Nvidia reports earnings and guidance beat as AI boom pushes data center revenue up 75%

Nvidia reported better-than-expected fiscal fourth-quarter outcomes on Wednesday, pushed by 75% revenue progress in its core data center enterprise. The inventory rose as a lot as 2% in prolonged buying and selling.

This is how the corporate did, in contrast with estimates from analysts polled by LSEG:

• Earnings per share: $1.62 adjusted vs. $1.53 estimated
• Revenue: $68.13 billion vs. $66.21 billion estimated

Nvidia’s whole revenue for the quarter climbed 73% from $39.3 billion a yr earlier. The corporate now will get over 91% of gross sales from its data center unit, which homes its market main synthetic intelligence chips.

Data center revenue got here in at $62.3 billion for the quarter, forward of expectations for $60.69 billion, in keeping with StreetAccount.

Web earnings virtually doubled to $43 billion, or $1.76 a share, from $22.1 billion, or 89 cents per share, in the identical quarter a yr in the past, the corporate mentioned in a press launch.

Guidance was additionally higher than anticipated. Nvidia mentioned revenue for the fiscal first quarter shall be $78 billion, plus or minus 2%. Analysts have been anticipating $72.6 billion. Nvidia mentioned it isn’t assuming data center revenue from China in its forecast.

Nvidia’s inventory is outperforming all of its megacap friends up to now this yr, as the corporate continues to be the main beneficiary of the AI boom. As of Wednesday’s shut, the shares are up 5% in 2026, whereas the Nasdaq is down 0.4%. The one different firm within the trillion-dollar membership to point out positive aspects this yr is Apple, which is up lower than 1%.

Wall Road obtained preview of what to anticipate from Nvidia when the 4 main hyperscalers — Alphabet, Amazon, Meta and Microsoft — reported quarterly outcomes a number of weeks in the past. Based mostly on their forecasts for capital expenditures together with analyst estimates, mixed capex for the yr might method $700 billion as the tech giants construct out their AI infrastructure.

In its CFO commentary, Nvidia mentioned hyperscalers “remained our largest buyer class,” accounting for simply over 50% of data center revenue.

Inside the data center enterprise, Nvidia reported $10.98 billion in gross sales for the corporate’s networking elements, that are used to attach lots of of graphics processing models. These gross sales have been up 263% yr over yr, reflecting sturdy adoption of the corporate’s NVLink networking expertise, as effectively as its Spectrum-X Ethernet switches with new offers from giants like Meta.

Nvidia’s gaming unit, which was its greatest, recorded revenue progress of 47% from a yr in the past to $3.7 billion, however fell 13% from the earlier quarter. Analysts have speculated that Nvidia might skip the launch of a brand new gaming GPU this yr, as reminiscence constraints power chipmakers to prioritize AI processors. For Nvidia, meaning AI accelerators largely offered in rack-scale methods just like the 72-GPU Grace Blackwell.

Reminiscence has been an space of potential concern for traders due to a world scarcity.

The corporate expects provide constraints to be a headwind to Nvidia’s Gaming enterprise “within the first quarter of fiscal 2027 and past,” finance chief Colette Kress wrote in her commentary.

Pleasure has been constructing for the upcoming launch of Nvidia’s next-generation Vera Rubin rack-scale methods, the successor to Grace Blackwell, later this yr. Kress mentioned on Wednesday’s name that the corporate “shipped our first Vera Rubin samples to clients earlier this week, and we stay on monitor to start manufacturing shipments within the second half of the yr.”

Vera Rubin is anticipated to ship 10 instances extra efficiency per watt, offering vitality effectivity at a time when data facilities face main energy constraints.

The corporate mentioned it is increasing its provide chain past Asia, the place it is concentrated, and into the U.S. and Latin America. Nvidia is now making Blackwell GPUs at Taiwan Semiconductor Manufacturing Firm’s new chip fabrication crops in Arizona, and a few of its rack-scale methods are assembled at a big new Foxconn plant in Mexico.

“These strikes are anticipated to strengthen our provide chain, add resiliency and redundancy, and meet the rising demand for AI infrastructure,” Nvidia mentioned in its monetary submitting. “Our capability to extend manufacturing capabilities will depend upon the native area’s manufacturing ecosystem’s capability to ramp manufacturing provide to the required quantity and on a well timed foundation.”

In automotive, which incorporates chips for automobiles and robots, Nvidia reported gross sales of $604 million for the quarter, up 6% from a yr earlier and under analysts’ expectations of $654.8 million, in keeping with StreetAccount.

For its skilled visualization enterprise, Nvidia reported revenue of $1.32 billion for the quarter, up 159% yr over yr and forward of expectations for $755.4 million, in keeping with StreetAccount.

Nvidia has been pouring cash into massive AI labs and different corporations within the trade, together with taking a big stake in chipmaker Intel. The corporate mentioned in its annual submitting that it invested $17.5 billion in personal corporations and infrastructure funds throughout the yr, “primarily to help early‑stage startups.” These investments “might not grow to be worthwhile within the close to time period, or in any respect, and there could be no assurance that we are going to understand a return on our investments,” Nvidia mentioned.

CEO Jensen Huang informed analysts on Wednesday that Nvidia continues “to work with OpenAI towards a partnership settlement and imagine we’re shut.” The 2 corporations introduced a $100 billion deal in September, however that deal has but to finalize.

In its quarterly submitting Wednesday, Nvidia famous that there isn’t any assurance “{that a} transaction shall be accomplished.”

That is breaking information. Please test again for updates.

— CNBC’s Salvador Rodriguez contributed to this report.

WATCH: First have a look at Nvidia’s subsequent AI system, Vera Rubin