Unitree’s Shanghai market debut delivered one of the clearest signs yet of investors’ enthusiasm for embodied artificial intelligence. It also opened an unusually wide gap between the value assigned to the company and the present commercial maturity of humanoid robots.
Shares in the Hangzhou-based manufacturer opened at RMB 1,100 on 19 August, 629.4% above the RMB 150.80 offer price. They ended the session at RMB 845, still 460.3% above the offer price but 23.2% below the opening level.
At the close, Unitree was worth about RMB 341.8 billion, or roughly $50 billion at the exchange rate used by Reuters. The IPO had valued it at RMB 61 billion and raised about RMB 6.1 billion. In a single session, the public market therefore multiplied the company’s offer valuation by 5.6.
That change did not reflect a comparable overnight change in Unitree’s factories, order book or robot capabilities. It reflected a sharp repricing of what the company might become.
Unitree’s first-day numbers
| Measure | Verified figure |
|---|---|
| First trading day | 19 August 2026 |
| Exchange and code | Shanghai STAR Market, 688836 |
| IPO price | RMB 150.80 |
| Opening price and reported intraday high | RMB 1,100 |
| Closing price | RMB 845 |
| Gain from IPO price at close | 460.3% |
| Closing market capitalisation | About RMB 341.8bn / $50bn |
| Gross IPO proceeds | About RMB 6.1bn / $904mn |
| Shares offered | 40.45mn, or 10% of post-IPO equity |
| Initially unrestricted shares | 30.09mn, or 7.44% of post-IPO equity |
A debut shaped by scarcity as well as conviction
Unitree Robotics is the first publicly traded humanoid robotics maker in mainland China, although other Chinese robot companies are listed in Hong Kong. That scarcity matters. Only 10% of Unitree’s post-IPO shares were sold in the offering, while the initially unrestricted portion represented 7.44% of the company.
The debut followed an offer that drew more than 8,000 times the initial online retail allocation. Robot Harbour’s earlier IPO report details the offer structure, allocation rate and distinction between Unitree’s trading and subscription codes.
A narrow tradable supply does not explain the scale of demand by itself, but it can magnify price movements when buyers compete for limited stock. Unitree’s offer price already represented 219.23 times the earnings measure used in its listing documents, according to the official disclosure. The first-day close placed a much larger premium on growth that has yet to be delivered.
Reuters reported that the shares opened about 11% lower on Thursday, their second trading day. That was an intraday observation rather than a closing result, but it underlined how unsettled price discovery remained after the debut.
The chief executive’s timetable is more sober
The morning after the listing, founder and chief executive Wang Xingxing told the World Robot Conference in Beijing that embodied intelligence was approaching a “ChatGPT moment”. His benchmark is not another choreographed backflip. It is a robot entering an unfamiliar setting and completing most requested tasks from ordinary voice or text instructions.
Wang said he hoped a robot could eventually perform about 80% of tasks in an unfamiliar household. Yet he also put the major software advance required to reach that point two to three years away in an optimistic case, or five to 10 years away at the outer end of his estimate.
Unitree’s largest commitment of capital and people is now to world models, which help machines interpret and act within physical environments. Wang acknowledged that the company was behind in applying physical-AI models in the real world and said today’s humanoids were not capable enough for mass deployment. The principal constraint, in his account, is the AI controlling decisions and interactions rather than the ability to build an agile machine.
Production leadership is not deployment proof
Unitree has stronger foundations than many early robotics ventures. Reuters describes it as the world’s largest producer of robot dogs and the second-largest humanoid maker by shipments. It is profitable, and its machines are beginning to appear in industrial settings.
But most customers identified in its IPO prospectus are universities and research institutions. That distinction is important. A shipment to a laboratory, a televised performance and a robot doing dependable work through a full factory shift are not equivalent commercial milestones.
The IPO proceeds are intended for robot-model and hardware research, new products and manufacturing capacity. Those investments can strengthen Unitree’s position, but the decisive evidence will be repeatable deployments: machines completing useful work safely, reliably and at an economic advantage over alternatives.
The company must also navigate a more difficult international market. The United States has restricted future imports of foreign-made humanoid and quadruped robots on national-security grounds, limiting one potential route to overseas growth.
What the valuation now has to prove
Unitree’s debut demonstrates that capital is available for a Chinese robotics leader before general-purpose humanoids have reached mass commercial adoption. It does not demonstrate that the technical inflection point has arrived.
The market is effectively pricing a future in which Unitree combines low-cost hardware, manufacturing scale and better robot intelligence to become a platform for physical work. The company’s own chief executive has identified the missing element with unusual clarity: software that generalises beyond rehearsed tasks and controlled environments.
For now, the 460% rise is evidence of investor appetite. Whether it becomes evidence of durable value will depend on what happens away from the trading screen—in homes, laboratories, warehouses and factories. Unitree’s primary listing status is also recorded in Robot Harbour’s Robotics Market Tracker.
This article is an editorial analysis and does not constitute investment advice.