Nvidia CEO’s swagger isn’t enough to support the stock, as tiny cracks emerge

By Therese Poletti

Nvidia Corp. co-founder and Chief Executive Jensen Huang wasn’t able to completely assess all the concerns from Wall Street on Wednesday, as the chip designer is seeing some of its heady growth slow amid a slightly delayed new-product launch.

Nvidia (NVDA) reported better-than-expected fiscal second-quarter earnings and revenue, but its forecast for the third quarter did not exceed the consensus estimates on Wall Street. Its shares fell during the company’s call with analysts, and was down nearly 7% in after-hours trading. The could translate into its worst post-earnings performance in 2 1/2 years if the pattern continues in regular trading on Thursday.

After major increases in revenue and earnings over the past two years due to strong sales of its graphics processors for use in AI-focused data centers, Nvidia is seeing some of that rampant growth slowing down. Growth rates of both gross margins and data-center sales fell on a sequential basis, compared with the fiscal first quarter.

One issue is that volume production of the company’s next-generation chip family, code-named Blackwell, was delayed by a quarter. Chief Financial Officer Colette Kress told analysts that a “mask change” was made to improve the chip’s production yields, and that Blackwell would ramp in volume in the fourth quarter. Previously, the company had Blackwell expected to be in volume production in the third quarter.

“Hopper demand remains strong and the anticipation for Blackwell is incredible,” Huang told analysts Wednesday, referring to the earlier generation of GPUs, the H100 Hopper family.

But Lucas Keh, an analyst with Third Bridge, said in a note that Nvidia has been cutting its prices on the H100 line, citing that as one potential reason why gross margins had fallen sequentially.

Some analysts also asked Huang about the concerns among some on Wall Street that investors are starting to look for some evidence of return on investment from all the spending that has gone into AI in the past year and a half. After a previous long-winded answer on the same topic, Huang was a bit more concise a second time, explaining that the shift to a new computing paradigm that uses more GPUs rather than just central processing units is more efficient.

“When they build out Hopper-based infrastructure, and soon Blackwell based infrastructure, they start saving money,” Huang said. “That’s tremendous return on investment. And the reason why they start saving money is because data processing saves money…The world of general-purpose computing is shifting to accelerated computing.”

But it’s not clear if that response was good enough, and he may have also made some investors nervous with a reference to all the generative-AI startups that are customers of the cloud-service providers who are buying chips from Nvidia.

“Of course, the number of generative-AI startups is generating tens of billions of dollars of cloud (services), representing opportunities for our cloud partners,” Huang said earlier in the call, perhaps reminding some on Wall Street of the ill-fated Sun Microsystems, which boomed by selling servers to startup companies during the dot-com bubble and suffered mightily when that bubble burst.

Nvidia is pointing to a very big year ahead once its Blackwell family is in volume production, and Wall Street wants to believe that capital spending on chips and equipment will continue. But a few tiny cracks in the story are emerging. Whether they are temporary and inconsequential or warning signs of a larger slowing ahead is a big question that no one can answer quite yet.

-Therese Poletti

This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.

(END) Dow Jones Newswires

08-28-24 2117ET