Panos Panay is the man in charge of Microsoft Windows and Microsoft hardware, and he has CEO Satya Nadella’s ear. So when he says that AI is the future of Windows, we’re definitely paying attention.
“AI is going to reinvent how you do everything on Windows, quite literally,” he told an audience at AMD’s CES 2023 press conference, pausing for effect.
The context: AMD has just announced its new Ryzen 7000 mobile processors, and the company’s boasting that they’re the first x86 chips to contain a dedicated AI engine — one that, it claims, also happens to be 20 percent faster than the one in Apple’s MacBooks with M2 chips.
(Dragging Apple was a bit of a theme during the early part of AMD’s keynote, though it’s worth noting that Apple is overdue to refresh its laptops with M2 Pro chips, and its existing MacBooks have largely unbeatable battery life. AMD CEO Lisa Su claimed AMD’s new chips can offer 30 hours of battery life, though a slide behind her clarified that they can reach 30 hours of video playback specifically.)
But back to AI: “You’ve told me: every TOP I’m going to put on that machine, I’m going to use,” Su related. “I’m going to try!” Panos laughed during the presentation.
Unfortunately, he didn’t go into much detail about how Windows might use them — for today, his one suggestion was that a PC’s webcam could add background blur, detect eye contact, and automatically frame someone on a conference call while using far less battery, all part of Windows Studio Effects. He namedropped natural language models as well.
That’s not all that compelling by itself, but it’s intriguing to think that Microsoft might have bigger plans, especially because AMD’s mobile chips won’t be the only x86 ones with onboard AI engines for long. They’ll also be included in Intel’s Meteor Lake chips, due later this year, and they’ve been a part of ARM-based chips for some time now.
Correction, January 5th: AMD says its AI engine is up to 20 percent faster than the M2, not the M1 Pro; the M1 Pro comparison it made was in rendering, not AI. We regret the error.