Nvidia beats on $46.7B revenue — and no China H20 sales

Nvidia has turned quarterly earnings into a kind of market-moving ritual, and Wednesday’s numbers didn’t disappoint. The chipmaker reported fiscal second-quarter 2026 revenue of $46.7 billion and non-GAAP earnings per share of $1.05, topping Wall Street’s consensus forecasts of about $46 billion and $1.01. For a company that has already rewritten expectations several times over, the bar was high — and still, it managed to raise it.

AI drives the quarter

The numbers underline just how firmly Nvidia sits at the center of the AI gold rush.

The top line revenue numbers were up 56% from a year earlier and 6% from the prior quarter, powered almost entirely by its data-center business and underscoring the still-insatiable demand for Nvidia’s AI accelerators. That division — the beating heart of the AI boom — booked $41.1 billion in sales, also up 56% year-over-year (and 5% sequentially). Within that, the company’s new Blackwell chips ramped up even faster than expected, notching a 17% sequential increase.

“Blackwell is the AI platform the world has been waiting for, delivering an exceptional generational leap — production of Blackwell Ultra is ramping at full speed, and demand is extraordinary,” CEO Jensen Huang said in the press release. “The AI race is on, and Blackwell is the platform at its center.”

Gaming, Nvidia’s legacy business, contributed $4.3 billion in revenue, up nearly 50% from last year, proving that the AI wave hasn’t completely eclipsed the company’s consumer roots. Other divisions also delivered double-digit growth. Professional Visualization revenue rose 32% from a year ago to $601 million, while Automotive climbed 69% to $586 million, reflecting Nvidia’s push into robotics and autonomous driving platforms. While those areas are still small compared with the company’s data center business, those lines are increasingly positioned as future growth pillars.

A beat, but no bounce

Yet beneath the blockbuster growth, there were hints of tension. Shares fell around 2% in after-hours trading, as markets weighed a slight miss in data-center revenue despite the overall beat.

Analysts had penciled in slightly higher data-center sales, around $41.3 billion, making the reported $41.1 billion just shy of some expectations. That hairline miss helps explain the stock’s fall. Options traders had already priced in a swing of about 6% either way, a sign that markets were braced for another monster quarter but not expecting fireworks. With a market capitalization north of $4.4 trillion, Nvidia has become the single most important stock in the S&P 500 — accounting for roughly 8% of the entire index — so even a beat can look routine when expectations are this inflated.

Story Continues

Margins on steroids

Profitability also surged. Net income nearly hit $26.4 billion (GAAP) for the quarter, up 59% year-over-year. Gross margins climbed to 72.7% on a non-GAAP basis, excluding one-time adjustments linked to its H20 chip inventory. Free cash flow reached $13.5 billion, giving Nvidia ample room to fund both research and shareholder returns.

Huang told analysts, “You’ve heard me say before that in a lot of ways, the more you buy, the more you grow. And because our performance per dollar is so incredible, you also have extremely great margins.”

And shareholder returns are now a headline in their own right. The company returned $24.3 billion to investors in the first half of the year through buybacks and dividends, and its board just added another $60 billion to its repurchase authorization — without expiration. That buyback firepower is among the largest ever announced by a U.S. corporation.

China stays off the map

The geopolitical wild card increasingly lives in Washington, not Santa Clara. U.S. export restrictions continue to block Nvidia’s most advanced chips from reaching China. The company has created workarounds such as the H20, but those shipments have been whipsawed by shifting licensing rules.

In the second quarter, no H20 sales were made to China, though Nvidia booked about $650 million in H20 sales to a non-Chinese customer and recorded a one-time $180 million release of previously reserved H20 inventory. Meanwhile, Beijing regulators have discouraged adoption over security concerns, highlighting how policy rather than technology can dictate demand.

Nvidia’s CFO Colette Kress made clear on the post-release earnings call that China is still a question mark, not a revenue line, as the company continues to navigate a “dynamic external environment.” Kress said that the U.S. government began reviewing licenses for H20 chip sales in late July and that while some Chinese customers have technically been approved, no shipments have gone out under those licenses. She told investors that, “if geopolitical issues subside, we should ship $2-5 billion in H20 revenue in Q3 — and if we had more orders, we can build more.” She also pressed U.S. officials to green-light Blackwell sales into China, arguing that every licensed deal would strengthen America’s AI stack while keeping Nvidia’s momentum intact.

Huang put the potential in stark terms, calling China a “$50 billion opportunity this year” if Nvidia were able to compete with its most advanced products — and he said that if that’s what the market is today, it could swell by roughly 50% a year, in line with global AI growth.

“It’s fairly important, I think, for the American technology companies to be able to address that market,” Huang said on the call. “We just have to keep advocating the sensibility of and the importance of American tech companies to be able to lead and win the AI race and help make the American tech stack the global standard.”

Earlier this month, Nvidia and the U.S. government agreed to a deal that would give Washington a 15% cut of H20 revenue, though no regulation has codified that agreement and made it official yet.

One stock, market-size stakes

Looking ahead, guidance will be just as closely watched as the backward-looking numbers. Nvidia told investors it expects revenue of about $54 billion (plus or minus 2%) in the third quarter, with non-GAAP gross margins in the 73–74% range. That outlook assumes no H20 shipments to China, suggesting any policy shifts could create upside. Kress said the company expects to exit the fiscal year with margins in the mid-70s, pointing to sustainable profitability even amid volatile geopolitics.

“Over the next couple of years,” Huang said on Nvidia’s earnings call, “we’re going to scale into a $3–$4 trillion AI infrastructure opportunity.”

Beyond the quarter, Nvidia’s results are a proxy for the broader AI economy. Microsoft, Google, Amazon, and Meta are projected to spend more than $325 billion in capital expenditures in 2025, much of it on AI infrastructure. Each cloud capex update now serves as a shadow earnings guide for Nvidia, with Huang noting that the capex “of just the top four hyperscalers has doubled in two years as the AI revolution has gone into full steam.”

“We’re in every cloud for a good reason,” Huang said on the call. “There are a lot of reasons why Nvidia is chosen by every cloud and every startup and every computer company.... We’re a full-stack solution for AI factories.”

Still, the longer Nvidia dominates, the more rivals circle the moat. AMD is pushing its MI300 accelerators, while hyperscalers are designing their own silicon. Even modest defections could shave billions off Nvidia’s trajectory. And because Taiwan’s TSMC manufactures most of its chips, any disruption in the supply chain — from fabs to high-bandwidth memory — could rattle results.

For now, though, the story remains familiar: Nvidia delivers another blowout quarter, and the market keeps rewarding it. But with a sky-high valuation riding on quarterly beats — and policy risks that can rewrite its order book overnight — the stakes for Huang’s empire have never been higher.

View Comments