The dramatic descent of Unitree Robotics' stock price from 1,100 yuan to 500 yuan serves as a stark stress test for the valuation logic underpinning the entire embodied intelligence sector. On September 10, the company's shares fell below the 500 yuan mark during morning trading, hitting an intraday low of 497.88 yuan — a record low since its listing. By market close that day, the stock settled at 498.55 yuan, down 2.99%, bringing the total market capitalisation to approximately 201.6 billion yuan, evaporating over 240 billion yuan from its peak of 444.9 billion yuan on debut day.
The Capital Extravaganza
On August 19, amid widespread anticipation, Unitree Robotics, the world's leading humanoid robot shipper by volume, officially debuted on the Shanghai Stock Exchange's STAR Market, earning the title of "China's first humanoid robot stock". Priced at 150.80 yuan per share, the stock surged to 1,100 yuan at the opening bell, with market value briefly exceeding 450 billion yuan. Founder Wang Xingxing, holding approximately 121.4 million shares, ascended to the ranks of China's wealthiest post-90s entrepreneurs, while lucky lottery winners saw paper profits of nearly 480,000 yuan per lot — setting a new record for single-lot gains since the A-share market's registration system was introduced.
Yet this celebration proved fleeting, lasting barely a day. The stock retreated to 845 yuan by the close of trading day one, followed by successive declines over subsequent sessions. By August 26's close, Unitree Robotics shares had fallen to 591.5 yuan, down roughly 46% from the opening price, erasing approximately 200 billion yuan in market value. Despite occasional bounces, the downward trajectory persisted. By September 10, having breached the 500 yuan threshold less than a month after listing, the stock had lost approximately 55% from its peak — a textbook "halving".
Analysts attribute this volatility to multiple converging factors: the company's headline-grabbing narrative fuelled intense IPO subscription enthusiasm, yet only about 30.09 million shares (7.44% of total) were in free float, with a lottery winning rate of just 0.018%. Limited tradable supply met with massive speculative capital — compounded by the STAR Market's no price-limit rule during the first five sessions — driving the stock to extreme levels. However, this surge quickly triggered profit-taking pressures. Some investors liquidated positions to free up capital for upcoming mega-IPOs like YMTC and Enflame Technology, while others, recognising that Unitree Robotics' valuation had diverged sharply from fundamentals, chose to bank their gains. When questioned about the stock's performance, Wang Xingxing stated his hope that investors buy the company's shares "because they identify with the company's value, not for speculative purposes". On the pressures of going public, he responded, "Pressure brings motivation," pledging to deploy raised funds efficiently and accelerate investment project implementation to reward shareholders with solid performance.
The Reality Behind the Profits
Public information reveals that Unitree Robotics, founded by the 1990-born Wang Xingxing on August 26, 2016, counts a star-studded shareholder roster including Tencent, Alibaba, the Robot Fund, and China Internet Investment Fund — often described as "half of China's investment circle". Viewed purely through financial metrics, Unitree Robotics is a profitable enterprise. For full-year 2025, the company generated revenue of 1.699 billion yuan, up 332.64% year-on-year; non-GAAP net profit attributable to parent reached 591 million yuan, up 652.78%, while gross margin on core business improved from 56.74% in 2024 to 60.13%. Humanoid robot shipments exceeded 5,500 units globally — ranking first worldwide. Among seven embodied intelligence companies that disclosed interim reports for the first half of 2026, Unitree Robotics was the only one reporting positive net profit attributable to shareholders. According to its prospectus, quadruped robot sales reached 697 million yuan in 2025, accounting for 41% of revenue; humanoid robot sales hit 868 million yuan, or 51.78%, marking the first time humanoids surpassed quadrupeds as the primary revenue source.
However, digging deeper into the revenue structure reveals a more nuanced "profitability" picture. In the first three quarters of 2025, a stunning 73.6% of Unitree Robotics' humanoid robot revenue came from university and research institution orders, with commercial consumer applications at 17.4% (primarily store displays and cultural tourism performances). Genuine industrial applications — such as smart manufacturing and intelligent inspection — contributed a mere 15.7 million yuan, just 2.64% of revenue. For every 100 yuan of humanoid robots sold, nearly 74 yuan stem from research funding procurement by universities and institutes. This ratio suggests that Unitree Robotics' current humanoid business more closely resembles an "educational tools enterprise" than a "productivity business". These two business models follow entirely different purchasing logic: the former draws on research budgets driven by academic project needs without strict labour cost accounting, while the latter demands rigorous return-on-investment calculations. During this phase — when competitors remain immature, customers show price insensitivity, and demand is fuelled by academic buzz and policy tailwinds — the 60% gross margin enjoys a clearly defined window of opportunity. Data from the Beijing Academy of Artificial Intelligence shows that between April 8, 2021 and March 2025, among over one hundred winning bids by Unitree Robotics, research-focused universities were the primary buyers, with nearly 30 universities having purchased its products. By 2025, the customer mix had begun shifting, with increased orders from state-owned enterprises, central enterprises, and industrial clients. In Q1 2026, revenue reached 423 million yuan, up 68.49% year-on-year, but net profit attributable to parent stood at just 50.01 million yuan, down sharply by 47.69%. The company has been investing heavily in robot hardware and structural R&D, embodied intelligence large models, and motion control algorithms, driving a significant increase in capital expenditure. Revenue grows while profits contract — a warning signal that cannot be ignored for a company selling on the strength of its 60% gross margin.
Four Layers of Risk
Alongside the share price pressure, Unitree Robotics' commercialisation path faces increasingly intense scrutiny. The potential of the humanoid robot industry and Unitree Robotics' technological innovation are both genuine, but the gap between current commercialisation progress and market expectations is widening. As one analyst put it, "Investor expectations for Unitree are too high, to some extent mortgaging its future."
At the second World Humanoid Robot Games held in August 2026, Unitree Robotics' team posted a 100-metre preliminary time of just 12.41 seconds, finishing last in their heat; in the 400-metre final, Tiangong Ultra claimed gold with 38.15 seconds, leaving Unitree empty-handed in sprint events. Just one year earlier, at the inaugural Games, Unitree had swept four gold medals. The company explained that time constraints, the number of new robot models, and pre-competition testing limitations led the team to scale back participation in certain events, with energy focused on mass-production products. In the current embodied intelligence landscape, the "cerebellum" — corresponding to motion control technology — has reached a high level of maturity, but the "brain" responsible for scene understanding and long-horizon task planning remains firmly constrained by massive scene data requirements and technical bottlenecks. As of early 2026, globally compliant, usable real-robot effective data stands at only approximately 500,000 hours — a gap exceeding 99% of what is needed. The industry currently resides in a phase of "mature cerebellum, missing brain intelligence", and Unitree Robotics' accumulation at the "brain" level remains relatively insufficient.
More direct external pressure comes from the regulatory front. According to multiple media reports, following the violent share price swings after Unitree Robotics' listing, the China Securities Regulatory Commission has issued informal "window guidance" to certain investment banks and institutions, raising the bar for IPO approval of humanoid robot startups. Regulators now require candidate companies to satisfy at least one of three conditions: generation of relatively stable recurring revenue, a sustained narrowing trend in losses, or genuine innovation in core technology, product capability, and commercialisation. The regulatory objective is to filter out homogeneous companies that rely primarily on low-price competition and short-term shipment volumes, preventing a flood of similarly modelled, insufficiently profitable enterprises from crowding into the capital markets. Following Unitree Robotics, multiple robot companies have already initiated listing processes on both mainland and Hong Kong exchanges. Statistics show that seven companies whose primary businesses encompass robotics have filed for Hong Kong IPOs, while CloudMinds and Leju Robotics have also submitted A-share IPO applications. The dramatic fluctuations of the industry bellwether inevitably cast a shadow over the valuation expectations of these later entrants.
Between Euphoria and Reflection
The listing and subsequent share price turbulence of Unitree Robotics epitomise a stress test for China's humanoid robot industry. According to UBS statistics, Chinese investment in humanoid robots surged from 450 million yuan in 2022 to 4 billion yuan in 2024 — nearly a ninefold increase in two years. The China Academy of Information and Communications Technology projects the global humanoid robot market will reach 17 billion yuan in 2025, with China's share exceeding 8.5 billion yuan — more than half of the global total. Of the 13,000 humanoid robots delivered worldwide in 2025, the majority came from Chinese enterprises. Industry statistics indicate that in 2025, the sector completed 463 financing rounds with over 150 companies entering the field, yet shipments grew just 17% while valuations soared 300% — a divergence that typifies classic bubble warning signals. Capital is flooding into this track at unprecedented speed, but as capital influx accelerates, the humanoid robot industry may soon face homogenised competition. "When a new product enters the market, customers don't first ask 'how advanced is it?' but 'why do I need it?'" Over the next one to two years, humanoid robots are more likely to undertake relatively repetitive tasks within "semi-structured environments" — such as intra-factory logistics, quality inspection, simple assembly, and pilot applications in hospitals and care facilities. As for broader general-purpose scenario adoption, that may still require three to five years. Ultimately, the success of humanoid robots lies not in flawless demonstrations on exhibition stages, but in consistently creating measurable value across real production lines and everyday environments.
Comments