Nvidia’s big show hits its marks but plays to tough crowd

The world, to be sure, has changed a lot over the past year since Nvidia hosted its previous GTC conference. Political and economic instability both domestically and globally have made investors edgy and damped the appetite for riskier tech stocks in particular. The Nasdaq Composite was down nearly 8% for the year to date ahead of Nvidia Chief Executive Jensen Huang’s Tuesday GTC keynote address, having more than doubled in 2024.

The company’s stock slid again by more than 3% Tuesday, as Huang’s keynote came during another broad market selloff. The stock clawed back some of that ground on Wednesday, but Nvidia is still down 12.5% for the year compared with a blistering 81% gain at this point last year. The five other megacap tech giants commanding market caps above $1 trillion have averaged a 9% drop so far this year.

For its own business, Nvidia has also had to contend with worries about increased competition and developments such as China’s DeepSeek, which has made claims that seem to undercut the need to spend billions on Nvidia’s GPU chips to produce advanced AI models and tools.

Huang answered both at this year’s GTC, spending a good portion of his keynote address demonstrating that the sort of AI reasoning popularized by DeepSeek actually requires more powerful computing capabilities. “The understanding of R-1 was completely wrong," he told financial analysts on Wednesday, speaking of the open-source large language model that DeepSeek developed.

As far as the competition goes, analysts roundly agree that Nvidia remains far ahead. “The roadmap looks really solid, and their capability gap vs competitors across their entire massive stack continues to widen," wrote Stacy Rasgon of Bernstein in a note to clients on Wednesday. In his report, Mark Lipacis of Evercore ISI said operators of major data centers have told him that Nvidia has an eight-year lead over competitors.

CoreWeave, a provider of AI-focused data centers, said in initial-public-offering papers filed earlier this month that it uses only GPU chips from Nvidia because “our current customers have contractually specified our use of Nvidia GPUs."

Rivals will have a hard time closing that distance. Nvidia’s financial muscle is unmatched in the chip sector with nearly $61 billion in trailing 12-month free cash flow, triple that of Broadcom and 25 times as large as Advanced Micro Devices, or AMD. Nvidia also seems to have plenty of room to throttle up its own investment; the company’s annual R&D outlay is only 10% of revenue, compared with an 18% average for chip companies on the PHLX Semiconductor Index, according to data from S&P Global Market Intelligence.

Nvidia’s toughest competitor seems to be its own past performance and the expectations that the past creates for the future. Several analysts noted that the big announcements from this year’s conference—including the Blackwell Ultra chips coming later this year and the new Vera Rubin chips launching in the latter half of next year—were largely as expected.

Wall Street expects Nvidia’s revenue to grow 57% to a little more than $205 billion this fiscal year, which is impressive for a company that size but still half the pace that Nvidia managed last year.

Strong chip companies often test the laws of physics, but even the best among them can’t always fight the laws of gravity.

Write to Dan Gallagher at dan.gallagher@wsj.com