Nvidia, dominant in AI data centers, is looking at consumer PCs again

Nvidia reports quarterly results on Wednesday afternoon. While the computer chip designer has repeatedly posted eye-popping sales figures amid the buildout of AI data centers, the company is now looking toward a future when AI happens outside those data centers, too.

Nvidia started out designing graphics chips that powered high-end PCs, for things like video games. The company then figured out those chips could be adapted for AI. 

“It just had the right technology at the right time,” said Jacob Bourne, an analyst with eMarketer. “I mean, Nvidia has just crushed the AI data center market. That is really what has pushed it to become the world’s most valuable company.”

And while the AI boom has been very, very good for Nvidia, Matt Bryson, managing director of hardware and semiconductors at Wedbush Securities, said investors are starting to wonder.

“Is there the potential that competition changes that dynamic?” he said.

Competition from, say, AMD — another chip maker that announced a partnership with Meta this week. Google and Amazon are also designing data center chips of their own. So, Nvidia is trying to stay ahead by looking beyond data centers, according to Angelo Zino, senior equity analyst with CFRA Research.

“As AI evolves, I think you're going to see an increasing amount of compute necessary at the edge level,” he said.

The edge, as in, actual devices that people use — not data centers. Running AI locally can speed up some computations. Zino said that’s mission-critical for things like robots or autonomous vehicles.

“You're gonna really need to react instantaneously, and you're gonna need to have the compute power right at the device level,” he said.

And for those users who don’t plan to let AI drive them through busy streets, local computing can have benefits even on your standard laptop computer, per eMarketer’s Jacob Bourne.

“First of all, security,” he said.

Users working with sensitive information, like health care or financial data, can more easily keep it private. “Not sharing it through the cloud, but keeping it on your device,” Bourne said.

Nvidia wants more of its chips on those devices as data center competition heats up, he added.