Unitree’s $9 Billion IPO Tests China’s Bet on Humanoid Robots

Retail participation was especially intense. Reuters reported that the public tranche was more than 8,000 times oversubscribed, while another report based on an exchange filing put the figure at 5,526 times. The discrepancy means the precise multiple should be treated carefully, but both accounts point to the same conclusion: humanoid robotics has attracted unusually strong speculative and thematic interest among Chinese investors.

The offering also gives the sector a valuation reference point. High-profile technology-linked backers and strategic investors have helped put Unitree at the center of China’s AI-robotics narrative, but financial support or affiliation is not proof that customers will buy robots at scale.

Unitree has hardware traction—but not yet a mature market

Unitree makes both humanoid and quadruped robots. Its prospectus, as reported by Forbes, said the company shipped 5,511 humanoid units globally in 2025, more than ten times its 2024 volume. The company also reported more than 33,000 robot-dog sales since its founding.

Those figures give Unitree a meaningful position in a young market: it has visible products, manufacturing experience and a shipment record that many newer competitors have not yet established. But shipment volume alone does not show that humanoid robots are ready for mass adoption.

A separate analysis of the prospectus reported that much of Unitree’s 2025 humanoid volume went to universities and research customers, with a much smaller share reaching industrial applications. That distinction matters because research sales can demonstrate demand for platforms and experimentation without proving that robots are delivering attractive returns in factories or service businesses.

The company’s earlier filing also acknowledged that commercial applications for its quadrupeds and humanoids remained limited. The central question is therefore not whether Unitree can build and ship robots. It is whether customers will use them intensively enough to justify their purchase, maintenance and integration costs.

Where humanoid robots could earn revenue first

The near-term opportunity is more likely to be business-to-business than consumer-led. Potential applications include:

• material handling and selected factory tasks;
• warehousing and logistics;
• inspection and security;
• hazardous or difficult environments;
• research, testing and data collection; and
• training physical-AI systems in real-world settings.

Humanoid form factors are attractive because they are designed to work in environments built for people. In theory, a robot that can navigate existing workspaces may require less infrastructure redesign than a specialized machine. In practice, that advantage matters only if the robot is reliable, safe, dexterous and affordable enough for a specific task.

Consumer uses—such as household assistance, elder care, education, entertainment and companionship—are a longer-term possibility rather than an established market. They require higher standards for safety, autonomy, customer support and usability, often at a lower price than industrial buyers can tolerate.

The opportunity is a data-and-deployment loop

China’s strongest embodied-AI case is not simply that it can manufacture robot bodies. It is that dense industrial supply chains and a large manufacturing base could create many places to deploy robots, collect operating data and improve performance.

A U.S. congressional advisory report describes this as a potential physical-AI data loop: deployment in manufacturing, robotics and research generates specialized real-world data that can improve later systems. If the loop works, more deployments could produce better control and perception, which could make additional deployments easier and more valuable.

China also brings substantial hardware and industrial capabilities to the effort. Analysts point to its manufacturing depth, engineering talent, component ecosystem and ability to iterate products at volume. China’s electricity and grid infrastructure may also support the broader growth of AI systems that require substantial power.

These advantages do not guarantee leadership. They describe the conditions for scaling physical AI, not proof that any one company has solved reliability, autonomy or unit economics.

Why the valuation remains risky

At roughly RMB 61 billion, Unitree’s IPO valuation places considerable weight on future growth. The company had reported 2025 revenue of about RMB 1.7 billion and adjusted net profit of roughly RMB 590 million, according to the Shanghai Stock Exchange’s summary of its prospectus. Reuters also reported a forecast of up to RMB 283 million in first-half attributable profit excluding one-off items.

Those results show that Unitree is not merely a pre-revenue robotics concept. However, current profitability does not by itself validate a valuation built around a much larger future market. The investment case depends on several difficult transitions:

1. From prototypes to dependable products: robots must perform consistently outside demonstrations.
2. From research buyers to operating businesses: customers must use them for tasks with measurable economic value.
3. From unit sales to utilization: revenue may increasingly depend on service, software, maintenance and recurring deployments—not only hardware shipments.
4. From early enthusiasm to repeat orders: customers must buy additional systems after pilot projects.
5. From expensive machines to scalable automation: costs must fall without sacrificing safety or performance.

The risk is that humanoids remain expensive machines suited to narrow roles. The Robot Report noted that Unitree’s average selling price declined sharply between 2023 and the first nine months of 2025, a sign of improving affordability but also a potential pressure on hardware margins. Lower prices can expand the market, yet they may require a much larger volume of sales to preserve profitability.

Unitree is part of a wider Chinese robotics funding cycle

Unitree’s IPO could make public markets more accessible to the wider robotics supply chain, including actuator and motor companies, sensor suppliers, integrators and other embodied-AI developers. It may also intensify competition for engineers, manufacturing capacity and strategic capital.

The listing is not an isolated event. AgiBot, another Chinese humanoid-robot maker, has initiated a Hong Kong IPO process, according to Reuters. The parallel listing activity suggests that investors and companies are treating humanoid robotics as an emerging sector rather than a single-company story.

That funding matters because robotics requires capital well before broad profitability. Companies must finance hardware iteration, manufacturing, training data, safety testing, customer deployments and working capital. A successful public offering can extend that runway while giving early investors a possible exit and private competitors a market benchmark.

China’s position in the global physical-AI race

Unitree’s IPO also complicates the usual framing of the U.S.-China AI competition. Leadership in foundation models and leadership in embodied AI are related, but they are not identical.

The United States retains important advantages in advanced compute, investment conditions and frontier-model performance. Analysts also identify continuing Chinese constraints in advanced chips and lithography. North Asian economies retain major strengths in semiconductor manufacturing, precision components, industrial automation and specialized robotics.

China’s opportunity is to combine those hardware and software inputs with domestic manufacturing scale, industrial customers and real-world deployment data. A 2026 analysis from the U.S.-China Economic and Security Review Commission specifically argues that widespread embodied-AI deployment could help China build proprietary industrial data advantages over time.

That is a strategic possibility, not a settled outcome. The United States may lead on some of the “brain” layer—models and compute—while China competes aggressively on the “body” layer: manufacturing, integration and deployment. The eventual winners may be the companies that connect both layers effectively.

What investors should watch next

Unitree’s IPO shows that investors are willing to fund the humanoid-robot thesis before the market has fully proven itself. The next evidence will need to come from operations rather than subscription statistics:

• the share of shipments going to industrial customers rather than research institutions;
• repeat orders and robot utilization rates;
• customer economics compared with conventional automation;
• hardware gross margins as prices decline;
• reliability, safety and autonomy in unscripted environments; and
• recurring software, service and maintenance revenue.

The clearest interpretation of Unitree’s IPO is therefore balanced. It validates strong confidence that physical AI can become commercially important, and it gives China’s robotics ecosystem capital and visibility. But the roughly $9 billion valuation also assumes that Unitree—and the sector around it—can cross the difficult gap between impressive demonstrations, research shipments and safe, profitable, high-volume operation.