Unitree’s IPO Whiplash Tests China’s Humanoid Robot Boom

Unitree Robotics’ turbulent first week as a public company is a useful reality check for the humanoid-robot industry. Its shares closed 460% above their IPO price on Shanghai’s STAR Market on August 19, 2026, after an even larger intraday jump; the company briefly reached about 445 billion yuan, or $66 billion, in market value. Within days, the stock had fallen roughly 45% from its debut peak. 1719

That volatility is not evidence that China’s push into humanoid robotics will fail. China has a powerful manufacturing base, a dominant share of global shipments, and clear strategic reasons to automate work as its population ages. But Unitree’s trading made one issue impossible to ignore: investors are valuing a future market for capable general-purpose robots while much of today’s market remains early-stage, narrow, and difficult to assess by conventional operating metrics. 24

A valuation built on future capability

Unitree’s IPO was priced at a valuation equal to 219 times its 2025 earnings and 36 times sales, according to Reuters. After the post-listing decline, Reuters’ Breakingviews still described the company as trading at more than 400 times annualized forecast first-half earnings. Such multiples imply unusually high expectations for growth, margins, and eventual market size. 31

The market’s enthusiasm was based on a credible long-term story: humanoid robots could eventually perform valuable physical work across factories, logistics, services, and homes. Yet a large theoretical market does not establish current earnings power. A sharp correction can therefore be read as a repricing of the distance between that promise and today’s commercially proven use cases.

It is also important to distinguish the company’s closing valuation from its intraday high. Unitree closed its first day at 845 yuan per share, up 460% from its 150.80-yuan IPO price and worth roughly $50 billion by Reuters’ estimate. The approximately $66 billion figure refers to the earlier intraday peak. 2019

Shipment leadership is not the same as end-market demand

China’s production advantage is real. BofA Global Research estimated that the global humanoid-robot industry shipped about 20,000 units in the previous year, with Chinese manufacturers accounting for 95% of that total. China’s industry ministry expected the country to build more than 100,000 humanoids in 2026. 2

But shipment counts alone do not reveal whether customers can earn an attractive return from the machines. Reuters reported that many Unitree products are sold to research institutions, while Chatham House said Unitree’s prospectus showed the largest share of its humanoid revenue came from universities, academic bodies, and research organizations. 208

Those buyers can be important early adopters. They help fund experimentation, generate training data, and encourage component development. Still, research purchases and publicly supported pilots are different from repeat orders by factories or service businesses that can demonstrate lower labor costs, higher throughput, or better safety after accounting for supervision, maintenance, and downtime.

That is the central commercial question: not whether China can build and ship robots, but whether a broad base of customers will keep buying them without relying on subsidies, demonstrations, or experimental budgets.

State support can build capacity before it proves demand

Humanoid robotics is a strategic priority for Beijing, which sees automation as a response to an aging population and a shrinking workforce, as well as a way to strengthen advanced manufacturing. Reuters reported that authorities had allocated at least $20 billion in development subsidies since 2024. 4

Such support can be productive. It can lower component costs, expand supply chains, speed up iteration, and help domestic companies reach manufacturing scale. China’s strong shipment share is consistent with those industrial advantages. 2

The same approach also carries a familiar risk: supply can expand faster than economically sustainable applications. Chinese officials have warned of bubble risks in a market with around 150 humanoid-robot manufacturers. A competitive shakeout would not necessarily be a failure of the technology; it may be the process through which a crowded field narrows to companies with differentiated products, reliable operations, and viable unit economics. 4

The hard problem is reliable embodied intelligence

Humanoids can perform impressive, choreographed actions, but operating autonomously in ordinary, changing environments is far harder. Unitree founder Wang Xingxing said robots still struggle to generalize across unfamiliar tasks and execute precise movements. 13

Research from MERICS similarly found that Chinese humanoids still lack precision and dexterity, are mostly used for limited tasks and site-specific trials, and remain far from fully autonomous machines that can perceive and respond in real time in the physical world.

These limitations matter because a useful commercial robot needs more than a striking demo. It must work consistently around people and clutter, cope with variation, recover from mistakes, and do so with acceptable uptime and maintenance costs. Until those conditions are met, many deployments will remain specialized rather than general-purpose.

The “ChatGPT moment” is a practical benchmark

Wang has described the industry’s potential breakthrough as a robot placed in an unfamiliar home that can complete about 80% of tasks through text or voice commands. He said that milestone could arrive in two to three years if progress is fast, or five to 10 years if it is slow. 13

That benchmark is useful because it connects technical progress to real demand. If robots can safely complete a wide range of unfamiliar tasks with little setup or supervision, potential buyers could expand far beyond laboratories and pilots. If they cannot, the industry may still grow—but through more constrained applications with slower revenue expansion than lofty valuations assume.

What investors should watch next

The best evidence that the sector is maturing will be operational rather than promotional:

• repeat orders from commercial customers;
• deployments that extend beyond research and demonstration settings;
• improving uptime, safety, and task success rates;
• lower total ownership costs, including maintenance and supervision; and
• improving margins alongside growing revenue.

The closest historical analogy is not a simple verdict that humanoids are either a bubble or the next industrial revolution. China’s electric-vehicle industry shows how policy support and manufacturing scale can create globally competitive companies, while a crowded field still produces losers. The dot-com comparison makes a related point: a transformative technology can endure even when many early expectations and valuations do not. 14

Unitree’s decline is therefore best understood as a demand for proof. China may retain a major advantage in building humanoid robots at scale. Whether today’s leading valuations are justified depends on the next step—turning that production strength into reliable, repeatable, economically compelling work in the real world.