Nvidia posts $96.2 billion quarter as AI chip demand keeps outrunning forecasts

Nvidia reported second-quarter revenue of $96.2 billion on Wednesday, a 106 percent jump from a year earlier that beat Wall Street's already elevated expectations and reinforced the chipmaker's position at the center of the artificial-intelligence spending boom. The results, disclosed in a quarterly earnings release filed after markets closed, showed data-center revenue alone reaching $89.0 billion, up 117 percent from the same period last year, as cloud providers and AI developers continued to buy Nvidia's processors faster than the company can ship them.

The report, covering the fiscal second quarter ended July 26, was disclosed in an 8-K filing with the Securities and Exchange Commission. It arrived as investors, still digesting a separate government inflation report earlier in the week, looked to Nvidia to validate the enormous sums that Microsoft, Amazon, Google and Meta are pouring into AI data centers. Nvidia's own guidance suggested that spending is not slowing: the company forecast third-quarter revenue of $108.0 billion, plus or minus 2 percent, which would mark growth of roughly 89 percent from a year earlier.

A quarter that outran forecasts

Nvidia's non-GAAP earnings came to $2.22 per share, more than double what it reported in the same quarter last year. On a GAAP basis, the company earned $2.46 per share and $59.7 billion in net income, on gross margins of 75.0 percent. The results topped the consensus revenue estimate of roughly $92.2 billion that analysts had penciled in ahead of the report, according to figures compiled by financial data providers and cited in coverage from Euronews.

Beyond the headline data-center number, Nvidia's combined gaming, professional-visualization, automotive and OEM businesses — now reported together as an "edge computing" segment — brought in $7.2 billion, up 27 percent from a year ago but a much smaller share of the total than in years past. The imbalance underscores how thoroughly the company's fortunes now hinge on hyperscale cloud operators and AI labs buying GPUs by the tens of thousands rather than on gamers buying graphics cards one at a time.

Nvidia also disclosed that it returned $26.0 billion to shareholders in the quarter through buybacks and dividends, leaving roughly $99.0 billion of repurchase authorization still on the books, and declared a quarterly dividend of $0.25 per share payable October 1. The full breakdown of the quarter, along with segment detail and the company's forward-looking statements, is laid out in the underlying SEC filing index for the release.

Why it matters beyond one company

Nvidia's results function as a proxy for the health of the broader AI buildout. The company's chips sit inside the data centers that Microsoft, Amazon, Google, Meta and a wave of smaller "neocloud" operators such as CoreWeave are constructing at a pace that has strained power grids, real-estate markets and, in some cases, corporate balance sheets. A miss from Nvidia would have raised fresh doubts about whether that spending is sustainable; a beat, instead, keeps the capital flowing toward suppliers, contractors, utilities and the smaller chip and networking firms that feed into Nvidia's supply chain.

The quarter also arrived against a shifting policy backdrop in Washington. After the Trump administration barred sales of Nvidia's China-specific H20 chip in 2025, the Commerce Department's Bureau of Industry and Security later revised its licensing policy to allow case-by-case approval of the more capable H200 chip for vetted Chinese customers, including Alibaba, Tencent and ByteDance, with the U.S. government taking a cut of the resulting sales. Nvidia's third-quarter guidance explicitly excludes any data-center compute revenue from China, meaning the $108 billion forecast could prove conservative if those licenses translate into shipments — or could remain unrealized if Beijing continues steering its companies toward domestic chipmakers instead.

• Data-center revenue: $89.0 billion, up 117 percent year over year
• Total revenue: $96.2 billion, up 106 percent year over year and 18 percent from the prior quarter
• Non-GAAP earnings per share: $2.22; GAAP earnings per share: $2.46
• Third-quarter guidance: $108.0 billion in revenue, plus or minus 2 percent, excluding China data-center sales

A muted, choppy reaction on Wall Street

The market's initial response was more subdued than the numbers alone might suggest. Nvidia shares wobbled in extended trading immediately after the release, a pattern that has followed several of the company's recent reports even when results beat estimates, before firming as the earnings call got underway and Chief Financial Officer Colette Kress walked through the quarter in more detail. Investors have grown increasingly focused on whether Nvidia's eye-popping growth rates can be sustained once the current wave of data-center construction matures, and on how quickly the company's next-generation Vera Rubin platform, which Nvidia said is ramping into full production, will replace existing Blackwell-based systems in customer order books. Live coverage of the call from Kiplinger's markets desk described shares turning sharply higher only once Kress began detailing the quarter, rather than on the headline numbers alone.

"AI has reached its inflection point. It's doing useful work. Its tokens are productive and profitable. Now, compute is revenue," said Jensen Huang, Nvidia's founder and chief executive.

That framing — that AI spending is shifting from speculative infrastructure-building to revenue-generating deployment — is central to how Huang has tried to reassure investors who worry about a bubble in AI capital expenditure. Whether that argument holds will depend heavily on public disclosures from Nvidia's largest customers in the coming weeks about their own AI returns, and on whether enterprise and government demand for AI applications keeps pace with the hardware being installed.

Who feels the effects

The results carry direct consequences well outside Nvidia's Santa Clara headquarters. Contract manufacturers such as Foxconn, memory suppliers like SK Hynix and Micron, and networking companies that build the interconnects linking thousands of GPUs together all stand to benefit from continued data-center demand at this scale. Cloud customers renting Nvidia-powered capacity, from large AI labs to smaller startups building on top of rented compute, are watching the same guidance for signs of whether GPU scarcity and pricing will ease or persist into next year. And retail and institutional investors holding Nvidia stock, now one of the most heavily weighted names in major indexes, have a direct stake in whether the company's growth trajectory can extend into fiscal 2028 as the initial AI infrastructure buildout matures.

Workers across the supply chain are affected as well. Nvidia itself has continued hiring engineers to support its expanding product roadmap, while suppliers ramping up production of Blackwell and now Vera Rubin systems have added shifts and headcount to keep pace with orders. A slowdown in Nvidia's growth, by contrast, would ripple quickly through firms whose revenue depends on continued GPU demand.

What comes next

Attention now turns to the third-quarter guidance and to how quickly the Vera Rubin platform, along with the newly announced Vera CPU for AI agents and the Groq 3 LPX inference accelerator that Nvidia said entered full production, translate into shipped revenue. Nvidia's forecast assumes zero data-center compute revenue from China, a deliberately conservative baseline that leaves room for upside if Washington's licensing regime continues to open and Chinese buyers resume large orders. Analysts will also be watching gross margins, guided to 74.0 percent for the current quarter, for signs of whether rising component costs and the transition to new chip generations are beginning to squeeze profitability even as revenue keeps climbing.

More broadly, the report sets the tone for a string of earnings calls from Nvidia's largest customers, whose own capital-spending disclosures in coming quarters will show whether the AI infrastructure buildout that has powered Nvidia's growth is beginning to plateau, or whether, as Huang argues, it is only beginning to generate the revenue that will justify the spending that built it.