If you thought humanoid robots were still parked in science fiction, RBC Capital has other ideas. Its new Imagine report puts the total addressable market for humanoids at $9 trillion by 2050, with another $3 trillion attached to software and services once a robotics app-store economy takes hold.
RBC’s core argument is straightforward. Humanoids succeed because they fit into a world built for humans. They can move through existing environments, use existing tools and avoid the costly retrofitting required by other robotic forms.
Early deployments will concentrate on warehouses, agriculture, logistics and repetitive industrial tasks. The more ambitious household and elder-care applications remain more than twenty years away, although development is already accelerating.
The numbers give the story its heat. Households alone could represent a $2.9 trillion market. China is expected to account for 61% of global demand thanks to manufacturing scale, state backing and a high willingness among consumers to embrace new technology.
RBC expects average unit costs to fall toward $25,000, roughly on par with a mid-range car. In the long run, humanoid robots may be purchased with the same regularity.
The competitive landscape is crowded. Tesla’s Optimus is targeting vertical integration with production expected in 2026. Boston Dynamics continues to push performance with Atlas.
Unitree is pursuing mass-scale adoption with sub-$14k models. Apptronik’s Apollo is already active in Mercedes factories. Figure AI is demonstrating both industrial and domestic use cases, including laundry folding. No dominant player has emerged. This is a global contest running at full sprint.
The economics favour software. Hardware will open the door, but software will capture most of the margin. Apple’s 2025 services share of 26% is cited as the model. Skill packs, training modules and third-party robotics applications form the backbone of the additional $3 trillion TAM.
If RBC is right, humanoids are not a curiosity. They are the next general-purpose computing platform. The race now is not only to build the robots, but to own the operating systems that will govern them.