Nvidia’s most recent quarter, the second quarter of fiscal 2027 (April 27 to July 26, 2026), produced $96.2 billion in revenue, up 106% from a year earlier, with $89.0 billion from Data Center. The risk that sits behind that growth is concentration: in its quarterly filing, Nvidia says one direct customer made up 16% of Q2 revenue and that it is “currently experiencing certain supply constraints.” Nvidia reported the results on August 26, 2026, and guided the third quarter to $108.0 billion, plus or minus 2%, without assuming any Data Center compute revenue from China. (Sources: Nvidia’s August 26, 2026 earnings release and its 10-Q for the quarter ended July 26, 2026.)
This post replaces our February 25, 2026 coverage of the fourth quarter of fiscal 2026 (ended January 25, 2026). That quarter is now two reports old, so the numbers below are the current ones, with the earlier quarter shown for comparison.
Which quarter is this, and what did Nvidia report?
Nvidia’s fiscal year ends in late January, so fiscal 2027 runs from January 26, 2026 to late January 2027. The second quarter is the one that ended July 26, 2026. Figures from the earnings release:
- Revenue: $96.2 billion, up 18% from the first quarter and up 106% from the second quarter of fiscal 2026.
- Data Center revenue: $89.0 billion, up 18% from the prior quarter and up 117% from a year earlier.
- Gross margin: 75.0% on both a GAAP and non-GAAP basis.
- Diluted earnings per share: $2.46 GAAP and $2.22 non-GAAP.
- Returned to shareholders: about $26.0 billion in the quarter through buybacks and dividends.
How does it compare with the quarter we covered in February?
In the fourth-quarter fiscal 2026 release, Nvidia reported $68.1 billion in revenue and $62.3 billion in Data Center revenue for the quarter ended January 25, 2026, and guided the next quarter to $78.0 billion, plus or minus 2%. Our arithmetic from the first-half totals in the 10-Q ($177,837 million for six months, less $96,221 million for the second quarter) puts the first quarter at $81.6 billion, above that $78.0 billion outlook. The new $108.0 billion outlook is about 12% above the second quarter by the same arithmetic.
What has happened since February 25, 2026
- February 25, 2026: Nvidia reported the fourth quarter and unveiled the Rubin platform, which it said delivers “up to a 10x reduction in inference token cost, compared with the NVIDIA Blackwell platform” (same release as above).
- August 26, 2026: The second-quarter results above. Chief executive Jensen Huang said in the release that Vera Rubin is “now in full production.”
- Per the 10-Q: Vera Rubin “began production shipments in the third quarter of fiscal year 2027,” which began July 27, 2026, and Blackwell still accounted for the majority of system shipments in the second quarter.
- September 28, 2026: Nvidia announced a $150 billion increase to its share repurchase authorization, bringing the remaining authorization to $235 billion, which it expects to execute through fiscal year 2028 (Nvidia press release).
What about the “tenfold” inference-cost claim?
It is a vendor claim with a ceiling attached. Nvidia’s February release says “up to” 10x lower inference token cost for Rubin compared with Blackwell; it does not say customers have measured that. We found no customer-reported figure for Rubin, and the 10-Q says shipments only began after the quarter closed. Treat the number as a design target until buyers or independent benchmarks report real cost per token. The earlier version of this post described the saving as an expected outcome; we have removed that framing and the worked savings example built on it.
What risks do the filings actually name?
These come straight from the 10-Q:
- Customer concentration: Nvidia states its revenue “is concentrated among a limited number of direct and indirect customers.” One direct customer was 16% of second-quarter revenue; across the first half, three direct customers were 16%, 15% and 13%.
- Receivables: Five direct customers made up 22%, 14%, 13%, 11% and 10% of accounts receivable on July 26, 2026, against three customers at 25%, 18% and 13% on January 25, 2026. For some investment-grade customers, Nvidia says it may extend payment terms from 90 days up to one year.
- Supply: Nvidia says it is experiencing certain supply constraints and has made large commitments to secure inventory and capacity.
- China: The outlook assumes no Data Center compute revenue from China. The 10-Q says licensed H200 shipments to China-based customers were less than 1% of Data Center revenue in the quarter, after a $0.4 billion charge in the first half for excess H200 inventory and purchase obligations.
- Regulators: Nvidia reports information requests from competition regulators in the European Union, the United States, the United Kingdom, China and South Korea.
Why this matters to a small business
Most small businesses never buy a GPU; they rent AI through software and cloud plans. Our read: the practical link is indirect. If a few buyers absorb most of Nvidia’s output and supply stays tight, the prices and limits on the AI tools you pay for are set upstream, and we would not assume they fall just because a chip maker claims lower cost per token. A sensible habit is to choose tools you could swap, and to keep your team’s know-how in your own documents rather than inside one vendor. For context on buying AI as a service, see our guides to AI for small business, AI software for small business, and Microsoft 365 Copilot for small business. This is analysis, not investment advice, and we do not comment on whether any stock is a good buy.
What to watch next
The next scheduled report covers the third quarter of fiscal 2027, the quarter that began July 27, 2026, against the $108.0 billion outlook. We have not confirmed a date for that report. The figures to compare are the share of revenue from the largest customer, the receivables concentration, and whether Nvidia’s filings describe supply constraints as easing.
