It’s no surprise that the conversation has moved beyond plain AI and data centres to the companies now called humanoid makers. One of them is Unitree, the Hangzhou firm taking on Tesla and Boston Dynamics with its G1, H1 and R1 robots. It has just priced a Shanghai IPO that values the company at 219 times its 2025 earnings and 36 times sales. Retail demand was more than 8,000 times oversubscribed. Is that valuation simply too high?
Unitree, founded in 2016 by engineer Wang Xingxing, first made its name with relatively inexpensive quadruped robots. Its humanoid models have gone viral for running, dancing and performing martial arts. Revenue rose more than fourfold to nearly 1.7 billion yuan ($252 million) in 2025, with adjusted net profit of about 600 million yuan. Overseas sales already contribute more than 40 percent of the total. The company is seeking to raise 6.1 billion yuan ($904 million) at 150.80 yuan a share and will become the first mainland-listed humanoid robot maker. Proceeds are earmarked for embodied-AI model development, robot-body research and manufacturing expansion.
On paper this looks like another chapter in Beijing’s push to dominate what is called embodied intelligence. In practice it looks more like a familiar capital-markets move: raise large sums from public investors while commercial demand is still mostly universities, government-backed demonstration projects and early industrial pilots. All while the technology itself still has well-documented reliability problems.
AgiBot has already overtaken Unitree on shipments
According to Smart Analytics Global data, global humanoid robot shipments reached roughly 19,100 units in the first half of 2026. Up 272 percent year-on-year. AgiBot shipped about 8,400 units (44 percent share), overtaking Unitree’s 5,900 units (31 percent). Together the two Chinese firms controlled roughly 75 percent of worldwide shipments. Industrial and commercial uses now account for more than 70 percent of the total. Full-year 2026 volume is projected to approach 60,000 units, with China accounting for more than 97 percent of units shipped and more than 85 percent of demand.
That dominance is impressive. So are the implications. Unitree’s own prospectus states that large-scale commercial adoption remains uncertain. Challenges listed include task generalization, endurance, safety and the ability to operate reliably in unstructured environments. Humanoids still struggle with reliability, dexterity and performing varied tasks for long periods without human intervention.
The robots that break toes and get in trouble with the police
A Unitree G1 already has a public life of its own in Europe. In Warsaw the same model, given a cheeky Polish nickname as “Edward Warchocki” by two local creators, has become a social-media phenomenon. It walks the streets, runs its own accounts, gets told off by the police, chases wild boars, and has gone viral globally as a result. The actions of Edward have sparked exactly the debate the creators intended: what happens when these machines leave the lab and enter public space.
Other public failures are less charming. Chinese humanoids have face-planted, smashed plates, chased wildlife and, in at least one documented case involving a Unitree G1, broken a person’s toe. China may control around 90 percent of the humanoid market by some measures, yet the machines still frequently cannot walk reliably on uneven surfaces or navigate unexpected obstacles for sustained periods.
Why list now?
The capital needs are genuine. Developing embodied-AI models, collecting training data, hiring engineers and expanding manufacturing capacity is expensive, and it is happening years before clear factory or household demand is proven. Public markets in China are providing that capital while government support and retail enthusiasm remain high. State-backed funds hold meaningful stakes (Beijing Robotics Industry Development Fund 3.83 percent, China Internet Investment Fund 2.11 percent), though direct state ownership is listed at only 0.67 percent. Major private backers include Meituan, Tencent, Alibaba and others.
Unitree itself warns that geopolitical tensions and potential U.S. trade, procurement or sales restrictions could weigh on its international expansion. This is the same overseas revenue that already contributes more than 40 percent of sales. Other Chinese players are also racing: Leju has filed for a Shenzhen ChiNext listing; AgiBot has begun preparations for a Hong Kong IPO.
Capital markets disruption or classic over-valuation?
Humanoid robots are not new to capital markets. Tesla is developing Optimus and Boston Dynamics continues to advance Atlas. What sets the Chinese makers apart is cost. They produce sophisticated machines at prices far below their overseas rivals.
Paying 219 times 2025 earnings and 36 times sales for a company whose own prospectus flags uncertain large-scale adoption is aggressive even by Chinese tech-IPO standards. Retail oversubscription of more than 8,000 times tells you more about domestic liquidity and narrative than about proven unit economics or technological readiness.
Will Unitree (and the wider cohort) justify the multiple by turning demonstration robots into reliable industrial tools? Or will the listing wave simply transfer risk from venture and state funds onto public shareholders while the machines are still learning not to fall over?
Only the next few years of actual deployments, not some viral videos, can answer that.
Author: Andy Samu
See Also:
Who is Edward Warchocki: the Robot Taking Over Warsaw
How Did AGIBOT Overtake Unitree in Humanoid Robots, and Why Is the Market Exploding? – MRKT3.0
