Nvidia Q2 earnings: a $92B test of the AI boom

The Nvidia Q2 earnings report, due after the US market close on Wednesday 26 August 2026, is the clearest test yet of whether the AI infrastructure boom is still accelerating. The company has guided to fiscal second-quarter 2027 revenue of $91.0 billion, plus or minus 2%, a number that assumes zero data center compute revenue from China, while Street consensus compiled by Visible Alpha and Bloomberg clusters near $92 billion with adjusted earnings per share around $2.09.

Results are due around 1:20 p.m. PT, followed by the analyst call at 2 p.m. PT. Nvidia enters the print having beaten consensus estimates for 14 straight quarters, but a simple beat may not settle the questions investors are asking, because the complications sit inside the Nvidia Q2 earnings guide itself: a reported 15%-plus increase in AI server prices, the slowest sequential revenue step in a year, and a profit mix that now includes markups on equity stakes.

Options markets are pricing a move of about 5% in the stock after the print, an unusually wide band for a company that routinely clears expectations. Given Nvidia's weight in major indices, the report has the power to move the broader market, and the S&P 500's push toward record highs is partly riding on it. Benchmark raised its July-quarter revenue forecast to $92 billion with EPS at $2.10 ahead of the release, nudging the bar slightly above the Street.

The numbers behind the Nvidia Q2 earnings report

Nvidia's own guidance, issued in May, targets $91.0 billion in revenue with a non-GAAP gross margin near 75%, a level that would put year-over-year growth at roughly 95%. Consensus compilers sit slightly higher. Visible Alpha's consensus is $92.2 billion, Bloomberg's is near $92 billion and Zacks' is $91.8 billion. Projections for net income run above $51.5 billion, up about 95% year over year, and data center revenue estimates range from $83 billion to $85.7 billion, with Citi's figure at the top of that range implying growth of roughly 108%.

Data center sales are the health check for the entire AI trade, and Nvidia's new reporting framework makes that line easier to read. Since last quarter, the company splits data center revenue between hyperscalers and a second group it calls ACIE, covering AI clouds, industrial and enterprise customers. The split shows whether the spending cycle is broadening beyond the largest cloud buyers or concentrating in a few accounts, which is the difference between a durable buildout and a capex bubble in one part of the market.

Margin pressure arrives before the price increase

The pricing story is the most concrete complication in this Nvidia Q2 earnings cycle. Bloomberg's 22 August report disclosed the scale of the change: Nvidia had alerted its major customers that many AI server configurations would carry price increases above 15%. The new rates take effect on systems shipped in early 2027.

The timing gap is the crux of the margin question. The memory costs that drove the hike are already reflected in the bill of materials, so near-term margins compress while the higher prices are still months from appearing in revenue. Nvidia guided to a non-GAAP gross margin near 75% in May, before the scale of the memory pressure was public, and the first quarter's 74.9% GAAP margin shows how little room the target leaves. A guide that holds at 75% would signal that Nvidia can pass costs through; a cut would signal that memory makers have captured part of the value.

The first trading session after the price-hike news split the market's attention. Nvidia fell 2.91% on Monday 24 August, but its memory suppliers dropped far harder: Seagate lost 6.51%, SanDisk 6.45%, Micron 5.83% and Western Digital 5.24%, against an S&P 500 down just 0.28%. The pattern is a margin-transfer signal. Tight DRAM and HBM supply hands pricing power to memory makers, and the open question is whether Nvidia's ~75% gross margin absorbs the added cost or passes it through to customers.

There is a second strain on the profit picture. Roughly a quarter of last quarter's GAAP profit came from marking up equity stakes in portfolio companies rather than from selling chips. That gives the report a quality-of-earnings dimension: with net income projected above $51.5 billion, investors will want to see how much of the total comes from the core business. Nvidia's balance sheet has become a financing tool for the AI buildout, with more than $500 billion in targeted financing behind infrastructure projects, and the call will be read for how the company deploys that capital.

Sequential growth is slowing where it used to be fastest

The pace of expansion is the second watch item in the Nvidia Q2 earnings report. Nvidia's guided sequential revenue addition of $9.4 billion is smaller than the $10.3 billion, $11.1 billion and $13.5 billion added in each of the three preceding quarters, the slowest quarter-over-quarter step in a year of rapid growth. The base has grown so large that each step now carries more weight, which makes the deceleration in the adds more consequential than it looks in percentage terms.

Demand is also rotating. In the first quarter, hyperscalers contributed $37.9 billion of data center revenue against $37.4 billion from ACIE customers, but hyperscaler growth slowed to 12% quarter over quarter while the rest of the base grew 31%. The new reporting framework will show whether that rotation continues, and the pace at which the ACIE group absorbs demand determines whether hyperscaler capex remains the cycle's only engine.

China is the most visible structural break. The Nvidia Q2 earnings guidance assumes zero data center compute revenue from the country, a floor set by export restrictions rather than demand. Basing the target on zero China revenue makes it easier to defend, since any Chinese orders that do materialize would land as upside. It also means the company is ceding a market that had been a major source of data center demand in earlier years, and the speed with which buyers outside China absorb the slack will decide whether the slowdown in sequential adds is temporary.

The third signal for 2027 is the architecture transition. Systems shipping early next year sit at the boundary between Blackwell and the next-generation Vera Rubin platform, and the reported price increase straddles that shift. Comments from chief executive Jensen Huang on Rubin's ramp and timing will be the first concrete indication of 2027 supply and pricing, and of whether Nvidia's pricing power survives the transition or gets squeezed between memory costs and a new generation of silicon. If the increase is framed as covering memory costs, the ~75% margin holds and the hike is a pass-through; if it is framed as pricing power ahead of the Rubin ramp, margins could improve through next year.

Why this matters

For decision-makers, the Nvidia Q2 earnings report is a stress test of the AI buildout's economics, with consequences that reach past one company's income statement. If Nvidia holds a gross margin near 75% while passing memory costs through, pricing power across the supply chain stays intact. If the sequential slowdown, the China exclusion and the Rubin transition squeeze that margin, the cost of AI infrastructure rises for hyperscalers, cloud providers and the enterprises buying compute. The guide, the margin and the Rubin commentary together set the price of AI hardware for 2027.

Photo by Brecht Corbeel on Unsplash

Related Articles

• Samsung Q2 2026 Earnings: 89.4 Trillion Won on AI Chip Boom
• Nvidia AI server price hike tops 15% as memory costs surge
• AMD data center revenue doubles as AI chip race tightens

✔Human Verified

Researched and cross-referenced against primary sources by the Bytevyte editorial team. This article was generated with the assistance of artificial intelligence and reviewed by the Bytevyte editorial team.