Barclays: Humanoid Robot Market May Hit $200B in a Decade; China Holds 85% of Deployments

Barclays frames AI's move from digital intelligence to physical intelligence as Automation 3.0. Automation 1.0 began in 1961, when General Motors installed Unimate, the world's first industrial robot. These machines performed single, repetitive tasks in highly structured environments, forcing factories to be redesigned around them.

Automation 2.0 is the digital AI era, in which ChatGPT, large language models and AI agents automate office workflows, data processing and software development. This raises cognitive productivity but stays within the digital realm. Automation 3.0 takes AI into the physical world. Humanoid robots differ because they are built to adapt to human environments, rather than requiring humans to rebuild the world for machines.

Barclays argues that adapting machines to humans is more economical than rebuilding the physical world. The world has about 8 million factories and 2.8 billion buildings. Nearly all physical infrastructure—stair heights of 170–180 mm, door widths of 800–900 mm and control panel heights—is standardized around human ergonomics. Rebuilding that infrastructure to suit robots would be astronomically expensive; giving robots a human form is more economical and scalable.

Real-world data is also human data. Barclays cites Omdia data showing that by 2025 the internet will contain 29 billion videos, most showing people using tools, moving, working and interacting with their surroundings. A humanoid form is naturally suited to learning actions, behavior and environmental interaction from this vast existing video data.