Artificial intelligence chipmaker Nvidia Corp. delivered a better-than-expected earnings and revenue beat as it posted its hotly anticipated third quarter financial results.

It also offered guidance for the coming quarter that was higher than expectations, yet its stock fell almost 2% in extended trading.

The company reported earnings before certain costs such as stock compensation of 81 cents per share, beating Wall Street’s projection of 75 cents. Revenue for the period rose 94% year-over-year, to $35.08 billion, ahead of the $33.16 billion analyst target.

The numbers show Nvidia is continuing to grow rapidly on rampant enterprise demand for its AI processors, though investors will have noted that its revenue growth slowed in comparison to the previous three quarters, when sales rose by 122%, 262% and 265%.

Even so, the profits kept pouring into the company. Net income rose to $19.3 billion in the quarter, up from just $9.24 billion in the year-ago quarter.

For the current quarter, Nvidia said it’s expecting to see revenue of around $37.5 billion at the midpoint of its guidance range, just above the analysts’ consensus estimate of $37.08 billion. That forecast implies year-over-year revenue growth of about 70% from the previous year, slowing from annual growth of 265% in the year-ago quarter.

AI drives demand

Nvidia’s rampant sales have been driven by incredible demand from cloud infrastructure providers and other enterprises for its iconic graphics processing units, which power the vast majority of AI workloads today. The company has been the single largest beneficiary of the technology industry’s race to embrace AI, and it has emerged to become the world’s most valuable publicly traded company.

Most of the growth stems from Nvidia’s data center business, which covers sales from AI chips and related components and makes up the lion’s share of its revenue. The company attributed $30.8 billion of its sales to that segment, up 112% from a year earlier, beating the Street’s forecast of $28.82 billion.

Not all of those sales were GPUs, though. The company said around $3.1 billion of revenue attributed to the business comes from the sale of networking components such as its InfiniBand products.

“The age of AI is in full steam, propelling a global shift to Nvidia computing,” said Nvidia founder and Chief Executive Jensen Huang (pictured).

Nvidia also reported a gross margin of 73.5%, just above the Street’s estimates, which was attributed to the sale of increasing numbers of data center chips.

Third Bridge analyst Lucas Keh said Nvidia’s data center unit delivered sequential growth of 17% during the quarter, and he expects the company to be able to maintain that rate for some time to come. “Our experts believe this growth rate could continue throughout 2025 as cloud providers increase their appetite for data center GPUs,” he said.

Elsewhere in Nvidia’ business, sales of GPUs for gaming delivered $3.28 billion in revenue, beating the Street’s forecast of $3.03 billion. According to the company, this was the result of increased demand for high-powered GPUs in personal computers and laptops, plus rising console-related revenue.

The much smaller automotive and professional visualization businesses also posted strong results. Sales in the former rose 72%, to $449 million, while in the latter, they increased 17%, to $486 million.

First Blackwell chips shipped

Analysts were also looking for Nvidia’s guidance for its next-generation Blackwell GPUs, which promise to deliver a 30-times performance boost versus the existing H100 chips while reducing energy consumption by up to 25% on some workloads. The company had initially planned to ship the Blackwell chips in the second half of this year, but its plans came unstuck when a design flaw was revealed, causing the launch date to be pushed back to early 2025.

Earlier this week, reports emerged that claimed Blackwell might be subject to further delays from supposed overheating problems when testing clusters of multiple chips linked together in data center server racks, though the company played down such fears.

Nvidia Chief Financial Officer Collette Kress shrugged off the report in a conference call with analysts, saying the company intends to expand production of Blackwell in the fourth quarter and accelerate production through fiscal 2026.

She added that most customers, including the likes of Microsoft Corp., OpenAI and Oracle Corp., have already started receiving samples of the Blackwell GPUs, with 13,000 shipped out so far. “Every customer is racing to be the first to market,” Kress said. “Blackwell is now in the hands of all of our major partners, and they are working to bring up their data centers.

Keh said it was interesting to note that the concentration of revenue attributed to public cloud providers increased on a sequential basis during the quarter. “This points to positive signs of hyperscale adoption for Blackwell, despite the delays,” he explained.

According to Kress, Blackwell shipments are expected to ramp up significantly in the next year, and she thinks the company is on track to generate “several billion dollars” of revenue in the fourth quarter. She added that sales of the current-generation H200 chip also increased substantially in the quarter. The company recently announced the launch of a new and improved version of the H200 GPUs, called the H200 NVL, which provides a 1.5-times memory efficiency increase and 1.2-times bandwidth increase over the H100 NVL.

“Both Hopper and Blackwell systems have certain supply constraints, and the demand for Blackwell is expected to exceed supply for several quarters in fiscal 2026,” Kress added.

Keh said there are indications that Nvidia’s pricing strategy for Blackwell won’t be as aggressive as analysts had previously thought. He said the sequential decline in data center GPU margins was because customers are shifting from older H100 systems to more complex systems such as the H200 and the Blackwell-based B200 chips.

“If the transition to these GPUs is negatively impacting Nvidia’s margins, this confirms Blackwell pricing will not be as aggressive initially, likely to spur adoption against competitive threats like AMD,” he pointed out.

Photo: Robert Hof/SiliconANGLE