A technician aligns a copper liquid-cooling plate inside an unbranded AI server chassis.

NVIDIA's revenue mix now reflects the industrial scale of AI computing infrastructure.

Companies & Earnings Technology & Innovation

NVIDIA Revenue Breakdown: Data Center Reaches 92.5%

NVIDIA is now overwhelmingly a Data Center company by revenue. In the quarter ended July 26, 2026, Data Center produced USD89.023 billion of NVIDIA's USD96.221 billion in total revenue—92.5% by ATG Press's calculation.

That percentage is useful, but it is not the full revenue breakdown. NVIDIA now separates its market platforms into Data Center and Edge Computing, then splits Data Center between Hyperscale and `AI Clouds, Industrial, & Enterprise`, or ACIE. The company still reports two accounting segments—Compute & Networking and Graphics—so readers need to keep the two disclosure views separate.

This guide uses NVIDIA and U.S. Securities and Exchange Commission disclosures available on August 28, 2026. It is designed as a reusable revenue map, not a stock-price prediction or a recap of the market's reaction to earnings.

Key Takeaways

Large unbranded data-center chillers and pipework stand under violet evening light.
Data Center revenue increasingly connects chips, networking, power and cooling infrastructure.
  • Data Center accounted for 92.5% of Q2 FY2027 revenue, up from 87.9% one year earlier on NVIDIA's recast market-platform series.
  • Inside Data Center, Hyperscale produced USD48.710 billion, or 54.7%, while ACIE produced USD40.313 billion, or 45.3%.
  • Edge Computing generated USD7.198 billion. NVIDIA says this platform includes AI PCs and workstations, gaming, professional visualization, automotive and robotics.
  • The new presentation does not mean gaming disappeared. It means gaming is no longer disclosed as its own market platform.
  • Mix concentration is only one risk test. NVIDIA's latest available Form 10-Q showed three direct customers at 21%, 17% and 16% of Q1 FY2027 revenue, while current CFO commentary disclosed USD279 billion of supply and capacity commitments.

NVIDIA Revenue Breakdown: The Short Answer

A compact industrial robot works inside a fenced electronics test cell.
Edge Computing combines several end markets, including workstations, gaming, automotive and robotics.

NVIDIA provides two different lenses on the same consolidated revenue.

The reportable-segment view is used in the financial statements. For Q2 FY2027, Compute & Networking produced USD88.299 billion and Graphics produced USD7.922 billion. Together, they equal total revenue of USD96.221 billion.

The market-platform view explains where NVIDIA's computing products and systems are deployed. Data Center produced USD89.023 billion and Edge Computing produced USD7.198 billion. Those figures also equal total revenue.

Swipe horizontally to view all columns.

Q2 FY2027 revenue viewCategoryRevenue (USD billions)Share of total*
Reportable segmentsCompute & Networking88.29991.8%
Reportable segmentsGraphics7.9228.2%
Market platformsData Center89.02392.5%
Market platformsEdge Computing7.1987.5%

\*ATG Press divides each category by USD96.221 billion and rounds to one decimal place. The two views overlap economically and should not be added together.

This distinction prevents a common mistake. Data Center is a market platform, not a GAAP reportable segment. A Data Center system can include compute, networking, software and other components that sit inside the accounting segment structure.

Five Quarters Show Data Center Concentration Rising

Unbranded high-bandwidth memory packages sit in a precision thermal test tray.
Rapid platform growth requires memory supply, inventory and manufacturing commitments as well as processor demand.

NVIDIA's recast quarterly revenue trend makes the direction clear. Data Center's share of total revenue rose in every quarter in this five-quarter window, from 87.9% in Q2 FY2026 to 92.5% in Q2 FY2027.

Swipe horizontally to view all columns.

Fiscal quarterData Center (USD billions)Edge Computing (USD billions)Total revenue (USD billions)Data Center share*
Q2 FY202641.0965.64746.74387.9%
Q3 FY202651.2155.79157.00689.8%
Q4 FY202662.3145.81368.12791.5%
Q1 FY202775.2466.36981.61592.2%
Q2 FY202789.0237.19896.22192.5%

\*ATG Press calculation using NVIDIA's recast market-platform data. Percentages may not sum to exactly 100% because of rounding.

The numerator and denominator both expanded. Data Center revenue grew 117% year over year, while Edge Computing grew 27%. Edge was not shrinking in dollars; it was becoming a smaller proportion because Data Center grew much faster.

The recast matters. NVIDIA changed its market-platform presentation in Q1 FY2027, then reclassified one company from ACIE to Hyperscale in Q2 and recast prior periods. A historical comparison should use the company's current series rather than splice old gaming and automotive labels directly into the new categories.

Hyperscale and ACIE Split the Data Center Platform

Data Center itself is not a single demand source. NVIDIA divides it into two groups:

  • Hyperscale includes hyperscale cloud providers and large internet companies.
  • ACIE includes AI clouds, industrial customers, enterprises and sovereign customers.

In Q2 FY2027, Hyperscale was larger, but ACIE grew faster.

Swipe horizontally to view all columns.

Data Center componentQ2 FY2026 (USD billions)Q1 FY2027 (USD billions)Q2 FY2027 (USD billions)Year-over-year growthShare of Q2 Data Center*
Hyperscale24.16843.05048.710102%54.7%
ACIE16.92832.19640.313138%45.3%
Total Data Center41.09675.24689.023117%100.0%

\*ATG Press divides each Q2 FY2027 component by USD89.023 billion.

NVIDIA's Q2 CFO commentary attributes Hyperscale growth to Blackwell Ultra and ACIE growth to AI-native companies, enterprises, sovereign customers and hyperscalers using AI clouds. Those are management explanations, not independent forecasts.

The split gives investors a better question than “Is AI demand strong?” Hyperscale measures the largest cloud and internet buyers, while ACIE tests whether demand is broadening into other deployment channels. A healthy mix can still be concentrated among a small number of direct distributors, system builders or cloud customers.

What Happened to NVIDIA Gaming Revenue?

Gaming products still exist. The disclosure changed.

Beginning in Q1 FY2027, NVIDIA's market-platform presentation grouped activity into Data Center and Edge Computing. The Q1 FY2027 Form 10-Q explains that Edge Computing includes AI PCs and workstations, gaming, professional visualization, automotive and robotics.

That means the USD7.198 billion Edge figure cannot be read as gaming revenue. It combines several end markets. NVIDIA separately reports Graphics as an accounting segment, but Graphics and Edge Computing are not interchangeable labels.

For readers researching the broader chip-production chain, ATG Press's Rigaku semiconductor-metrology analysis shows why inspection and process-control suppliers respond to a different set of demand and execution signals than an AI platform company does.

Three Concentration Tests Beyond Data Center Mix

The 92.5% figure describes product-market concentration. It does not describe who ultimately buys the systems, how cash is collected or how much future capacity NVIDIA has committed to secure.

1. Direct customers

In NVIDIA's latest available quarterly filing, three direct customers represented 21%, 17% and 16% of Q1 FY2027 revenue. Together they accounted for 54%. NVIDIA defines direct customers broadly to include distributors, original design and equipment manufacturers, cloud providers, AI model makers and system integrators.

Those shares should not be assigned to named companies. A direct customer may buy on behalf of other economic end users, and NVIDIA does not disclose the identities in the filing. The three Q1 percentages also should not be presented as Q2 figures.

2. Geography based on customer headquarters

The same Q1 filing reported USD63.769 billion from U.S.-headquartered customers out of USD81.615 billion total, or 78.1% by ATG Press's calculation. This is not the same as final delivery location.

NVIDIA's FY2026 Form 10-K illustrates the problem. The company estimated that 76% of Data Center revenue from Taiwan-headquartered customers in FY2026 was attributable to end customers in the United States and Europe. Headquarters data are useful, but they are not a map of where every system runs.

3. Commitments and working capital

Current CFO commentary says supply and capacity commitments increased from USD119 billion in the prior quarter to USD279 billion, mainly for memory procurement. It also reported accounts receivable of USD63.1 billion, days sales outstanding of 60 days and inventory of USD31.6 billion.

The company attributed the higher receivable period to extended payment terms on certain large, multi-quarter agreements with investment-grade customers. Inventory increased as NVIDIA prepared for Vera Rubin.

These disclosures do not prove demand will weaken. They show that fulfilling rapid growth requires larger commitments, more inventory and more financing across the ecosystem. The USD279 billion is a future commitment schedule extending across several fiscal years, not a current-period expense or a revenue forecast.

What to Watch in the Next NVIDIA Revenue Breakdown

Four checks can keep the next update disciplined:

1. Data Center share: does it stay above 90%, and is the change driven by Data Center growth or Edge weakness? 2. Hyperscale versus ACIE: does ACIE continue growing faster, or do the largest cloud buyers regain share? 3. Cash conversion: compare revenue growth with receivables, days sales outstanding, inventory and operating cash flow. 4. Commitment growth: compare supply, cloud-service, lease and guarantee exposure with the timing and quality of customer demand.

NVIDIA guided Q3 FY2027 revenue to USD108.0 billion, plus or minus 2%, and said the outlook assumes no Data Center compute revenue from China. That is company guidance, not an ATG Press forecast. The most useful future article update will replace guidance with actual results and preserve the same recast definitions.

For a Japan-based contrast in how investors map AI exposure, ATG Press's SoftBank versus SoftBank Group guide separates operating revenue from portfolio net asset value. NVIDIA's figures should likewise be read through the metric the company is actually reporting, not through a familiar brand label.

TradingView data may be delayed. The chart is market context only; NVIDIA and SEC disclosures are the sources for revenue, concentration and commitments.

Sources

Investment Disclaimer

This article is for informational purposes only and does not constitute investment advice. Company guidance and forward-looking statements are uncertain, and actual results may differ materially. Verify current filings and consult qualified advisers before making investment decisions.

-Companies & Earnings, Technology & Innovation
-, , , ,