Intel says it’s shipped ‘slightly north’ of 100 million AI PCs, which equates to roughly four ‘ZOPS’, or ‘Zetta OPS’ of AI processing power
Here’s a topic we all know and love: AI PCs. The often nebulous term refers primarily to PCs with dedicated NPUs built to perform local AI computing, which I know we in the PC gaming community are all very excited about.
Ah, enough snark. It’s certainly represented a major change in the way the tech industry promotes new systems, and thanks to NPUs cropping up in many of the latest chips, the number of AI PCs has risen dramatically. According to Intel fellow Tom Peterson, its figures are now ‘slightly north’ of 100 million.
“I do think that’s the critical question… you’ve got all these ZOPS out there, are they useful yet? Initially, the floating point units inside PCs were dormant for decades, and then one day, they were just always there.
“And for me, the trigger is Microsoft,” Petersen continued. “So once Microsoft Copilot+ becomes more useful and more ubiquitous, based on the NPU and the GPU, that’s one of these useful options.”
Intel has certainly taken the “build it and they will come” approach to NPUs, although as it stands in 2026, it’s difficult to see why your average user would buy a PC based on its local AI processing capabilities. Dell has recently said that consumers are ‘not buying based on AI’, and I’d say that 100 million+ AI PC figure is likely more reflective of Intel’s huge chip market share, rather than a massive demand for consumer AI hardware.
Intel CPUs turn up in a lot of PCs, and PCs still sell in huge numbers, AI be damned. Anyway, I’ll be going to bed this evening thinking of ZOPS. I’ve never liked the TOPS acronym, but stick a Z in front of it? Now we’re talking.
