Nvidia Q2 earnings landed with numbers that would look implausible on any other company’s income statement. Revenue for the quarter ending July 26, 2026, reached $96.2 billion, up 18 percent from the previous three months and more than double the same period a year earlier.

The company also guided to $108 billion for the current quarter, which would mark another sequential jump of roughly 12 percent.

The Headline Figures

Gross margin came in at 75.0 percent on both a GAAP and non-GAAP basis, up from 72.4 percent a year ago.

Earnings per diluted share reached $2.46 on a GAAP basis and $2.22 non-GAAP. The GAAP figure more than doubled year over year.

Operating income hit $63.7 billion, a 124 percent increase from the prior year. Net income reached $59.7 billion.

For the six-month period, revenue totalled $177.8 billion against $90.8 billion in the comparable stretch last year.

Data Center Remains Everything

The data center segment produced $89.0 billion, up 18 percent sequentially and 117 percent year over year.

That single line accounts for roughly 92 percent of company revenue. Whatever else Nvidia does, this is the business.

Edge computing contributed $7.2 billion, growing 13 percent from the prior quarter and 27 percent annually. Healthy growth by ordinary standards, dwarfed by comparison.

What Huang Said

Founder and CEO Jensen Huang framed the quarter as a turning point rather than a continuation.

He argued that AI has reached its inflection point and is now performing useful work, with tokens that are productive and profitable. His formulation was that compute has become revenue.

He also described a structural change in who is buying. A year ago, he said, a single lab was driving the buildout. Now there are multiple frontier labs scaling simultaneously, a wave of new AI startups, a thriving open-model ecosystem, and physical AI coming online, with momentum across the United States and internationally.

The Vera Rubin platform, he noted, is in full production and was designed for exactly this moment.

The China Absence

One detail in the outlook deserves attention.

Nvidia is assuming zero data center compute revenue from China in its third-quarter guidance.

That means the projected $108 billion excludes what was once a substantial market entirely. The company is forecasting continued growth while writing off a major geography.

Where the Money Went

Nvidia returned approximately $26.0 billion to shareholders during the quarter through buybacks and dividends. Roughly $99.0 billion remains authorised for further repurchases.

The next quarterly dividend of $0.25 per share is payable October 1 to holders of record on September 10.

Free cash flow for the quarter was $21.3 billion, down from $48.6 billion in the prior quarter, largely reflecting a $22.3 billion increase in accounts receivable.

Signs of Strain in the Working Capital

The balance sheet tells a story the income statement does not.

Accounts receivable climbed to $63.1 billion from $38.5 billion at the start of the fiscal year. Inventories rose to $31.6 billion from $21.4 billion.

Both figures are consistent with a company shipping enormous volume and building ahead of demand, but they also mean substantial cash is tied up in the pipeline.

Nvidia also raised nearly $24.9 billion in new debt during the quarter, pushing long-term borrowings to $32.4 billion from $7.5 billion in January.

The Financing Play

Perhaps the most strategically significant announcement had nothing to do with chips.

Nvidia disclosed partnerships with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR to create independent compute financing platforms, aiming to mobilise more than $500 billion in third-party capital for AI infrastructure over time, subject to definitive agreements.

The logic is straightforward. Customers cannot buy what they cannot finance. Helping arrange the capital keeps demand from hitting a funding ceiling.

Product Announcements Worth Noting

The quarter brought a long list of launches. Several stand out.

• Vera Rubin racks are running at CoreWeave, Google Cloud, Microsoft Azure, Oracle Cloud Infrastructure, and Nebius
• Nvidia Vera, described as the first CPU built for AI agents, is headed for broad adoption
• Groq 3 LPX, an inference accelerator, entered full production
• The DSX platform gives infrastructure builders a complete framework for constructing AI factories
• A partnership with Microsoft aims to rebuild the Windows PC around RTX Spark, a one-petaflop superchip

On the physical AI side, Nvidia launched Cosmos 3, an open frontier omnimodel, and the Isaac GR00T reference humanoid robot.

Going Global

The geographic expansion is notable.

Nvidia announced sovereign AI infrastructure partnerships in Korea with SK Telecom, NAVER, and Brookfield, and a multiyear memory partnership with SK hynix.

In Japan, it partnered with the government and industrial leaders on what it describes as the world’s first national AI infrastructure.

In Europe, 35 new AI HPC supercomputers are in development.

It also secured land, power, and shell capacity in Ohio through a partnership with SB Energy.

What to Watch

The guidance implies continued acceleration, but a few things bear monitoring.

Gross margin is projected to slip slightly to 74.0 percent. Operating expenses are rising, guided to $9.2 billion GAAP. Receivables and inventory growth will need to normalise eventually.

And the fundamental question remains whether customer spending on AI infrastructure can sustain this pace, or whether the buildout eventually meets the limits of what the resulting applications actually earn.

For now, Huang’s answer is that demand is accelerating and the buildout is running at full steam.