Physical AI sounds like a contradiction in terms. A computer, but a body?

But for the marketing architects, it’s the latest term of art, a buzzword meant to point us citizens toward a bright and promising technological future.

Back here on earth, the term is maybe most useful as a way to understand how automotive companies are thinking about themselves right now: as tech pioneers. It’s also a handy shortcut to understanding how appetizing the automotive industry is for the companies that make chips—what could be a $123 billion opportunity by 2032, up some 85 percent from 2023. The giant CES consumer tech showcase that just took place in Las Vegas always has its share of goofy robot demos, but this year’s presentations showed how the world of robots, cars, and chipsets are growing ever closer.

First, to define (marketing) terms: “Physical AI” is the way tech developers eventually hope that autonomous systems interact with the real world, by using camera and sensor data to truly understand and reason through what’s going on around them, and perform complex tasks to respond. Physical AI is humanoid robots putting in a day’s work on the Hyundai factory floor, as Google DeepMind, Boston Dynamics, and the Korean automaker announced they would do in the coming months. It’s a car driving itself in complex traffic situations, or taking an arguably more complicated job: seamlessly handing off control between a human driver and a software-powered one. Physical AI lets autonomous systems like cameras, robots, and self-driving cars perceive, understand, reason, and perform or orchestrate complicated actions in the real world.

It is no accident that the companies making the loudest noise about physical AI are chipmakers, including Nvidia and ARM. The former announced ...