A body for AI: Why investors are betting on humanoid robots

Robots were impressive long before they were useful

Humanoid robots are hardly new. Honda introduced ASIMO in 2000, demonstrating a machine that could walk, climb stairs and interact with people. Boston Dynamics later made humanoids a viral phenomenon with videos of Atlas running, jumping and doing parkour.

The commercial case was less convincing. Engineers could build machines capable of remarkable movements, but getting them to understand unfamiliar environments and perform different tasks reliably remained much harder.

AI may be changing that equation. New models can interpret language and images, plan actions and learn from demonstrations. Instead of programming every movement separately, developers are working toward robots that can understand instructions and adapt their behavior.

China wants to turn humanoids into an industry

China has emerged as the center of the manufacturing race. Unitree is already producing humanoids at scale, while competitors including AgiBot and UBTech are expanding production. Chinese manufacturers also benefit from a supply chain for motors, batteries, sensors and other components built around industries such as electric vehicles and consumer electronics.

Unitree's shares closed 460% above their offer price on their first day of trading, showing how much enthusiasm has already built around the sector.

Factories may come before robot servants

The popular vision is a humanoid that cooks, cleans and handles everyday chores. The nearer-term business opportunity looks considerably less glamorous.

Factories and warehouses provide controlled environments, repetitive tasks and clear labor costs against which a robot can be measured. Tesla is developing Optimus partly for manufacturing, while Figure AI and other developers are targeting industrial and logistics work.

Some robots are already getting a serious test. During an 11-month deployment at BMW's Spartanburg plant in South Carolina, Figure AI's Figure 02 worked on an active production line, moving more than 90,000 components and contributing to the production of more than 30,000 vehicles.

The humanoid shape also has a practical advantage. Factories, stairs, tools and workstations were designed around people, so a machine with roughly human dimensions could work within existing infrastructure.

Investors may be running ahead of the robots

That promise has not yet been proved at mass-market scale. At this week's World Robot Conference in Beijing, humanoids boxed, danced and played table tennis. One robot attempting something more mundane struggled to fold a shirt.

China's rapid expansion also makes it harder to separate genuine commercial demand from the excitement surrounding the industry. Government support and a growing startup ecosystem are helping the technology mature, while manufacturers are still trying to prove that humanoids can deliver enough value for businesses to adopt them at scale.

Unitree captures that tension. The company has moved beyond selling robots as technological curiosities, yet much of the industry still depends on demonstrations, research and pilot projects rather than large-scale commercial deployments.

Physical AI could become the next stage of the trade

Humanoids offer investors something software AI cannot: exposure to the physical economy. Successful machines could eventually work across manufacturing, logistics, healthcare and services, while aging populations in many developed economies add pressure to automate jobs that may become increasingly difficult to staff.

They could also broaden the AI investment chain. Data centers created demand for chips, power and cooling; physical AI adds motors, sensors, batteries and manufacturing capacity.

The industry has already proved that humanoid robots can put on extraordinary demonstrations. Unitree's IPO suggests investors increasingly believe they can become a real business as well. The next stage will depend on something much more ordinary: whether enough companies can find useful work for them to do.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.