The processors powering AI PCs

We’ve all become used to cores and gigahertz as the key stats for processors, but as we slide deeper into 2024 that is set to change. Now, the silicon makers are starting to talk about TOPS, which stands for trillions (or tera) operations per second. The sheer magnitude of the number here gives you some idea of AI PCs’ potential power, as – in the right situation – they can achieve so much in a fraction of a second.

Here, we provide a quick look at what AMD, Apple, Intel and Qualcomm have to offer with their latest processors.

APPLE M3 18 TOPS (NPU only)

When it comes to NPUs built into silicon, Apple had a head start on its rivals. The M1 appeared in 2020, and while Apple didn’t state its power the internet’s best guess (from @t3mporarybl1p on X) is 11.3 TOPS. It’s perhaps surprising, then, that the M3 is stuck on 18 TOPS. Apple claims the new MacBook Air ( see p54) is the “world’s best consumer laptop for AI”. While that may not be true of the hardware, Apple can rely on a strong software ecosystem, not least because developers have been exploiting the Neural Engine for so long.

AMD RYZEN 7040 AND 8040 SERIES 7040: 33 TOPS 8040: 39 TOPS

AMD beat Intel to the AI punch by almost a year, with its CEO Lisa Su announcing the Ryzen 7040 series way back in January 2023 and the first laptops available by March last year. Then in December, it announced the updated 8040 series, with the promise of a slight boost in performance, too. It’s a little confusing, as not every 7040 and 8040 series chip includes Ryzen AI – AMD’s name for its NPU – so you should check each processor’s listing before buying.