Nvidia earnings recap: Stock rises after earnings blow past Wall Street estimates
Kim Soo-hyeon/Reuters
Nvidia has a message for Wall Street: Relax, the AI train isn't slowing down.
The world's most valuable company by market cap reported its earnings for its fiscal fourth quarter on Wednesday, handily surpassing Wall Street forecasts.
"Enterprise adoption of agents is skyrocketing," said Nvidia CEO Jensen Huang. "Our customers are racing to invest in AI compute."
Nvidia reported $68.13 billion of revenue for the quarter, compared to estimates of $65.91 billion. Earnings per share were $1.62, beating the consensus forecast of $1.53. In its key data center unit, Nvidia reported revenue of $62.3 billion, higher than estimates of $60.36 billion.
Looking ahead, Nvidia breezed past first-quarter analyst forecasts. It sees revenue of $76.44 billion to $79.56 billion, a range well above the estimated $72.78 billion.
The results come at a sensitive time for the AI trade, and Huang looked to sooth Wall Street jitters by touting continued demand for AI compute and its Blackwell and forthcoming Rubin chip platform. CFO Colette Kress said the company shipped "our first Vera Rubin samples" to customers earlier this week.
"'It's really important to realize that inference equals revenues now for our customers," Huang said on a call with analysts, emphasizing that AI was at an "inflection point." OpenClaw got a shoutout too.
On the China front, Nvidia said it has yet to generate any revenue from its H-200 chips, which it hopes to sell in the region.
Huang also gave an update on Nvidia's partnership agreement talks with OpenAI, saying he believed they were "close" to being finalized.
Asked about the concept of data centers in space, an idea that Elon Musk is investing in, the Nvidia CEO said the economics "are poor today," but will "improve over time."
Nvidia stock rose about 3% in after-hours trading following the earnings release, though it pared some of those post-market gains as the earnings call progressed.
Scroll on for the play-by-play from Nvidia's Q&A with analysts:
The stock is trading up .84% as the earnings call ends.
The Nvidia CEO concludes his answer to the final question by saying the AI boom only works if it makes real money.
As companies monetize AI tokens and roll out smarter agents that solve real-world business problems, he says, compute is increasingly turning into revenue.
Huang says because he has a fundamental belief that "compute drives revenues," the capex boom isn't going away.
"Fundamentally, because every single company depends on software, every software will depend on AI," he says. " And so every company will produce tokens."
In order to produce tokens, companies will have to continue to buy more compute, putting Nvidia in a prime position, the CEO says.
"This new way of doing computing is not going to go back," he says. "And so to the extent that we believe that producing tokens is going to be the future of computing, which I believe and I think largely the industry believes, then we're going to be building out this capacity from this point forward. And continue to expand from here."
Jensen Huang is asked about previously outlining the potential for Nvidia to get to $3 to $4 trillion of data center capex by 2030.
On revenue diversity, Kress says Nvidia's top five cloud providers make up about half of its data center revenue, with the other half coming from a diversifying group of customers, including AI startups, enterprises, and sovereign governments — a sign that demand is spreading beyond Big Tech.
Elon Musk is convinced data centers in space are the answer to AI's massive energy requirements. OpenAI CEO Sam Altman has voiced doubt they'll be meaningfully viable for many years.
Huang says the economics aren't great at the moment but will improve.
"In space, the energy, the heat dissipation, it's cold in space, however, there's no there's no air flow, and so the only way to to dissipate here is through conduction, and the radiators that you need to create are fairly large," Huang says.
"So the methods that we use here on Earth are are a little different than the way we would do it in space, but there are many different computing problems that really wants to be done in space, and so Nvidia is already the world's first GPU in space. Hopper is in space," he adds.
Huang says "one of the best use cases of of GPUs in space is imaging."
When asked about Nvidia's growing list of strategic investments — from Anthropic to OpenAI and others, which are also Nvidia customers — Huang says it all comes back to the ecosystem. The goal, he said, is to ensure everything from large language models to robotics is built on Nvidia's platform.
"We want to take the great opportunity that we have as we're in the beginning of this new computing era," Huang adds.
Nvidia's CFO says it's too early to project what Nvidia's gaming business will look like given supply chain constraints on memory.
"As much as we would love to have additional, more supply, we do believe for a couple quarters it is going to be very tight," Kress says during the call. "If things improve by the end of the year, there is an opportunity to think about what that is from a year over year growth, but it's still too early for us to know at this time."
Nvidia was originally a gaming-focused company that turned its focus the high-performance chips foundational to the AI industry. Like other companies, Nvidia is facing supply constraints on memory, and that impacts its gaming chips.
Discussing Nvidia's NVLink technology and engineering advancements, Huang stresses the importance of inference gains.
Inference equals revenue "because agents are generating so many tokens, and the results are so effective," he says.
"When the agents are coding, it's off generating 1000s, 10s of 1000s, hundreds of 1000s, because they're running for, you know, minutes to hours," Huang says. "And so these systems, these agentic systems, are spawning off different agents as a team."
"The number of tokens being generated is really, really gone exponential," he adds. "And so, so we need to inference at a much higher speed, and when you're inferencing at a at a much higher speed, and each one of those tokens are dollarized, it directly translates into revenues."
The stock is now trading up around .4% after hours, down from earlier gains of roughly 1.5%.
Huang says he's not worried about AI companies running out of steam or, more importantly, money, as they spend more than ever on capex.
"I am confident in their cash flow growing, and the reason for that is very simple. We have now seen the inflection of agentic AI and the usefulness of agents," Huang says during the call.
Huang says the industry has reached another "inflection point." In order for companies to continue generating tokens to seize the moment, he says they need to continue investing in compute.
"And that translates directly to growth," he says. "And that translates directly to revenues."
Huang and Kress are now fielding live questions from analysts as the stock trades up 1.47% after hours.
Nvidia CEO Jensen Huang says the company is close to finalizing a partnership with OpenAI. The collaboration was initially outlined last year as a potential $100 billion AI infrastructure effort. Recent reports suggest Nvidia may also make a roughly $30 billion investment.
Huang says Nvidia remains thrilled with its long-standing relationship with OpenAI, which he describes as a once-in-a-generation company.
Nvidia's CEO talks various AI companies and projects, and mentions OpenClaw, the buzzy open-source AI agent whose founder recently got hired by OpenAI.
"AI adoption between Claude Cowork and OpenClaw compute demand is skyrocketing," he says.
"And I'm certain that at this point, with the product productive use of Codex and Claude Code and the the excitement around Claude Cowork, and, you know, just the incredible enthusiasm about OpenClaw and the enterprise versions of them, all of the enterprise ISVs, who are now working on agentic systems on top of their tools platforms — I'm certain at this point that we are at the inflection point."
Kress says Nvidia continues to expect "most of our growth to be driven by data center, consistent with last quarter."
She says the chipmaker is not assuming any data center compute revenue from China, in line with the uncertainty surrounding their business there.
Kress also says that Nvidia will start to include "stock-based compensation expense in our non-GAAP results."
Nvidia says supply for its latest chip architectures remains tight — reflecting broader industry constraints around advanced components like high-bandwidth memory — but adds that long-standing supplier partnerships and forward supply commitments position the company to meet demand beyond the next several quarters.
Kress says that Nvidia shipped "our first Vera Rubin samples" to customers earlier this week. Kress adds that Nvidia expects its next generation of chips to ship to customers in the second half of 2026.
"We expect every cloud model builder to deploy, Vera Rubin," the CFO says.
Kress warns that newly public Chinese AI competitors are making progress and could become a competitive threat.
"Our competitors in China, bolstered by recent IPOs, are making progress and have the potential to disrupt the structure of the global AI industry over the long term," she says.
She adds that the US needs to remain the de facto platform for developers and companies across the globe — including in China, and pledges to continue engaging with both the US and Chinese governments.
Nvidia is making some sales in China, though the chipmaker is generating revenue nowhere near what CEO Jensen Huang has said is possible there.
However, Nvidia also said it has yet to generate any revenue from H-200 chip sales in China, which the Trump administration separately approved earlier this month with some conditions.
According to documents Nvidia filed with the Securities and Exchange Commission, Nvidia has generated roughly $60 million of H20 revenue since the Trump administration reapproved sales of the less-powerful chip in August 2025.
"USG officials expressed an expectation that the USG will receive 15% or more of the revenue generated from licensed sales of our products, but the USG did not publish a regulation codifying such requirement," the company said.
President Donald Trump has asked Nvidia to give the US government a cut of H-200 sales, though the chipmaker said it has yet to see a formal regulation formalizing that arrangement.
"We expect sequential revenue growth throughout calendar 2026 exceeding what was included in the $500 billion Blackwell and Rubin revenue opportunity we shared last year," Kress says.
"We believe we have inventory and supply commitments in place to address future demand, including shipments extending into Calendar 2027."
"Total revenue of 68 billion was up 73% year over year, accelerating from Q3 growth on a sequential basis," CFO Colette Kress says.
CEO Jensen Huang and CFO Colette Kress are on the call, which will begin with the executives reading through prepared remarks on the quarter followed by a live Q&A with analysts.
Key numbers:
Revenue: $68.13 billion, $65.91 billion estimated
EPS: $1.62, $1.53 estimated
Data center revenue: $62.3 billion, $60.36 billion estimated
First-quarter revenue forecast: $76.44 billion to $79.56 billion, $72.78 billion estimated
Full fourth quarter results
- Revenue $68.13 billion, +73% y/y, estimate $65.91 billion
- Data center revenue $62.3 billion, +75% y/y, estimate $60.36
billion - Gaming revenue $3.7 billion, +48% y/y, estimate $4.01 billion
- Professional Visualization revenue $1.3 billion vs. $511
million y/y, estimate $770.7 million - Automotive revenue $604 million, +6% y/y, estimate $643.2
million - Adjusted gross margin 75.2%
- Adjusted operating expenses $5.10 billion, +51% y/y, estimate
$4.96 billion - Adjusted operating income $46.11 billion, +81% y/y, estimate
$44.56 billion - R&D expenses $5.51 billion, +48% y/y, estimate $5.38 billion
- Adjusted EPS $1.62
- Free cash flow $34.90 billion vs. $15.52 billion y/y
There are high expectations — and then there are Nvidia expectations. Analysts expect the chipmaker to report a blockbuster quarter, but that's no longer enough. "The market is focused on durability — specifically, evidence that AI infrastructure demand remains strong and broadening," said Alvin Nguyen, a senior analyst at Forrester.
Capex commitments from Big Tech so far this year suggest infrastructure demand is robust, but strong forward guidance from Nvidia will be needed to lift what has been a fluctuating stock.
Anna Moneymaker/Getty Images
Earnings may serve as a prelude as Nvidia readies its GPU Technology Conference (GTC) in about three weeks.
At the developer event, kicking off on March 16, Huang will deliver one of his trademark keynote speeches, and the company will likely unveil its latest chips, AI tools, and major customer partnerships. On the earnings call, Nvidia could also offer hints about how its Blackwell rollout is progressing so far and what's coming next.
Additionally, any signals about customer demand could help shape expectations heading into the annual event.
Though Nvidia has a dominant hold on the AI chip market, the race is heating up as Big Tech diversifies.
This week, AMD announced a deal to supply 6 gigawatts' worth of its chips to Meta. It follows a similar deal the semiconductor firm made with OpenAI in October. Plus, Google, Meta, Amazon, and Microsoft continue to produce their own custom AI chips.
Daniel Morgan, a senior portfolio manager at the financial services firm Synovus, said Nvidia's management "doesn't seem overly concerned about competition given its strong performance lead." He cited the dominant performance of Nvidia GPUs on benchmarks, and said the company's ecosystem advantage "would be hard to match."
Spencer Platt/Getty Images
The stock market has been pummeled by the AI scare trade recently, as investors react to updates from AI makers like Anthropic. Nvidia's earnings could be catalyst to help lift the market out of its panicked state, one investor says,
Michael Landsberg, CIO of Landsberg Bennett Private Wealth Management, spoke to CNBC about what he'll be listening for on the Nvidia earnings call.
"I think at the end of the day, obviously Nvidia is kind of going to be the bellwether here, and hopefully straighten the ship a little bit in terms of what you would expect going forward," he said.
Goldman Sachs analysts highlighted a recent "dislocation" between Nvidia earnings and the chipmaker's stock price. They say it showcases what they call the challenges of perceived "over-earning."
"A dominant market position in the near-term can eventually give way to fears of over-earning, as competition increases and uncertainty about the sustainability of demand grows," the analysts wrote.
They went on to say that the dynamic can lead to earnings strength in the near term but a contraction of multiples over time. They noted that in the last five months, Nvidia's share price has been roughly flat despite a 37% increase in forward earnings estimates.
Patrick T. Fallon/AFP via Getty Images
One big question as Nvidia reports earnings: What's next on the China front?
The Trump administration initially signaled it would allow sales of less advanced Nvidia Blackwell GPUs in China, but later reversed course amid concerns they could be used to further Beijing's military capabilities. In December, it decided to allow Chinese firms to buy the second most advanced AI chips, though these shipments have stalled.
A new twist came earlier this week, when Reuters reported that Chinese AI lab DeepSeek had trained its newest model on Nvidia's top-tier Blackwell chips at a data center, likely in Inner Mongolia. While Nvidia CEO Jensen Huang has continued to advocate for expanded access to the Chinese market, the latest development underscores how China remains one of Nvidia's biggest geopolitical flashpoints heading into earnings.
Nvidia has led the AI boom thanks to demand for its GPUs to train large language models. But, as my colleague Ali Barr recently pointed out, the game is changing from training models to actually running them — a task known as inference.
In December, Nvidia acquired Groq, an inference-focused chipmaker, sending a big signal that it's shifting with the times. However, there's still an opportunity for rivals to take advantage of this transition. Investors will want reassurance that Nvidia's crown isn't about to slip as inference takes off.
Ben Hider/Getty Images
Even if Nvidia beats earnings estimates, the chipmaker's numbers are more likely to be "heavily scrutinized" by investors after a report from Citrini Research went viral earlier in the week, according to Hardika Singh, an economic strategist at Fundstrat Research.
In the report, written as a hypothetical look back from 2028, the research firm outlined a scenario in which the AI boom leads to a white-collar recession and a stock market crash sometime in the next two years.
"I don't expect too many fireworks this time around," Singh wrote in a client note, pointing to Nvidia's historical performance after reporting earnings. "Investors are already on edge about the AI trade, given how a simple hypothetical scenario in a Substack article caused a selloff in shares of everything from DoorDash to Mastercard."
"The timing particularly sucks for Nvidia, which just can't seem to catch a break," she added.
The bar is high for Nvidia, even more so this quarter as investors rethink parts of the AI trade. Scrutiny will be on not just a solid earnings beat but also on guidance for the current quarter and beyond.
"Though AI spending shows little sign of easing, Nvidia and fellow chip firms face higher bandwidth memory costs amid sector shortages, and this could affect projected margin growth," oe Mazzola, the head trading and derivatives strategist at Charles Schwab, said
On the revenue front, he said that data centers will be the most important driver to watch, and guidance must also beat the average forecast.
"Nvidia remains under pressure every quarter not to just exceed quarterly consensus, but to guide for growth exceeding analysts' average estimates."
via Smith Collection/Gado/Getty Images
Anthony Saglimbene, the chief market strategist at Ameriprise, said to keep an eye out for updates on how successfully Nvidia is ramping up production of its new Rubin model.
"Execution on next-generation platforms, like Rubin, needs to give investors confidence that the company is on track with deployment schedules and that lead times are manageable."
He added that the firm needs to demonstrate that "it can continue to successfully manage an industrial-scale manufacturing ramp not often seen in history."
Joe Moore, Morgan Stanley's analyst covering Nvidia, says he would buy the stock heading into earnings on Wednesday, as industry contacts remain optimistic about the company.
Moore has a $250 price target on the stock, implying 28% upside.
In the weeks following the company's earnings report, Moore said to keep an eye on Jensen Huang's appearances at Morgan Stanley's TMT conference and a GTC developer event for more details about the firm's Rubin platform, as they could be upside catalysts.
Markus Schreiber/AP
Dan Ives, the bullish Wedbush tech analyst, said he expects Nvidia to handily beat earnings estimates.
But, perhaps above all else, investors will be paying attention to CEO Jensen Huang's tone and outlook for chip demand going forward.
"There is one company that is the foundation for the AI Revolution and that is Nvidia with the Godfather of AI Jensen having the best perch and vantage point to discuss overall enterprise AI demand and the appetite for Nvidia's AI chips looking forward," Ives wrote in a client note on Monday.
Fourth quarter
- Revenue estimate $65.91 billion (Bloomberg Consensus)
- Data center revenue estimate $60.36 billion
- Compute revenue estimate $51.61 billion
- Networking revenue estimate $9.02 billion
- Gaming revenue estimate $4.01 billion
- Professional Visualization revenue estimate $770.7 million
- Automotive revenue estimate $643.2 million
- OEM & other revenue estimate $179.4 million
- Adjusted gross margin estimate 75%
- Adjusted operating expenses estimate $4.96 billion
- Adjusted operating income estimate $44.56 billion
- R&D expenses estimate $5.38 billion
- Adjusted EPS estimate $1.53
- Capital expenditure estimate $1.62 billion
FIRST QUARTER - Revenue estimate $72.78 billion
- Adjusted gross margin estimate 75%
- Adjusted operating expenses estimate $5.33 billion
- Capital expenditure estimate $1.65 billion
2027 Year - Revenue estimate $334.81 billion
- Capital expenditure estimate $7.54 billion