China’s humanoid robots have become remarkably good at grabbing attention. They can dance, perform kung-fu, box, run, jump and even execute backflips that would leave most humans reaching for an ice pack.
But there is a rather important question behind all the impressive videos: what can they actually do that somebody is prepared to pay for?
The answer is increasingly encouraging — although it is considerably less glamorous than kung-fu.
The robots are coming
Chinese humanoid robots are already beginning to move into factories, warehouses and other controlled environments. Some are being used for repetitive tasks such as moving components, loading machines, inspecting products and sorting goods.
One Chinese electronics production trial reported a humanoid robot completing 2,283 operations during an eight-hour shift without errors.
Cup of tea anyone?
That is where the real commercial opportunity lies. A robot does not need to be ‘clever’ to make dinner, walk the dog and discuss the economy.
If it can reliably perform one repetitive task for hours without getting tired, injured or demanding a tea break, it can potentially save a company money.
China is particularly well placed to exploit this. It has enormous manufacturing capacity, established electronics and battery supply chains and a huge domestic industrial market.
The objective is increasingly to make humanoid robots cheaper and produce them in large numbers.
Unitree
There are signs that money is already being made. Unitree, one of China’s best-known robot manufacturers, reported 1.7 billion yuan in revenue in 2025 and was profitable. Its forthcoming Shanghai listing has attracted extraordinary investor enthusiasm.
But this does not mean the robot revolution has arrived in your kitchen.
The biggest problem is versatility. A robot can be extraordinarily impressive at one carefully prepared task while struggling with the chaos of an ordinary home.
Picking up identical components on a production line is one thing; finding a dropped sock under the sofa, loading a dishwasher and working out which cupboard contains the washing-up liquid is another.
That is why the immediate future is likely to involve robots as workers rather than robots as servants.
Work ethic
Factories, warehouses, logistics centres, hotels, shops and perhaps hospitals offer predictable environments where a machine can be trained to perform specific jobs.
Home robots will probably take longer because homes are messy, unpredictable and full of objects designed for humans rather than machines.
So, can China’s robots make money? Absolutely — but probably not because they can do backflips.
The backflips sell the dream. The boring eight-hour shift is where the business case is being tested.
And if Chinese manufacturers can make these machines cheap enough, reliable enough and useful enough, the robots really could become everywhere — not dancing on stage, but quietly doing the jobs nobody wants to do.
And the U.S.?
The U.S. is very much in the robot race, and in some respects it may be ahead of China — particularly in combining humanoid robots with advanced AI.
The interesting question is whether America can turn that technological lead into mass production and profitable businesses.
The leading U.S. names include Tesla and Optimus, Figure AI, Agility Robotics and Digit, and Apptronik with Apollo.
Apptronik, for example, raised more than $935 million in its latest funding round to scale Apollo, with investors including Google, Mercedes-Benz, John Deere and AT&T Ventures.
Agility’s Digit is probably one of the clearest examples of an American humanoid moving beyond the demonstration stage.
Digit has been used commercially in logistics, including work for GXO, where robots have been handling totes. Agility is now expanding its manufacturing and AI development capacity in the U.S.
Then there is Figure AI, which has attracted enormous investment and is concentrating on robots capable of learning a range of tasks rather than simply performing one pre-programmed movement.
Figure has demonstrated robots working in industrial environments, including BMW’s manufacturing operations.
Tesla
And, of course, there is Tesla’s Optimus. Tesla has something its rivals desperately want: enormous manufacturing experience, a huge AI operation and the potential ability to produce robots at scale.
Elon Musk’s ambition is considerably bigger than building a warehouse worker — he ultimately envisages a general-purpose robot that can work in factories and homes.
The fascinating difference is that China appears to have an advantage in manufacturing scale and cost, while the U.S. has extraordinary strengths in AI, software, robotics research and access to investment capital.
That makes this less like a traditional technology race and more like a three-way contest:
China: Can we manufacture millions cheaply?
America: Can we make them intelligent?
Everyone else: Can we work out how use and pay for them?
And there is an important reality check. The global humanoid industry is still tiny. Only around 13,000 humanoid robots were shipped worldwide in 2025, although forecasts suggest shipments could rise dramatically over the next decade.
So the U.S. is not losing the robot race. If anything, it is running a different race.
China may be trying to make humanoid robots into mass-produced industrial products.
America is trying to make them into AI-powered workers.
Whichever approach produces a robot that can reliably work an eight-hour shift — and costs less than employing a human to do the same job — could ultimately win.
The dancing and backflips are impressive.
But the real championship event is the payslip.