The trend in one sentence
Humanoid robotics is moving from demonstration toward selective deployment, but the useful question is not how human the robot looks. It is whether it can perform economically valuable work reliably.
Why humanoid form matters
Much of the physical world was designed around the human body: stairs, doors, shelves, tools, workstations and warehouses.
A general-purpose robot with a roughly human form could theoretically operate in those environments without requiring the environment itself to be rebuilt.
That is the attraction.
Why caution matters
A compelling demonstration is not the same thing as an economically reliable worker.
Real deployment depends on safety, uptime, dexterity, battery life, maintenance, training, integration and total cost. Narrow industrial and logistics tasks are therefore more useful evidence than spectacular consumer demonstrations.
Jim Carroll’s own 2026 retrospective explicitly treated the category as partly right and partly hype-inflated, including some of his own earlier enthusiasm.
That distinction belongs in the trend.
What to watch
Watch paid deployments, hours worked without intervention, task success rates, cost per productive hour, deployment in structured environments and the speed at which AI improvements transfer into physical reliability.
The strategic question
Where does a humanoid form create enough economic advantage to beat a specialized machine?