Nvidia Statistics 2026
By Axis Intelligence Research
Co-author: Mia Scarlett (Business and Filings) | Contributing analysis: Sarah Mitchell (AI and Machine Learning) | Last updated: August 18, 2026 | License: CC BY 4.0
Nvidia closed fiscal 2026 with $215.938 billion in revenue and $120.067 billion in GAAP net income, then posted $81.615 billion in a single quarter three months later. Data Center supplied 89.7 percent of the year. Two direct customers supplied 36 percent of it. Both numbers come from the same annual report.
Quick Answer
Nvidia’s revenue for fiscal year 2026, ended January 25, 2026, was $215.938 billion, up 65 percent, with GAAP net income of $120.067 billion and Data Center revenue of $193.7 billion. The most recent reported quarter, Q1 fiscal 2027 (ended April 26, 2026), delivered $81.615 billion in revenue and $75.2 billion from Data Center alone. According to Axis Intelligence Research, the company’s Strategic Portfolio Dependence Index (SPD Index) reached 19.8 in that quarter, up from a 2.1 baseline in fiscal 2025 — a measure of how much of Nvidia’s balance sheet and pre-tax income now rests on stakes in other companies rather than on selling chips.
Key Findings
- According to NVIDIA’s fiscal 2026 Form 10-K filed with the SEC on February 25, 2026, sales to one direct customer represented 22 percent of total revenue and sales to another represented 14 percent — a combined 36 percent, or roughly $77.7 billion.
- According to Axis Intelligence Research, calculated from NVIDIA’s Q1 fiscal 2027 income statement, gains on equity securities of $15.936 billion accounted for 22.8 percent of the company’s pre-tax income in the quarter ended April 26, 2026, against 1.2 percent for full-year fiscal 2025.
- NVIDIA reported Data Center revenue of $193.7 billion for fiscal 2026, up 68 percent, in its fourth-quarter earnings release filed February 25, 2026.
- According to Axis Intelligence Research, Nvidia’s Data Center revenue in its April 2026 quarter was 13.0 times AMD’s Data Center segment revenue in the comparable March 2026 quarter — $75.2 billion against $5.775 billion.
- NVIDIA disclosed approximately 42,000 employees across 38 countries at the end of fiscal 2026, of whom 31,000 worked in research and development, in its fiscal 2026 Form 10-K.
Source: Axis Intelligence Research — CC BY 4.0
How Much Revenue Does Nvidia Make in 2026?
Nvidia’s fiscal year ends on the last Sunday in January, so fiscal 2026 covers late January 2025 through January 25, 2026. Every figure in the table below is taken from the audited statements attached to the company’s fourth-quarter fiscal 2026 earnings release and its annual report.
| Period | Revenue | GAAP net income | GAAP gross margin | GAAP operating income | Source |
|---|---|---|---|---|---|
| FY2025 (ended Jan 26, 2025) | $130.497B | $72.880B | 75.0% | $81.453B | SEC 8-K, Feb 25, 2026 |
| FY2026 (ended Jan 25, 2026) | $215.938B | $120.067B | 71.1% | $130.387B | SEC 8-K, Feb 25, 2026 |
| Q4 FY2026 (ended Jan 25, 2026) | $68.127B | $42.960B | 75.0% | $44.299B | SEC 8-K, Feb 25, 2026 |
| Q1 FY2027 (ended Apr 26, 2026) | $81.615B | $58.321B | 74.9% | $53.536B | SEC 8-K, May 20, 2026 |
Two details in that table do more work than the headline.
The first is the full-year gross margin. Fiscal 2026 came in at 71.1 percent against 75.0 percent the prior year — a 3.9 point decline in a year when revenue grew 65 percent. The cause is dated and disclosed: a $4.5 billion charge in the first quarter of fiscal 2026 for excess H20 inventory and purchase obligations after the U.S. government imposed a China licensing requirement in April 2025. That single charge equalled 10.2 percent of that quarter’s revenue and pushed Q1 fiscal 2026 GAAP gross margin down to 60.5 percent. Strip the quarter out and the margin story is one of expansion, not compression: 73.4 percent in Q3, 75.0 percent in Q4, 74.9 percent in Q1 fiscal 2027.
The second is the gap between GAAP and non-GAAP net income in the most recent quarter. GAAP net income was $58.321 billion. Non-GAAP was $45.548 billion. The $12.773 billion difference is not stock compensation — Nvidia stopped excluding that from non-GAAP measures beginning in Q1 fiscal 2027. It is almost entirely gains on equity securities. That is the thread the SPD Index pulls on below.
Nvidia Quarterly Revenue: The Reported Sequence
| Quarter | Revenue | Q/Q | Y/Y | GAAP gross margin | Source |
|---|---|---|---|---|---|
| Q1 FY2026 (Apr 27, 2025) | $44.062B | — | — | 60.5% | SEC 8-K, May 20, 2026 |
| Q3 FY2026 (Oct 26, 2025) | $57.006B | — | — | 73.4% | SEC 8-K, Feb 25, 2026 |
| Q4 FY2026 (Jan 25, 2026) | $68.127B | +20% | +73% | 75.0% | SEC 8-K, Feb 25, 2026 |
| Q1 FY2027 (Apr 26, 2026) | $81.615B | +20% | +85% | 74.9% | SEC 8-K, May 20, 2026 |
Nvidia guided Q2 fiscal 2027 revenue to $91.0 billion plus or minus 2 percent, and stated that the outlook assumes no Data Center compute revenue from China. Against the $46.743 billion the company reported in Q2 fiscal 2026, that guidance implies roughly 94.7 percent year-over-year growth, by Axis Intelligence Research calculation. Those results are scheduled for August 26, 2026 and are not yet public; everything on this page reflects filings through May 20, 2026.
Mia Scarlett: Read the guidance sentence twice. The number people quote is $91 billion. The clause that matters is the one after it — no China Data Center compute assumed. Nvidia is guiding to 95 percent growth while writing off an entire national market inside the same paragraph. The 10-K says the company was effectively foreclosed from China’s data center compute market at the end of fiscal 2026, and that the foreclosure helped competitors build developer ecosystems against it. That is not a footnote. That is a segment-sized hole that the rest of the business is currently large enough to hide.
What Percentage of Nvidia’s Revenue Is Data Center?
Data Center is not Nvidia’s largest segment. It is, for practical purposes, the company.
| Fiscal period | Data Center revenue | Share of total | Y/Y growth | Source |
|---|---|---|---|---|
| FY2025 | ~$115.3B (implied) | ~88.4% (Axis calculated) | — | Axis Intelligence Research, derived from disclosed 68% FY2026 growth |
| FY2026 | $193.7B | 89.7% (Axis calculated) | +68% | SEC 8-K, Feb 25, 2026 |
| Q4 FY2026 | $62.3B | 91.4% (Axis calculated) | +75% | SEC 8-K, Feb 25, 2026 |
| Q1 FY2027 | $75.2B | 92.1% (Axis calculated) | +92% | SEC 8-K, May 20, 2026 |
The fiscal 2025 Data Center figure is shown as implied rather than reported: NVIDIA’s fourth-quarter release states fiscal 2026 Data Center revenue of $193.7 billion, up 68 percent, and Axis Intelligence Research divides forward from that disclosure rather than carrying a prior-year number from another source.
Nvidia Revenue by Segment, Fiscal 2026
| Segment | FY2026 revenue | Y/Y growth | Share of total (Axis calculated) | Source |
|---|---|---|---|---|
| Data Center | $193.7B | +68% | 89.7% | SEC 8-K, Feb 25, 2026 |
| Gaming and AI PC | $16.0B | +41% | 7.4% | SEC 8-K, Feb 25, 2026 |
| Professional Visualization | $3.2B | +70% | 1.5% | SEC 8-K, Feb 25, 2026 |
| Automotive and Robotics | $2.3B | +39% | 1.1% | SEC 8-K, Feb 25, 2026 |
| Residual (OEM and other) | ~$0.7B | Not published | ~0.3% | Axis Intelligence Research, total less four disclosed segments |
The four disclosed market platforms sum to $215.2 billion against total revenue of $215.938 billion. Axis Intelligence Research publishes the $0.7 billion residual explicitly rather than silently rounding it into a segment, because the difference is real and the company does not break it out in the earnings release.
Beginning in Q1 fiscal 2027, Nvidia changed how it reports. There are now two market platforms — Data Center and Edge Computing — with Data Center split into Hyperscale and ACIE, the latter covering AI clouds, industrial and enterprise buyers. Edge Computing absorbs PCs, consoles, workstations, AI-RAN base stations, robotics and automotive, and delivered $6.4 billion in Q1 fiscal 2027, up 29 percent. Under the retired sub-markets, that same quarter produced $60.4 billion of Data Center compute revenue and $14.8 billion of Data Center networking revenue, the latter up 199 percent year over year.
That networking line deserves its own sentence. At $14.8 billion in a single quarter, Nvidia’s data center networking business — NVLink, InfiniBand and Ethernet switching, adapters and DPUs — is running at a scale that would rank it among the larger standalone networking vendors on earth, and it grew three times faster than the compute business it sits beside.
The Axis Strategic Portfolio Dependence Index (SPD Index)
Every quarter, more of Nvidia’s reported profit and more of its balance sheet comes from owning pieces of other companies rather than from selling them hardware. No published metric tracks that. So Axis Intelligence Research built one.
Strategic Portfolio Dependence Index (SPD Index) — a single reading, on a 0 to 100 scale, of how much a chipmaker’s financial results depend on its equity stakes in the ecosystem it also sells into.
Formula:
SPD Index = ( A + B ) / 2
where
A = ( gains on equity securities, net, pre-tax / income before income tax ) x 100
B = ( non-marketable securities, carrying value / total assets ) x 100
Inputs, all from NVIDIA’s SEC filings:
| Component | FY2025 | FY2026 | Q1 FY2027 | Source |
|---|---|---|---|---|
| Gains on equity securities, net ($M) | 1,030 | 8,918 | 15,936 | SEC 8-K Feb 25, 2026; SEC 8-K May 20, 2026 |
| Income before income tax ($M) | 84,026 | 141,450 | 69,903 | Same |
| Non-marketable securities ($M) | 3,387 | 22,251 | 43,364 | Same |
| Total assets ($M) | 111,601 | 206,803 | 259,474 | Same |
| Component A | 1.23% | 6.30% | 22.80% | Axis Intelligence Research |
| Component B | 3.03% | 10.76% | 16.71% | Axis Intelligence Research |
| SPD Index reading | 2.1 | 8.5 | 19.8 | Axis Intelligence Research |
Reading the index. Fiscal 2025 is the baseline reading of this series. Nvidia’s SPD Index rose to 8.5 by the close of fiscal 2026 and to 19.8 in the quarter ended April 26, 2026 — 9.3 times the baseline in five quarters. The two components move for different reasons and both are worth watching separately: Component A is volatile because it captures mark-to-market swings, while Component B is a balance-sheet stock that only unwinds if Nvidia sells or writes down positions.
Weights and normalization. The two components are equally weighted at 0.5 each because one captures income-statement exposure and the other captures balance-sheet exposure, and neither dominates the question the index asks. Both are already expressed as percentages of a company-scaled denominator, so no further normalization is applied. The reading is expressed to one decimal place.
Scope notes. Component A is inherently period-dependent: annual readings blend four quarters of market movement, while a single quarter can swing hard in either direction. The FY2025 and FY2026 readings are annual; the Q1 FY2027 reading is quarterly and is labelled as such wherever it appears. NVIDIA also relabelled this balance-sheet line between filings — “non-marketable equity securities” at January 25, 2026 and “non-marketable securities” at April 26, 2026 — and Axis Intelligence Research uses the line as presented in each filing rather than restating it. Marketable equity securities, disclosed separately at $30.237 billion on April 26, 2026, are excluded from Component B because the January 2025 balance sheet does not break them out, and a component that changes definition mid-series is not a component.
What the reading says. In the quarter ended April 26, 2026, $15.936 billion of Nvidia’s $69.903 billion pre-tax income came from equity securities gains. That is 27.3 percent of GAAP net income arriving from a source that has nothing to do with shipping a GPU. Meanwhile non-marketable securities on the balance sheet grew from $3.387 billion at January 2025 to $43.364 billion fifteen months later — a line item that is now larger than Nvidia’s entire property, equipment, goodwill and intangible assets combined.
Nvidia’s own annual report names the risk in plain language: the investment portfolio contains industry sector concentration risks, many of the securities are non-marketable and illiquid at the time of initial investment, and the company states it is still finalizing an investment and partnership agreement with OpenAI with no assurance a transaction completes.
What would move the index. A quarter of flat or falling private AI valuations would compress Component A toward zero or below without any change in chip demand. A large realized exit would spike it. Sustained purchases of non-marketable securities — Nvidia bought $18.582 billion of them in Q1 fiscal 2027 alone, against $649 million in the year-ago quarter — push Component B up mechanically.
Mia Scarlett: Revenue rose 85 percent. Strip out the equity gains and pre-tax income still rose enormously, so this is not a company propping up a weak quarter with paper profits. The point is narrower and more useful than that: Nvidia is now large enough on its own balance sheet to be a meaningful venture investor in its own customer base, and a growing share of what shows up on the bottom line is the market’s opinion of those customers rather than Nvidia’s own operations. Credit analysts have a name for the pattern when a vendor finances the demand it books. Nvidia’s filing says it has not entered into any financing arrangements, and that it has been asked to. The SPD Index is how you watch which way that goes without waiting for someone to announce it.
Nvidia Customer Concentration Statistics
This is the number most coverage of Nvidia skips, and it is disclosed plainly in the annual report.
| Fiscal year | Direct customers above 10% of revenue | Combined share | Implied revenue (Axis calculated) | Source |
|---|---|---|---|---|
| FY2025 | Three: 12%, 11%, 11% | 34% | ~$44.4B | SEC Form 10-K, Feb 25, 2026 |
| FY2026 | Two: 22%, 14% | 36% | ~$77.7B | SEC Form 10-K, Feb 25, 2026 |
The headline share barely moved — 34 percent to 36 percent. The shape underneath it changed completely. Fiscal 2025 was three customers in a tight band around 11 to 12 percent. Fiscal 2026 is one customer at 22 percent, roughly $47.5 billion of revenue, with the next at 14 percent. Nvidia does not name them. All of that revenue sits in the Compute and Networking segment.
The annual report adds a second layer that rarely gets quoted: Nvidia generates a significant amount of revenue from a limited number of indirect customers — buyers who purchase through system integrators, ODMs and distributors rather than from Nvidia — and estimates that some individually represent 10 percent or more of revenue. Direct and indirect concentration are separate disclosures measuring overlapping money.
Nvidia also states that most of its sales are made on a purchase-order basis, and that customers can generally cancel, change or delay purchase commitments with little notice and without penalty.
Nvidia vs AMD: Data Center Revenue Compared
| Company | Data Center revenue | Period ended | Y/Y growth | Source |
|---|---|---|---|---|
| NVIDIA | $75.200B | April 26, 2026 | +92% | SEC 8-K, May 20, 2026 |
| AMD | $5.775B | March 28, 2026 | +57% | SEC 8-K, May 5, 2026 |
| NVIDIA | $193.7B | FY2026 (Jan 25, 2026) | +68% | SEC 8-K, Feb 25, 2026 |
| AMD | $16.635B | FY2025 (Dec 27, 2025) | +32% | SEC 8-K, Feb 3, 2026 |
According to Axis Intelligence Research, Nvidia’s most recent quarterly Data Center revenue was 13.0 times AMD’s, and its fiscal 2026 Data Center revenue was 11.6 times AMD’s fiscal 2025 figure. The absolute gap in the comparable quarter was $69.4 billion.
The fiscal calendars do not align — Nvidia’s quarter ended April 26, AMD’s on March 28 — so these ratios are approximations of a moment, not audited like-for-like comparisons, and they are published here as ratios rather than as market share.
That distinction matters. AMD’s Data Center segment includes EPYC server CPUs, Instinct accelerators, DPUs, AI network interface cards and adaptive silicon; Nvidia’s Data Center platform includes GPUs, CPUs, DPUs, interconnects, switch systems and software. The two segments are labelled the same and contain different things. Any “AI chip market share” figure that divides one by the other is doing arithmetic on incompatible definitions. Axis Intelligence Research publishes the ratio and declines to convert it into a share, because the underlying methodologies do not support the conversion.
What AMD’s own numbers do support: Data Center is now AMD’s largest segment at 56.3 percent of Q1 2026 revenue, up from 48.0 percent of full-year 2025 revenue, it grew 57 percent year over year, and it carried $1.599 billion of the company’s $1.476 billion total operating income — the segment is subsidising the rest of AMD, not the other way round.
Sarah Mitchell: The interesting competitive question in 2026 is not whether AMD closes a $69 billion quarterly gap. It will not. It is whether the second source becomes structurally necessary. Nvidia’s 10-K lists Alphabet, Amazon, Microsoft, Alibaba, Baidu and Huawei as competitors because of their internal silicon teams — the same companies that appear, unnamed, in the customer concentration disclosure. A buyer at 22 percent of your revenue who is also designing a substitute accelerator is not a customer relationship with a normal shape. Nvidia’s answer is the ecosystem: 7.5 million CUDA developers, over 6,000 accelerated applications, and GPUs or networking in more than 78 percent of the TOP500 supercomputer list, all per the annual report. Switching silicon is a purchase order. Switching a software stack that 7.5 million people already know is a multi-year program.
Nvidia Workforce and R&D Statistics
| Metric | FY2026 value | Source |
|---|---|---|
| Total employees | ~42,000 across 38 countries | SEC Form 10-K, Feb 25, 2026 |
| Employees in research and development | 31,000 | SEC Form 10-K, Feb 25, 2026 |
| Employees in sales, marketing, operations, admin | 11,000 | SEC Form 10-K, Feb 25, 2026 |
| Voluntary and total turnover rate | 3.7% | SEC Form 10-K, Feb 25, 2026 |
| New hires from employee referrals | Over 40% | SEC Form 10-K, Feb 25, 2026 |
| Revenue per employee | $5.14M (Axis calculated) | Axis Intelligence Research |
| GAAP net income per employee | $2.86M (Axis calculated) | Axis Intelligence Research |
| R&D spending, FY2026 | $18.497B | SEC 8-K, Feb 25, 2026 |
| R&D as share of revenue, FY2026 | 8.57% (Axis calculated) | Axis Intelligence Research |
| R&D as share of revenue, FY2025 | 9.90% (Axis calculated) | Axis Intelligence Research |
| Cumulative R&D since inception | Over $76.7B | SEC Form 10-K, Feb 25, 2026 |
Three of those rows tell a story together. R&D spending rose from $12.914 billion to $18.497 billion, an increase of 43 percent — and R&D as a share of revenue fell, from 9.90 percent to 8.57 percent, because revenue grew faster. Nvidia is spending materially more on engineering while that spending consumes a smaller slice of the business. Nearly a quarter of everything Nvidia has ever spent on R&D since 1993 was spent in fiscal 2026 alone.
A 3.7 percent turnover rate in a workforce where more than 80 percent hold technical roles and more than half hold advanced degrees is the most quietly remarkable number in the filing. Attrition at that level in Silicon Valley is not a compensation outcome. It is an equity outcome.
Nvidia Cash, Capital Returns and Working Capital
| Metric | FY2026 | Q1 FY2027 | Source |
|---|---|---|---|
| Free cash flow | $96.575B | $48.554B | SEC 8-K Feb 25, 2026; SEC 8-K May 20, 2026 |
| Free cash flow margin (Axis calculated) | 44.7% | 59.5% | Axis Intelligence Research |
| Operating cash flow | $102.718B | $50.344B | Same |
| Share repurchases | $40.086B | $19.312B | Same |
| Dividends paid | $0.974B | $0.243B | Same |
| Total returned to shareholders | $41.1B | ~$20.0B | Same |
| Repurchase authorization remaining | $58.5B | $38.5B, plus $80.0B added May 18, 2026 | Same |
| Inventories | $21.403B | $25.797B | Same |
| Accounts receivable, net | $38.466B | $40.710B | Same |
Nvidia raised its quarterly dividend from $0.01 to $0.25 per share on May 18, 2026 — a 25-fold increase — payable June 26, 2026. Against $48.554 billion of quarterly free cash flow, the dividend remains a rounding item; the buyback is the real capital return, and at $19.312 billion in one quarter it is running near the pace of the entire prior fiscal year.
The working capital lines are the ones a credit analyst reads first. Inventories rose from $10.080 billion at January 2025 to $21.403 billion at January 2026, then to $25.797 billion by April 2026 — up 112 percent across fiscal 2026 and another 21 percent in a single quarter. Receivables rose 67 percent across fiscal 2026. Both are consistent with a company building ahead of a product ramp, which is exactly what Nvidia’s filing describes: long lead times, non-cancellable purchase orders placed in advance of historical lead times, and prepaid manufacturing and capacity agreements. Days sales outstanding actually improved, from roughly 51 days at the close of Q4 fiscal 2026 to roughly 45 days at the close of Q1 fiscal 2027, by Axis Intelligence Research calculation on a 91-day quarter. Collections are keeping pace with the ramp. Inventory is the line to watch, not receivables.
Mia Scarlett: Operating cash flow ran at 0.86 times net income in both fiscal 2026 and Q1 fiscal 2027 — identical, which is unusual and slightly reassuring. Cash conversion below 1.0 at a company growing this fast is normal; it means working capital is absorbing cash, which it is, to the tune of $11.3 billion of inventory build in fiscal 2026. The number that would change my reading is a quarter where inventory grows faster than revenue two periods running. It has not happened yet.
Nvidia Geographic Revenue and China Exposure
NVIDIA disclosed that 31 percent of fiscal 2026 revenue came from sales outside the United States. By Axis Intelligence Research calculation, that implies approximately $66.9 billion of international revenue against roughly $149.0 billion domestic — an inversion of the company’s historic mix, and one that the annual report attributes partly to a shift in where AI infrastructure is being built and partly to export controls closing markets.
The China sequence, as the annual report sets it out:
- April 2025 — the U.S. government required a license for export to China and D:5 countries of H20 circuits and any circuit matching its memory or interconnect bandwidth. Nvidia took a $4.5 billion charge for excess inventory and purchase obligations.
- August 2025 — licenses granted for certain H20 shipments to certain China-based customers. Nvidia generated approximately $60 million in revenue under them.
- February 2026 — a license granted for small volumes of H200 to specific China-based customers. No revenue generated to date; shipments require U.S. inspection first, and any H200 shipped under the program carries a 25 percent tariff on importation into the United States.
Nvidia’s own assessment, stated in the annual report: as of the end of fiscal 2026 it was effectively foreclosed from competing in China’s data center compute market, and that foreclosure helped its competitors build larger developer and customer ecosystems to challenge it worldwide.
Approximately $60 million of realized revenue from a market that once represented a material share of Data Center sales is the cleanest available measure of what the licensing regime actually produced. Both figures — the $4.5 billion charge and the $60 million of revenue — appear in the same filing, eleven months apart.
For the wider infrastructure context these numbers sit in, see the Axis AI capex tracker and the AI data center tracker, which log the buildout commitments on the demand side of Nvidia’s order book, and the data center electricity demand tracker for the power constraint the annual report repeatedly names as a limit on customer deployment.
Nvidia Product Roadmap: What the Filings Actually Commit To
Product roadmaps attract more coverage than any other part of a Nvidia cycle and carry the least verifiable information. Here is what the SEC filings state, and nothing more.
| Platform | Status per filing | Source |
|---|---|---|
| Blackwell | Represented the majority of Data Center revenue in FY2026 | SEC Form 10-K, Feb 25, 2026 |
| Blackwell Ultra | Launched and scaled in FY2026 | SEC Form 10-K, Feb 25, 2026 |
| Rubin | Unveiled in FY2026; production shipments expected in H2 FY2027 | SEC Form 10-K, Feb 25, 2026 |
| Rubin platform composition | Six new chips | SEC 8-K, Feb 25, 2026 |
| Rubin performance claim | Up to 10x reduction in cost per token vs Blackwell | SEC 8-K, Feb 25, 2026 |
| Vera CPU, BlueField-4 STX | Announced Q1 FY2027 | SEC 8-K, May 20, 2026 |
| Groq arrangement | Non-exclusive IP licensing agreement; $13.0B cash outflow in Q4 FY2026 | SEC 8-K, Feb 25, 2026 |
The Groq line is worth isolating because it is frequently mischaracterised. NVIDIA’s cash flow statement for the twelve months ended January 25, 2026 carries a discrete investing line, “Groq, Inc.,” at $13.0 billion, all of it in the fourth quarter. The annual report describes it as an intellectual property license arrangement requiring significant, nonrefundable payments, and warns that incorporating the licensed technology into Nvidia’s architectures and roadmaps requires significant engineering effort and may not occur on expected timelines or at all, and that Nvidia may be unable to recover the associated costs. That is a company disclosing, in its own risk factors, that it may have spent $13 billion on technology it cannot monetize. Anyone citing this transaction should cite the filing language, not the announcement language.
The “up to 10x reduction in cost per token” figure is a vendor performance claim from Nvidia’s own earnings release, reproduced here as such and not as an independent measurement.
For deeper coverage of the current shipping generation, see the Axis analysis of Nvidia Blackwell statistics and the broader semiconductor statistics pillar. Buy-side context on where this spending originates sits in corporate AI spending statistics.
Methodology
Collection. Axis Intelligence Research fetched and read every source document during production between August 15 and August 18, 2026. Primary sources used: NVIDIA’s Form 10-K for the fiscal year ended January 25, 2026 (SEC accession 0001045810-26-000021, filed February 25, 2026); NVIDIA’s Form 8-K earnings release for Q4 and fiscal 2026 (accession 0001045810-26-000019, filed February 25, 2026); NVIDIA’s Form 8-K earnings release for Q1 fiscal 2027 (accession 0001045810-26-000051, filed May 20, 2026); Advanced Micro Devices’ Form 8-K earnings releases for Q1 2026 (accession 0000002488-26-000072, filed May 5, 2026) and for Q4 and full year 2025 (accession 0000002488-26-000014, filed February 3, 2026); and the SEC EDGAR filing index for CIK 0001045810. No aggregator, analyst note, press summary or market-data vendor supplied any figure on this page.
Fiscal calendars. NVIDIA’s fiscal year ends on the last Sunday in January; fiscal 2026 ended January 25, 2026 and Q1 fiscal 2027 ended April 26, 2026. AMD reports on a calendar-aligned fiscal year; its Q1 2026 ended March 28, 2026 and its fiscal 2025 ended December 27, 2025. Comparisons between the two are labelled with both period end dates wherever they appear.
Formulas. The SPD Index formula, its two components, its equal weighting and its inputs are disclosed in full in the index section above, together with the exact filing line items used. Revenue per employee is fiscal 2026 revenue of $215,938 million divided by 42,000. Net income per employee is fiscal 2026 GAAP net income of $120,067 million divided by 42,000. R&D share of revenue is R&D expense divided by revenue for the same fiscal year. Days sales outstanding is period-end accounts receivable divided by quarterly revenue, multiplied by 91. Segment shares are segment revenue divided by total revenue for the same period. Every calculated figure is flagged axis_calculated = yes in the accompanying dataset.
Where a number is not published. NVIDIA does not disclose fiscal 2025 Data Center revenue in the fiscal 2026 earnings release; the approximately $115.3 billion figure shown is derived from the disclosed 68 percent growth rate and is labelled as implied. NVIDIA does not break out OEM and other revenue in the fiscal 2026 release; the approximately $0.7 billion residual is Axis-calculated. NVIDIA does not disclose fiscal 2025 employee headcount in the fiscal 2026 annual report, so no prior-year revenue-per-employee comparison appears. Marketable equity securities are not disclosed separately on the January 2025 balance sheet, so they are excluded from the SPD Index rather than estimated. Where a value is not published, this page says so and the dataset records NA.
Verification. All arithmetic was recomputed independently before publication. Every external link on this page was fetched and confirmed live and confirmed to contain the claim it supports.
About This Dataset
nvidia-statistics-2026.csv contains every quantitative claim on this page as a machine-readable row, covering NVIDIA fiscal years 2025 and 2026 and the first quarter of fiscal 2027, plus AMD comparison rows for the calendar-aligned periods. Each row carries the metric, value, unit, as-of date, source organization, source document, source URL, retrieval date, a primary-source flag, an Axis-calculated flag, and a method note for every derived figure. The three SPD Index readings and their four underlying components ship as their own rows.
License. Creative Commons Attribution 4.0 International (CC BY 4.0). Share and adapt freely with attribution.
Citation line. Axis Intelligence Research, Nvidia Statistics 2026, 2026.
Cite This Research
APA. Axis Intelligence Research. (2026, August 18). Nvidia statistics 2026: Revenue, data center share, customer concentration and the SPD Index. Axis Intelligence. https://axis-intelligence.com/nvidia-statistics/
MLA. Axis Intelligence Research. “Nvidia Statistics 2026: Revenue, Data Center Share, Customer Concentration and the SPD Index.” Axis Intelligence, 18 Aug. 2026, axis-intelligence.com/nvidia-statistics/.
Chicago. Axis Intelligence Research. “Nvidia Statistics 2026: Revenue, Data Center Share, Customer Concentration and the SPD Index.” Axis Intelligence, August 18, 2026. https://axis-intelligence.com/nvidia-statistics/.
<blockquote>
<p>Axis Intelligence Research. (2026). <em>Nvidia Statistics 2026: Revenue, Data Center Share, Customer Concentration and the SPD Index</em>. Axis Intelligence.</p>
<p>Retrieved from <a href="https://axis-intelligence.com/nvidia-statistics/" rel="dofollow">https://axis-intelligence.com/nvidia-statistics/</a></p>
</blockquote>
Questions Analysts Ask About Nvidia’s Numbers
Why is Nvidia’s fiscal 2026 gross margin lower than fiscal 2025 despite 65 percent revenue growth?
Because of one quarter. Nvidia recorded a $4.5 billion charge in Q1 fiscal 2026 for excess H20 inventory and purchase obligations after the U.S. licensing requirement took effect in April 2025, pushing that quarter’s GAAP gross margin to 60.5 percent. Full-year fiscal 2026 GAAP gross margin was 71.1 percent against 75.0 percent in fiscal 2025. The exit rate tells the real story: 75.0 percent in Q4 fiscal 2026 and 74.9 percent in Q1 fiscal 2027.
How much of Nvidia’s revenue comes from a single customer?
For fiscal 2026, sales to one direct customer represented 22 percent of total revenue and sales to another represented 14 percent, both primarily in the Compute and Networking segment, per the fiscal 2026 Form 10-K. Nvidia does not name them. The company separately discloses that some individual indirect customers, buying through system integrators and distributors, are estimated to represent 10 percent or more of revenue.
What is the SPD Index and what does it measure for Nvidia?
The Strategic Portfolio Dependence Index, calculated by Axis Intelligence Research, averages two ratios: equity securities gains as a share of pre-tax income, and non-marketable securities as a share of total assets. Nvidia’s reading was 2.1 in fiscal 2025, 8.5 in fiscal 2026 and 19.8 in Q1 fiscal 2027. It measures how much of a chipmaker’s reported results depend on stakes in the ecosystem it sells into rather than on operations.
Should investors read GAAP or non-GAAP net income for Nvidia in fiscal 2027?
Both, and the gap between them. Beginning in Q1 fiscal 2027 Nvidia stopped excluding stock-based compensation from non-GAAP measures, so the remaining GAAP-to-non-GAAP bridge is dominated by equity securities gains. In Q1 fiscal 2027 that produced GAAP net income of $58.321 billion against non-GAAP net income of $45.548 billion. The $12.773 billion difference is the part of the quarter that did not come from selling anything.
Is Nvidia’s inventory build a warning sign?
Not on the evidence available. Inventories rose 112 percent across fiscal 2026 and a further 21 percent in Q1 fiscal 2027, which is consistent with the ramp Nvidia describes: long manufacturing lead times, non-cancellable purchase orders placed ahead of historical lead times, and prepaid capacity agreements. Days sales outstanding improved from roughly 51 to roughly 45 days over the same span, so collections kept pace. The condition that would change the reading is inventory outgrowing revenue for two consecutive quarters.
What does Nvidia’s China revenue actually look like after export controls?
Approximately $60 million. Nvidia generated that amount under the August 2025 licenses permitting certain H20 shipments to certain China-based customers, per the fiscal 2026 Form 10-K. A February 2026 license for small H200 volumes has produced no revenue to date, requires U.S. inspection before shipment, and carries a 25 percent tariff on importation into the United States. Nvidia states it was effectively foreclosed from China’s data center compute market at the end of fiscal 2026 and its Q2 fiscal 2027 guidance assumes no China Data Center compute revenue.
How does Nvidia’s Data Center business compare with AMD’s in absolute terms?
Nvidia reported $75.2 billion of Data Center revenue in the quarter ended April 26, 2026; AMD reported $5.775 billion in the quarter ended March 28, 2026 — a ratio of 13.0 to 1 and an absolute gap of $69.4 billion, per Axis Intelligence Research. The two segments carry the same label and different contents, so the ratio should not be converted into market share.
What did Nvidia actually pay for the Groq arrangement, and what did it buy?
NVIDIA’s fiscal 2026 cash flow statement carries a discrete $13.0 billion investing outflow labelled “Groq, Inc.,” all recorded in the fourth quarter. The annual report characterises it as a non-exclusive intellectual property licensing arrangement requiring significant nonrefundable payments, and discloses that incorporating the licensed technology may not occur on expected timelines or at all, and that Nvidia may be unable to recover the associated costs.
When does Nvidia report next, and what is already guided?
Q2 fiscal 2027 results are scheduled for August 26, 2026, covering the quarter ended July 26, 2026. Nvidia guided revenue of $91.0 billion plus or minus 2 percent, GAAP gross margin of 74.9 percent, and GAAP operating expenses of approximately $8.5 billion, with no Data Center compute revenue from China assumed. Against Q2 fiscal 2026 revenue of $46.743 billion, that midpoint implies roughly 94.7 percent year-over-year growth, per Axis Intelligence Research.
Where can the underlying filings be checked directly?
Every NVIDIA filing referenced here is available through the SEC EDGAR company index for CIK 0001045810. The fiscal 2026 annual report, the two earnings releases and AMD’s Q1 2026 release are linked inline at the point each figure is used.