Unitree Robotics entered public trading in Shanghai with the kind of first day that turns a technology company into a market symbol. The stock rose as much as 629% and closed at 845 yuan—460% above its 150.80-yuan offer price—after the company raised about 6.1 billion yuan, or $904 million.
The listing makes Unitree mainland China’s first publicly traded humanoid-robot maker. It also gives investors a highly visible benchmark for a category that has mostly been valued through venture rounds, demonstrations, shipment claims, and long-range forecasts.
The market is pricing a category, not just a company
Unitree’s debut reflects a broader bet on embodied AI: intelligent systems moving beyond screens and into machines that can navigate, manipulate objects, and work around people. China has made advanced robotics a strategic priority, while falling component costs and improving AI models have accelerated the pace of new prototypes.
Unitree brings more than a laboratory concept to that story. The Hangzhou company sells both quadruped robots and humanoids, reported roughly 1.7 billion yuan in 2025 revenue, and earned more than 40% of that revenue overseas. Its robots have also become unusually visible through acrobatics, martial-arts demonstrations, and major televised performances.
Visibility helps explain the excitement, but it can blur three different measures: technical capability, units shipped, and economically useful deployment. A robot can be impressive on stage, count as a shipment, and still be far from delivering repeatable value in a factory or warehouse.
Reliable work is the real benchmark
Humanoid robots face a tougher product standard than software. They must interpret messy environments, handle physical variation, recover from mistakes, operate safely near people, and keep working through thousands of cycles. Every failure has a cost in downtime, damaged equipment, or safety risk.
The decisive metric is therefore not whether a robot can complete a chore once. It is whether the system can complete that chore reliably enough, for long enough, at a total cost below the alternatives. Hardware price, maintenance, supervision, energy use, integration, and process redesign all sit inside that calculation.
This is why industrial and commercial deployments matter more than viral demonstrations. Factories and warehouses offer structured environments and measurable tasks, but even there the bar is high: uptime, cycle time, error rate, payload, and return on investment must survive ordinary operations rather than a controlled demo.
Scale can become China’s advantage
China’s robotics ecosystem combines component suppliers, electronics manufacturing, battery expertise, factories willing to pilot automation, and policy support. That creates a powerful feedback loop: more machines in the field produce more operating data, expose more failure modes, and give manufacturers reasons to reduce cost and improve service.
According to Omdia figures reported by the Associated Press, Unitree and AGIBOT each shipped more than 5,000 of the roughly 15,000 humanoid robots shipped globally in 2025. Yet analysts caution that many current machines are still used primarily in research, demonstrations, and performances. Production scale and productive scale are not the same thing.
Unitree says the IPO proceeds will fund advanced robotics research and an expanded manufacturing base. That is strategically coherent: the company now has capital to improve both the intelligence inside the machines and its ability to build them. The public market will expect those investments to become durable revenue and margins.
A valuation benchmark—and a pressure test
The first-day surge gives other robotics companies a powerful reference point, but it also raises the burden of proof. Unitree is no longer judged only against private expectations or engineering milestones. Public investors will eventually ask how much of its demand is repeatable, how quickly deployments expand, and whether service costs fall as the installed base grows.
For product builders, the useful lesson is to measure embodied AI like an operating system for physical work. Track successful task hours, human interventions, recovery time, deployment cost, and value produced—not only dexterity, model scores, or units delivered.
Unitree’s debut shows that capital is ready to believe in humanoid robotics. The next phase will show whether the machines can earn that belief one reliable shift at a time.
Relevant links
- Associated Press: Unitree shares soar in Shanghai trading debut
- Shanghai Stock Exchange: Unitree Robotics launches its STAR Market IPO
- Shanghai Stock Exchange: Unitree clears the listing review
- SunMarc: Nvidia and LG Put Physical AI on the Factory Floor
- SunMarc: AI Leaders Say the Singularity Is Near. What Should Builders Do?