Figure announced factory deployment. This is

Figure AI signed the deal with BMW.

Not a pilot program. Not a limited trial. Not a “we’ll put one robot in a corner and see what happens” arrangement. A deployment. Humanoid robots working on a factory assembly line alongside human workers.

I’ve been writing about this moment for three years. Since Figure 01 first walked. Since Figure 02 made coffee. Since the robot started learning tasks from video. The whole trajectory was pointing here, and now here is an actual date on an actual factory floor.

Why this is the moment

Demos are theater. Funding rounds are finance. Partnerships are PR. Deployment is commitment. When BMW puts Figure robots on their assembly line, they’re making a bet with real production targets and real consequences if the robots fail.

That level of commitment means something the demos didn’t. It means BMW tested these robots extensively. It means the failure rate is low enough to risk production quality. It means the economics work, at least for specific tasks.

We crossed a line. The line between “humanoid robots can work” and “humanoid robots do work.” Present tense.

The tasks

The robots aren’t building cars from scratch. They’re handling specific repetitive tasks: picking up parts, moving them to stations, placing them in fixtures. The kind of work that’s ergonomically punishing for humans over an 8-hour shift. The kind of work where consistency matters more than creativity.

That’s the wedge. Not the whole factory. A few tasks where the robot’s strengths (tireless, consistent, never calls in sick) outweigh its weaknesses (slow, limited dexterity, can’t improvise).

But wedges widen. That’s what wedges do.

What happens next

More factories. More automakers. More tasks. The progression from “one factory” to “many factories” will follow the same pattern as every industrial technology: slow adoption by risk-tolerant early movers, then rapid adoption once the economics are proven.

I think within three years, most major manufacturers will have humanoid robots in at least one facility. Within five, it’ll be standard. Within ten, a factory without humanoid robots will be unusual.

I’ve been wrong about timelines before. Usually by being too optimistic. But the Figure-BMW deployment isn’t a prediction. It’s happening. The timeline starts now.

The future of work just got a new coworker.

Related thinking: