The world has all the AI computing power it needs—for now

CEO and co-founder of Nvidia, Jensen Huang, has made big bets with the company’s future.
CEO and co-founder of Nvidia, Jensen Huang, has made big bets with the company’s future.
Image: Reuters/ Rick Wilking
We may earn a commission from links on this page.

The chipmaker Nvidia has always been a bellwether for the AI chip industry. Nvidia invested in the technology early by developing ways to run AI on its graphics cards as early as 2006, and as a result, has earned a place in nearly every major datacenter.

But amidst this AI boom—a cornerstone of what some call the fourth industrial revolution—Nvidia’s datacenter revenue has started to shrink. The company reported May 16 it generated $634 million in datacenter revenue in its most recent quarter, which ended April 28. That’s down nearly 10% from $701 million in revenue during the same period last year.

That’s bucking a trend for Nvidia. Even during its historically bad holiday quarter when revenue from gaming, its largest revenue segment, dropped by nearly half, the company’s datacenter segment still recorded year-over-year revenue growth.

So either Nvidia’s datacenter revenue is no longer the indicator it’s always been, or the tech industry doesn’t need to buy as much computing power as it once did.

Nvidia CEO Jensen Huang would have you believe it’s the latter. On a call with investors yesterday, he said that tech companies had actually invested too much in data centers last year, and now needed some time to “digest.”

“[Cloud computing companies] are digesting the capacity they have,” Huang said. “At this point, I think it’s fairly clear that in the second half of last year they took on a little bit too much capacity. And so, everybody is has paused to give themselves a chance to digest.”

This may be the case, but Nvidia isn’t the only game in town anymore. Tech companies like Google, Microsoft, Amazon, Tesla, and basically anyone else with a research budget is exploring creating their own AI chips that are custom-built for the task.

But graphics processors like the ones Nvidia sells are flexible, meaning companies like Facebook or Google could use the same hardware for AI and to support their video streaming platforms.

Nvidia expects its chip business to grow again in coming quarters, perhaps channeling former Chicago Bears coach Mike Ditka.

“The pause in hyperscale spending will pass,” Huang said. “Accelerated computing and AI are the greatest forces in computing today and Nvidia is leading these movements.”