transition are on track to maintain the 70%+ gross margin band that defines the company. Management's commentary on the call indicated that gross margin should remain around 70% as the Rubin architecture ramps, with operating expenses rising throughout the year on continued R&D investment.

Operating cash flow was $50.3 billion for the quarter. Free cash flow came in at $48.55 billion, a free cash flow margin of approximately 59%. That is one of the highest FCF margins ever recorded for a large-cap industrial. Annualized, the FCF run rate is approaching $200 billion. The implication for the capital structure is enormous: even with $80 billion of incremental buyback authorization, NVIDIA will continue to accumulate cash and marketable securities on the balance sheet unless it executes M&A at a scale the semiconductor industry has rarely seen.

The capital return math is simple. NVIDIA generated $48.6 billion of FCF in a single quarter. It returned $20.0 billion in that same quarter. That means the company is plowing roughly $28 billion per quarter into net cash, on top of the cash it already holds. At a $3.4 trillion market cap, the buyback yield is in the 2%+ range annualized, and that is before accounting for the 25x dividend increase. The shareholder-return profile now looks more like a mature, cash-rich blue-chip than a high-growth semiconductor, which is one reason why some investors are starting to frame NVDA as a "cash compounder" rather than a "growth stock."

Q2 Guidance: $91 Billion, Plus the H20 China Question

The Q2 FY2027 revenue guide of $91 billion, plus or minus 2%, was the most important number management released. It sits roughly $4.2 billion above the $86.84 billion LSEG consensus, a 5% positive surprise in the forward guide. The implied sequential growth from $81.6B to $91B is 11.5%, an aggressive ramp for a quarter that historically shows a seasonal step-down.

Two things are notable about the Q2 guide. First, it is set in the absence of any Data Center compute revenue from China. NVIDIA's Q1 FY2027 outlook did not assume any H200 or H20 shipments to China, and management has not added China back into the Q2 guide. This means the $91B is the floor, not the ceiling: any relaxation of the export-control regime during Q2 would create upside that is not currently in Street models.

Second, the guide does not include any material contribution from Vera Rubin, the next-generation GPU platform launching in 2H 2026. Rubin is still in sampling, with full production targeted for the second half of the fiscal year. The Blackwell Ultra refresh, which sits between Blackwell and Rubin, will carry the bulk of the H2 volume. This means NVIDIA is guiding to a $91B quarter on what is essentially a single product generation (Blackwell) plus the early Ultra refresh. The Rubin cycle is upside that is not in the Q2 number.

For broader market context, this Q2 print is happening against a backdrop where the Federal Reserve's June rate decision will set the discount rate that AI capex budgets are valued against. Lower rates extend the duration of the AI capex cycle; higher rates pressure the present value of forward earnings. That macro overlay is the single biggest external variable in the NVDA story right now.

Why the Stock Fell on a Beat: Three Forces Investors Read Wrong

Despite the headline beat, NVDA traded down 1.4% to 2.8% in the 24 hours after the print, depending on the intraday reference. The market reaction looks counterintuitive on the surface, but the sell-side note flow after the call clustered around three distinct concerns.

First, the China overhang. Investors were looking for any incremental commentary on H20 shipments to China after the early-2026 export-rule modifications. NVIDIA secured initial U.S. licenses for H200 shipments to China in February 2026, but the Q1 FY2027 results do not include any China Data Center compute revenue in the guide. Investors read the silence on incremental China upside as a signal that the export-control regime has not eased meaningfully since the February H200 license round and that the H20 chip remains effectively shut out of the largest single end-market for AI compute.

Second, the gross-margin guide. While the Q1 gross margin of 74.9% beat consensus, the forward commentary about Rubin-era margins staying "around 70%" was read as a setup for compression, not stability. The market had been modeling 71% to 72% as the long-term steady state; hearing 70% from the CFO reset that bar lower. Even though 70% gross margin would still be extraordinary for a semiconductor company, the directional signal mattered more than the absolute level.

Third, the capex-cycle concern. Hyperscaler AI capex is the input that drives every other variable in the NVDA model. The Q1 results showed that the big four hyperscalers (Microsoft, Google, Meta, Amazon) are still buying at a record pace, but the broader market is now starting to ask: what is the steady-state capex run rate, and when does the digestion phase start? Jensen Huang's call commentary, that Blackwell sales are "off the charts" and "cloud GPUs are sold out," was meant to dispel that concern, but the language also inadvertently reminded the market that the buildout is still in its early innings. Early-innings capex is great for NVIDIA in 2026; the digestion question is what comes after.

It is worth noting that the post-earnings selloff was modest in historical context. NVDA has now beaten revenue by $2.6 billion to $3.6 billion in each of the last four quarters, and the stock has reacted with a 1% to 5% pullback each time before resuming the trend. The pattern of "sell the news on a beat" is well-established and is not, by itself, a structural negative.

Wall Street Reaction: 51 Buys, One $300 Target, and the $445 Bull Case

Wall Street's response was as bifurcated as the post-earnings tape. Of 54 analyst ratings tracked by MarketBeat, 51 rate NVIDIA a Buy and 3 rate it a Hold; none rate it a Sell. The average 12-month price target sits at $305.38, implying 45% upside from a $211 reference. Marketscreener's broader consensus of 61 analysts shows a mean Buy rating, an average target of $296.81, and a 40.58% spread to the consensus. JPMorgan lifted its target to $280 after the print, while KeyBanc moved to $300 on the Blackwell-plus-Rubin catalyst stack.

On the more aggressive end, TIKR's internal model places a mid-case target of approximately $445, implying 124% total return and a 19% annualized IRR. The bull case rests on three legs: (1) Blackwell revenue compounding from $86.4B in calendar 2025 to $137.0B in calendar 2026; (2) the Vera Rubin transition layering an additional $40B to $60B of revenue in 2H 2026; and (3) the Free Cash Flow generation funding a buyback yield that compresses the share count at a 4% to 5% annualized rate.

Not every analyst is bullish. Simply Wall St's revised target moved to $181.39, implying that some models are still anchored to a more conservative 30x forward earnings multiple applied to FY28 numbers. The dispersion between the $181 bear case and the $445 bull case is the widest it has been in NVIDIA's history as a large cap, and that dispersion is itself a tradeable signal: the options market is pricing a realized move of roughly 8% to 10% per earnings event, which is consistent with the realized vol since the Q4 FY2026 print.

For a broader look at how the AI capex story fits into the macro picture, the U.S. national debt at $39.5 trillion and the $1 trillion interest bill is a useful reminder that the AI capex cycle is being financed against a deteriorating U.S. fiscal backdrop. The intersection of those two stories (record AI infrastructure spending and record U.S. interest expense) is the single biggest second-order risk to the NVDA bull case.

Vera Rubin, Blackwell Ultra, and the Path to $1 Trillion in Blackwell Revenue

The roadmap conversation is now centered on three dates: Blackwell Ultra in production today, Vera Rubin launching in 2H 2026, and Rubin Ultra in 2H 2027. The fourth date, Feynman, is the 2028 architecture that was announced at GTC 2025 and confirmed at the AI Infrastructure Summit. Each platform is positioned to deliver roughly a 2x improvement in inference token cost reduction and a 1.5x to 2x improvement in training throughput per watt versus its predecessor.

Rubin is the most consequential of these transitions. The Rubin GPU delivers 50 petaflops of FP4 performance per chip, with the rack-scale Vera Rubin platform offering up to 10x reduction in inference token cost versus Hopper. The new Vera CPU, NVIDIA's first custom Arm-based server CPU, is paired with Rubin in the NVL72 and NVL576 rack configurations, enabling NVIDIA to compete more directly with custom silicon from AWS (Trainium), Google (TPU), and Microsoft (Maia). On the Q1 FY2027 conference call, management noted that the Vera CPU platform is targeting the $200 billion agentic AI market opportunity, a market that did not exist 12 months ago.

The near-term revenue implication is straightforward. If Blackwell revenue compounds from $86.4B in calendar 2025 to $137.0B in calendar 2026, and Rubin adds a partial-year contribution in 2H 2026, the full-year FY27 revenue run rate is in the $360B to $400B range, far above the FY26 actual of $215.9B. S&P Global's analyst note frames this as the path to $1 trillion in cumulative Blackwell-family revenue across FY26, FY27, and FY28. That number, if achieved, would make NVIDIA the second U.S. company after Apple to clear a single trillion-dollar product cycle.

For context on the broader AI ecosystem, the Q1 results are also relevant to Apple's WWDC 2026 AI strategy. Apple's on-device AI push is a long-term competitive answer to NVIDIA's data center dominance, and the size of NVIDIA's data center line makes it clear that on-device and cloud AI are not competing for the same dollar; they are complementary, with cloud AI consuming the bulk of the silicon capacity NVIDIA can produce.

What Comes Next: A Trading Guide to NVDA Through Q2

The setup for the rest of 2026 is unusually well-defined. There are three near-term catalysts that will move the stock, in order of likely impact.

The first catalyst is the China export-control decision. Any incremental H20 or H200 license announcement would unlock $5B to $15B of upside to the FY27 revenue base, depending on the volume. The political calendar (midterm-year, U.S.-China trade negotiations) suggests the export regime is more likely to loosen than to tighten, but the timing is uncertain.

The second catalyst is the Blackwell Ultra ramp. The first quarter of material Ultra volume is widely expected to be Q2 FY2027, with the guide of $91B already embedding the early Ultra contribution. Any commentary at upcoming industry events (GTC Paris, Computex follow-up, SIGGRAPH) about Ultra supply-demand or pricing will reset the multi-quarter setup.

The third catalyst is Q2 FY2027 earnings in late August 2026. By that print, the Q2 guide of $91B will be in rearview, and management will offer the first real read on Rubin pull-in. A guide above $100B for Q3 would be a market-moving event; a guide below $95B would be the first sign of a digestion phase.

For investors, the key question is positioning. The consensus framework is: hold through the next two earnings events, watch for any softening in the hyperscaler capex commentary, and be ready to add on a 10%+ pullback tied to a non-fundamental headline (China export news, geopolitical event, or a large-cap tech earnings miss from a hyperscaler customer). The structural setup, a $360B to $400B revenue run rate, 70%+ gross margin, ~60% FCF margin, and a 25x dividend hike with $80B in buyback authorization, is one of the strongest in U.S. large-cap history.

Conclusion

NVIDIA's Q1 FY2027 was a beat on every important line item, a raise on Q2 guidance, and a step-change in capital return. The $81.6B revenue, $75.2B data center, $48.6B FCF, and $91B Q2 guide are the kind of numbers that define a generational company. The fact that the stock fell 1% to 2% on the print tells you more about the market's elevated expectations than about the underlying business. The China overhang, the 70% gross-margin guide, and the capex-cycle concern are all real, but they are not Q1 FY2027 problems; they are Q3 FY2027 and beyond problems.

The next 90 days will be defined by the China export decision, the Blackwell Ultra ramp, and the first Q2 earnings. The structural backdrop, a multi-year AI capex cycle, a clear product roadmap through 2027, and the most aggressive capital-return profile in U.S. semiconductors, remains intact. The market may need one or two more quarters of digestion to reset expectations, but the long-term setup is unchanged. NVDA at $211 is not a sell, but it is not a screaming buy either. The right posture is to hold the core position, add on a pullback, and wait for the Q2 print to confirm the $91B guide.

Sources & References:

• NVIDIA Official Q1 FY2027 Earnings Press Release (NVIDIA Investor Relations)
• NVIDIA Q1 FY2027 8-K Filing (SEC.gov)
• Nvidia's profit triples as Jensen Huang predicts further revenue jump (Reuters)
• Nvidia (NVDA) Q1 2027 earnings report: Live updates (CNBC)

Last Updated: June 01, 2026 | Source: NVIDIA Investor Relations (Official Press Release), SEC Filings, Reuters, CNBC