Does a robot cost $2 or $35 an hour? A multitrillion-dollar market depends on how the answer

A robot’s hourly cost sounds like a simple number, but the source material shows why it isn’t. Roland Berger puts one figure at $2 an hour. JPMorgan gets to $10 to $12. Reports from early industrial deployments push past $35. Those numbers only look like they’re contradicting each other. What they really show is that the physical AI business lives or dies on utilization.

A machine that sits around is expensive, no matter how cheap the model looks on paper. If a robot is bought or leased and then only works part of the time, the cost per useful hour climbs fast. The article’s example of a $90,000 industrial robot over five years makes that plain: at 2,000 working hours a year, the hardware alone comes out to about $9 an hour, and the total can get near $12 once maintenance and other costs are added. Cut that to 1,000 hours and the cost per productive hour doubles.

Agility Robotics offers the clearest public window into the problem. Its Digit robots have logged more than 65,000 hours across nine customer sites, including GXO, Schaeffler and Toyota Motor Manufacturing Canada. In its filing ahead of the merger with Churchill Capital Corp XI, Agility disclosed $1.78 million in 2025 net sales, a $140.2 million operating loss, and $300 million in milestone-contingent orders from a single customer. It also laid out its own cost model for a future robot: about $400,000 over five years. That works out to roughly $40 an hour on one shift of 2,000 hours a year, and close to $20 an hour on two shifts.

That is why the article keeps coming back to the factory floor. Demos are easy. Real work is messy. Boxes are misplaced, packaging changes, people step into aisles, and equipment moves. A robot that needs an engineer every time the setting changes isn’t labor. It’s a never-ending integration project.

So the more interesting business model is not selling a robot at all. It is selling work. In a Labor-as-a-Service setup, the customer buys pallets moved, trailers unloaded, parts delivered, or packages sorted. That puts pressure on vendors to keep machines productive, because the economics only improve when the hours add up. And the market is enormous if they can do it: the global manual labor market is measured in tens of trillions of dollars.