Nvidia forecasts upbeat sales on AI chip demand, talks up long-term prospects

By Arsheeya Bajwa and Stephen Nellis

Feb 25 (Reuters) - Chipmaker Nvidia forecast first-quarter revenue above market estimates on Wednesday, betting on Big Tech's unabated spending on its artificial-intelligence processors.

The company said it had secured enough chip inventory and capacity to meet demand beyond the next several quarters, seeking to alleviate concerns that a supply crunch at its chip contract maker TSMC was getting in the way of its growth. The shortage, though, will affect its gaming business, the company said.

Nvidia Chief Financial Officer Colette Kress said on a conference call with analysts that the company expects sales growth to exceed the $500 billion revenue pipeline for 2026 that the company disclosed in October, though she did not give a timeline beyond saying the company expected growth in each quarter of calendar 2026. 

Shares of the company rose over 3% in extended trading after the results, which were released 10 minutes after the expected time. 

"Our customers are racing to invest in AI compute — the factories powering the AI industrial revolution and their future growth," CEO Jensen Huang said in a statement.

The world's most valuable company expects fiscal first-quarter sales of $78 billion, plus or minus 2%, compared with analysts' average estimate of $72.60 billion, according to data compiled by LSEG.

The fourth-quarter results are good news for AI investors, who are looking to Nvidia's performance to gauge whether the hundreds of billions of dollars that Big Tech is pouring into data center infrastructure are paying off. Hyperscalers Alphabet, Microsoft, Amazon.com and Meta Platforms have forecast total capital expenditure of at least $630 billion in 2026, with most of the spending earmarked for data centers and processors. 

Businesses and governments are spending relentlessly in the race to develop the most sophisticated AI tech, or risk falling behind. 

"It’s clear from Nvidia’s latest numbers and their forecast that concerns about an AI slowdown simply are not showing up yet," said Bob O'Donnell, chief analyst at TECHnalysis Research. "Interestingly, the data center revenues are diversifying across more than just the biggest hyperscalers. This suggests there is still growth opportunity in more places, highlighting the ever-expanding interest in AI compute.”

Nvidia's sales concentration among a few key customers crept up during its just-ended fiscal 2026, with two customers making up 36% of sales. During the previous fiscal year, three customers made up 34% of sales.

NO CHINA REVENUE YET

Still, there are signs of risk to Nvidia's long-held dominance in making AI chips. Smaller rival AMD is set to unveil a new flagship AI server later this year and has clinched deals with Nvidia's top customers, including Meta. 

Meanwhile, Alphabet's Google has emerged as a top rival with a deal to provide Claude chatbot creator Anthropic with its in-house chips called TPUs. Google is also in talks to supply Meta, according to media reports.

Big Tech is increasingly turning inward in the quest for more computing power, dedicating resources to designing in-house chips that they are deploying in their data centers.

"We want to finally see this report be enough to spark the tech sector and really get moving past resistance but it most likely will be a fight," said Ken Mahoney, CEO at Mahoney Asset Management, which holds shares of Nvidia. "This was a good beat and raise, the usual for Nvidia, but based on the reactions preliminarily, it seems a lot was baked in to the cake so far." 

Nvidia reported January-quarter sales of $68.13 billion, beating estimates of $66.21 billion, according to LSEG data. It said adjusted profit came in at $1.62 per share, compared with estimates of $1.53, according to LSEG data.

Nvidia said its forecast for the current quarter did not include any expected revenue from sales of its data-center chips to China. However, the company said it had received licenses this month from the U.S. government to ship "small amounts" of its H200 chips to customers in China.

Analysts and investors were counting on the potential return of Nvidia's AI chip sales to China, earlier restricted due to export curbs placed by the U.S. government.

Huang said last month that he hopes China will allow the company to sell its powerful H200 AI chip in the country and that the license is being finalized.

Rival AMD has added sales of AI chips back to its forecast for the current quarter after it received licenses to ship some of its modified processors to China. 

The company also said it will include stock-based compensation expense in its non-GAAP financial measures, veering away from broader industry trends at a time when tech firms are fighting each other for top AI engineers and researchers.

"Stock-based compensation is a foundational component of Nvidia's compensation program to attract and retain world-class talent," the company said in a statement.

(Reporting by Arsheeya Bajwa and Zaheer Kachwala in Bengaluru, and Stephen Nellis and Max A. Cherney in San Francisco; Editing by Shinjini Ganguli, Sayantani Ghosh and Matthew Lewis)