Humanoid Robot Sales: Reality Versus Growth Forecasts
Humanoid robots dominate AI demonstrations, but Reuters reports that only about 7,000 were sold worldwide in 2025 for industrial and professional-service use.
This Short explains why the number is so surprising, what those robots are actually doing, and why many were purchased to collect real-world data for AI training rather than to perform productive work. Carmakers are still testing only small groups of humanoids, even as Bank of America forecasts 90,000 shipments in 2026 and 1.2 million by 2030.
We look at the enormous gap between today’s pilot projects and tomorrow’s forecasts, China’s role in the robotics race, and the practical hurdles—reliability, safety, cost, dexterity, and return on investment—that will determine whether humanoids become a real workforce or remain impressive demonstrations.
Source: Reuters, September 21, 2026. Reporting by Toby Sterling. Editing by Alexander Smith.
What to watch
Sales, shipments, experiments, and productive installations measure different things. A robot purchased to gather training data is not necessarily replacing a worker. Reliability, safe operation, useful task coverage, and total operating cost will determine whether pilots become sustainable deployments.
Related reading: our coverage of AT&T’s AI automation and workforce changes.
Watch and listen
Watch the YouTube Short above or listen to the full episode on Spotify.
