Nvidia Earnings Q2 2026: Profit and Revenue Beat Estimates, NVDA Stock Rises

Search Fintrev

Search our guides, reviews and glossary.

Nvidia’s second-quarter earnings beat Wall Street on both lines. The chipmaker reported revenue of $96.22 billion and adjusted earnings of $2.22 per share after the closing bell. Analysts tracked by Visible Alpha had expected $2.10 per share on $92.37 billion in revenue. Guidance also came in high: Nvidia forecast third-quarter revenue of $108 billion, above the consensus estimate. NVDA stock fell during the regular session, slipped briefly after the release, and then rose in after-hours trading. Investors appeared to accept the growth story even as the guidance pointed to slower expansion and as questions continued about how Nvidia’s customers are paying for the AI buildout.

This coverage covers the reported figures against expectations, the third-quarter outlook, the main points from the earnings call with CEO Jensen Huang and CFO Colette Kress, the debate over Nvidia’s customer financing, the memory supply constraint, and what the company is returning to shareholders. Three conventions help when reading the figures. First, Nvidia’s fiscal year ends in late January, so this “second quarter” belongs to fiscal 2027 and covers the middle months of calendar 2026. Second, the per-share profit figure that beat estimates is adjusted (non-GAAP) earnings, which excludes items such as stock-based compensation and certain investment gains or losses. Third, “consensus” means the average of analyst forecasts compiled by a data provider, here Visible Alpha, so a “beat” is a comparison against that average, not against a figure Nvidia promised.

Revenue and Profit Beat Estimates

Total revenue rose 106% from a year earlier to a record $96.22 billion, about $3.85 billion above the consensus figure. Working backward from the growth rate, the year-ago quarter produced roughly $47 billion in revenue ($96.22 billion divided by 2.06). Nvidia has more than doubled its quarterly sales in twelve months.

Adjusted earnings of $2.22 per share topped the $2.10 estimate by 12 cents, or a little under 6%. Going into the report, analysts had projected that both revenue and adjusted EPS would come in at roughly double the year-earlier levels. The company cleared both bars.

For context on the stock heading into the release, Nvidia shares finished the first half of 2026 up a little more than 7%, trailing the S&P 500. They had moved ahead by the eve of the report, up more than 12% year to date and narrowly outperforming the S&P 500 index. The shares were still below the record high set in May.

Data Center Revenue More Than Doubled

The data center segment produced most of the quarter’s revenue. It posted a record $89 billion, up 117% year over year and above the $85.90 billion analysts expected. That estimate assumed 109% growth, so the segment beat the consensus on both the dollar figure and the growth rate.

The rest of the business sits in what Nvidia now calls edge computing, a category that combines automotive, robotics and gaming. Edge revenue climbed 27% from a year earlier to $7.2 billion. Data center and edge together account for the $96.2 billion total. Data center made up about 92.5% of revenue and grew more than four times as fast as the edge category.

Gross Margin Met Expectations, but the Outlook Is Lower

Gross margin for the quarter was 75%, matching expectations and up from about 72.4% a year earlier. Gross margin is revenue minus the cost of producing the goods sold, expressed as a percentage of revenue. At this scale, each percentage point represents close to $1 billion of quarterly gross profit.

Costs below the gross profit line rose quickly. Operating expenses jumped 55% in the quarter. That was well below the 106% revenue growth rate, which means operating leverage still worked in Nvidia’s favor.

For the third quarter, Nvidia guided gross margin to 74%, plus or minus 50 basis points. A basis point is one hundredth of a percentage point, so the range runs from 73.5% to 74.5%. Even the top of that range sits slightly below the second-quarter figure. Before the report, Ben Reitzes, head of technology research at Melius Research, had flagged gross margin as the one quarterly detail that mattered most. In his view, a break in margins caused by memory costs would raise concerns. The guided step down is modest, but it points in the direction he described.

Third-Quarter Guidance Tops Projections

Nvidia expects third-quarter revenue of $108 billion, compared with the $105.39 billion that Visible Alpha’s tracked analysts had anticipated. That is about $11.8 billion more than the second quarter delivered, or sequential growth of roughly 12%.

The company forecast adjusted operating expenses of $9.0 billion, slightly above the $8.93 billion analysts expected. Before the release, the Street had also penciled in adjusted EPS of $2.40 for the third quarter. That is a consensus estimate and not company guidance, and it serves as the next benchmark the quarter will be judged against.

Why the Stock Wavered After a Clear Beat

The initial dip in after-hours trading came despite beats on revenue, earnings and guidance. Luke Lango, a technology analyst and publisher of Innovation Investor, attributed it to the growth trajectory. Year-over-year revenue growth had accelerated for five consecutive quarters and reached 106% in the second quarter. The third-quarter guidance implies year-over-year growth in the mid-to-high 80% range, and quarter-to-quarter growth is also expected to slow. In Lango’s words, any way you slice it, the growth story is slowing.

He also laid out the bull case. With a forward earnings multiple of about 18 times, near its lowest level in five years, Nvidia is not priced for perfection by that measure. A forward multiple divides the share price by expected earnings over the coming twelve months. A low reading means investors are paying less per dollar of projected profit than they have in recent years.

Lango pointed to three figures to argue that the AI buildout itself is not slowing: $89 billion in quarterly data center revenue, supply and capacity commitments that rose from $119 billion to $279 billion in a single quarter, and more than $500 billion in new financing partnerships. He framed the open question as Nvidia’s rate of capture within the cycle, meaning how much of the spending flows to Nvidia rather than whether the spending continues. He called it worth watching but not thesis-breaking.

Vera Rubin Ramp

Vera Rubin, the computing platform that succeeds Blackwell, began production shipments earlier in the month of the report. Kress said on the call that Nvidia has received Vera Rubin orders from every major hyperscaler, AI cloud provider and system OEM (original equipment manufacturer, meaning the server makers that build systems around Nvidia’s chips).

Kress said Nvidia expects Vera Rubin to be the fastest product ramp in the company’s history. She also said it should account for about 20% of data center revenue in the current quarter. That is a large share for a platform that only started shipping weeks earlier, and it means part of the third-quarter guidance depends on the new product’s production and delivery going to plan.

Revenue Seen Growing About 70% Next Fiscal Year

Looking beyond the current quarter, Kress said Nvidia sees revenue growing approximately 70% in the next fiscal year. She added that growth remains constrained by supply, chiefly a shortage of memory components. Huang put it more directly, telling analysts that demand is much higher than that and that he would need the help of the entire supply chain.

The 70% figure is therefore a supply-limited projection, not a demand forecast. If Nvidia can secure more components, the company’s own comments suggest there are orders to fill. If it cannot, the ceiling is set by what memory suppliers can deliver.

Memory Supply Remains the Bottleneck

Nvidia does not expect the memory constraint to ease soon. Kress said the company will work to close the gap between supply and demand but expects supply to remain a bottleneck at least through the end of fiscal 2028. She said Nvidia has long-standing relationships with all three major memory suppliers and is working with them to increase capacity.

Nvidia is backing that effort with money. Its supply and capacity commitments through 2032 grew to $279 billion from $119 billion, an increase of $160 billion in one quarter. Kress said the increase was primarily related to memory procurement, as component prices climb amid the global shortage. These commitments are agreements to buy components and capacity in future periods. They secure supply, but they also fix a large future cash outlay that Nvidia has to fund whatever demand looks like when the parts arrive.

Kress said annual commitment spending will peak at $92 billion this fiscal year and fall to $87 billion next year. Nvidia also expects equity investments to peak this year at $18 billion and capital expenditures to peak at $7 billion. Taken together, the guidance describes a company front-loading its spending to lock in supply, with outlays expected to ease after the current year.

The Customer Financing Debate

The most contested topic going into the report was Nvidia’s financial support for its own customers. Investors have watched a series of deals in which Nvidia invests in or provides financing to companies that then use the money to buy Nvidia products. Critics call this circular financing. Their concern is that if a supplier funds its customers’ purchases, some of its reported revenue depends on its own balance sheet rather than on independent demand, and a downturn could hit both sides at once.

The largest of these arrangements is a financing platform of roughly $500 billion assembled with major banks and private equity firms. According to the participants cited in coverage of the deal, these include Apollo, BlackRock, Blackstone, Goldman Sachs and KKR. Reitzes described the platform as using Nvidia chips as collateral. Full details of the structure had not been released at the time of the report. Some commentators compared it to mortgage-backed securities, the pooled-loan instruments that emerged from the U.S. housing market in the 1970s. BlackRock CEO Larry Fink reportedly made that same comparison.

Kress addressed the criticism directly on the call. She acknowledged the scale of Nvidia’s support and that some would call it circular financing, but said the company sees it differently. She described the recipients as once-in-a-generation companies with proven technology leadership and rapidly rising usage, and said Nvidia expects them to become the largest technology companies in history.

Reitzes, speaking before the report, offered a version of the bull argument. He said Nvidia chips and infrastructure are lasting longer and generating economic value for longer than people expected. In his view, that makes the hardware a financeable asset class, much as long-lived equipment can back a loan. He said the idea needs better explanation because, to a layman, it looks like circular financing. He also expects demand to rise sharply as the cost of AI tokens falls, leading to a deeper compute shortage rather than a glut. That would support the value of the collateral at this stage of the cycle.

The two views turn on one variable: how long the chips stay economically useful. If they hold value for years, chip-backed lending resembles conventional equipment finance. If newer generations such as Vera Rubin make older hardware obsolete faster than the loans run off, the collateral loses value while the debt remains.

Huang on OpenAI and Anthropic

OpenAI and Anthropic are both major Nvidia customers, and both are expected to go public. Huang said on the call that he expects them to remain customers and partners for a very long time and called them likely to be some of the most consequential technology companies in history. Asked about Nvidia’s stakes, he said his only regret was not investing more, and sooner.

The remark ties into the financing debate. Nvidia’s exposure to these companies runs through both product sales and equity investments, so their growth and eventual public valuations affect the company on more than one line of its financial statements.

Huang Says AI Demand Is Accelerating

In prepared remarks, Huang said demand is accelerating and that the AI infrastructure buildout is at full steam. He contrasted the current environment with a year earlier, when he said one lab alone was driving the buildout. He described today’s market as a golden age of new AI labs and startups, with multiple frontier labs scaling in parallel, a thriving open-model ecosystem and physical AI (AI applied to robots, vehicles and machines) coming online across the U.S. and internationally.

The argument behind the comment is about concentration. A buildout led by one customer carries the risk that the customer stops spending. A buildout spread across many labs, cloud providers and startups is less exposed to any single buyer.

$26 Billion Returned to Shareholders

Nvidia spent about $26 billion on share repurchases and dividends in the second quarter. It has about $99 billion remaining under its buyback authorization. An authorization is board approval to repurchase up to that amount. It does not commit the company to buying on any schedule.

The next quarterly dividend, 25 cents per share, is payable on Oct. 1 to shareholders of record as of Sept. 10. Since U.S. stock trades moved to one-business-day (T+1) settlement in May 2024, the ex-dividend date generally falls on the record date itself. To receive the payment, an investor needs to have bought the shares before that day.

Bank of America argued ahead of the report that Nvidia should return more. In a note that reiterated a buy rating and a $350 price target, the bank’s analysts said the payout looks modest at about 37% of projected free cash flow, based on Street estimates for 2027 and 2028. Free cash flow is operating cash flow minus capital expenditures. The analysts said raising the payout to as much as 75%, closer to Apple’s approach, could re-rate a stock they consider to be trading at a depressed multiple. They also said returning more cash would reduce the risk of overfunding the AI buildout, which could make the cycle last longer.

That suggestion runs into the spending plans described above. Cash committed to memory procurement, equity investments and capital expenditures is cash not available for buybacks. Nvidia’s guidance that those outlays will peak this year is the point where the two priorities meet.

Partner Stocks Rose Ahead of the Report

Shares of several AI hardware companies tied to Nvidia climbed before the results. Arista Networks (ANET), which makes Ethernet switches and software for large data centers, was among the best performers in the S&P 500, up close to 5% at one point. Nvidia-backed optical component maker Lumentum (LITE) rose 6%. Server maker Dell (DELL) gained, as did hard drive makers Seagate (STX) and Western Digital (WDC). Lumentum’s own earnings outlook and expected move show how closely the supplier’s results track Nvidia’s spending cycle.

These moves reflect how much of the AI hardware supply chain depends on Nvidia’s guidance. Networking, servers and storage are all sized around the GPU deployments Nvidia’s customers plan. A strong outlook from Nvidia is read as a demand signal for these suppliers, provided investors believe the demand will last.

Competition is the other side of that story. Rival chip designers are pitching alternatives to Nvidia’s accelerators, and AMD’s stock reaction to an analyst’s view of its new AI chips shows how the market prices the challenge to Nvidia’s lead.

How Much the Stock Was Expected to Move

Before the release, options pricing implied that NVDA could move up to about 6% in either direction by the end of the week. The implied move is usually estimated from the price of an at-the-money straddle, which combines a call and a put at the same strike and expiration, divided by the share price.

Hypothetical example: if a stock trades at $100 and a straddle expiring at the end of the week costs $6, options traders are pricing a move of roughly 6%, or about $94 to $106 by expiration. The figure describes the size of the move the market is paying for, not its direction, and the actual move can be larger or smaller.

The same method is applied to other large-cap reports. Microsoft’s expected post-earnings move is one example of how to read the setup.

Reading the After-Hours Reaction

Nvidia trades on the Nasdaq under the ticker NVDA. Because the company released results after the closing bell, the first market reaction took place in the after-hours session. Extended-hours trading generally has fewer participants and wider bid-ask spreads than the regular session, so prices can swing sharply on relatively small volume. The early slip followed by a rise was consistent with that: traders first reacted to the slowing growth rate in the guidance, then to the size of the beat and the commentary on demand.

A more reliable read on how the market judged the quarter comes from the next full regular session and the days that follow, which is also the window the options-implied move covers. Over that period the stock reflects the full conference call, including the Vera Rubin ramp, the supply-limited 70% growth outlook and Kress’s defense of the financing deals.

What Matters for the Next Report

The second quarter left a few open questions for the third-quarter release.

Growth rate versus growth level. Revenue is still rising in dollar terms, but the year-over-year rate is guided lower. The next report will show whether the deceleration matches the guidance or comes in better.

Margins under memory costs. The guided dip to about 74% is small. A larger slide would test Reitzes’s point that margins are the quarterly metric that matters most.

Vera Rubin’s share of data center revenue. Kress’s target of about 20% in the current quarter is a specific figure the next results can confirm or miss.

Supply commitments and cash returns. With commitment spending guided to peak at $92 billion this fiscal year, the balance between securing memory and returning cash to shareholders will show up in the next quarter’s buyback total and in any change to the payout.

Financing transparency. Details of the roughly $500 billion financing platform had not been released. How much of Nvidia’s future revenue is supported by arrangements it helped build is central to the circular financing question.

These five points take the reported beat beyond a single quarter. Huang and Kress made a case that demand will exceed supply through fiscal 2028. The next reports will show whether revenue, margins and cash flow follow that case.