Here is the complete English translation of the article:

Unitree's $440-Billion Debut: The Feasting, the Yardstick, and the Face of a Founder Who Refused to Smile

Unitree Technology, the Hangzhou-based company focused on humanoid robots, went public on the STAR Market. With an issue price of 150.80 yuan, the stock opened at 1,100 yuan, surging 629%, and its market value once touched 444.9 billion yuan. One board lot of 500 shares yielded a floating profit of 474,600 yuan.

On the day of the bell-ringing, the wealth-creation legend was accompanied across social media by another viral image: a candid shot of founder Wang Xingxing. Against a jubilant red backdrop of cheers, Wang's face was cold and stern, brows furrowed, showing no trace of joy.

Netizens joked he was a "billionaire who isn't happy at all." Some interpreted further: "That expression is like someone suddenly thrown onto a grill, forced to deliver a perfect answer sheet for this heaven-sent 400-billion valuation." Wang Xingxing's knitted brow has become the most honest footnote to the current embodied-AI track. Capital is pricing Unitree with the highest-spec sci-fi vision, but as the person who knows the hand best, Wang understands better than anyone: this windfall is drawing down the market's entire imagination for the future of general-purpose robots.

Four trading days later, that figure fell back to 243.9 billion yuan. Nearly 200 billion evaporated, a 45% drop from the peak.

The pressure is now on Wang. For him, the listing is only the beginning of a long challenge. If Unitree remains positioned merely as a research-and-education equipment maker, it will struggle to carry the banner of the embodied-AI vision.

01 The Feast and the Yardstick

On the morning of August 19, in the first minute of Unitree's debut on the STAR Market, the stock rose 629.44%, and total market value briefly touched 444.9 billion yuan.

A-share markets had never seen such a moment before.

The entire embodied-AI track had been hyped for two years in A-shares, yet it had always lacked a domestic whole-machine (integrated system) target to list. Valuations could only benchmark against the long-term vision of Tesla's Optimus, or diverge along component-concept plays; private-market financings referenced each other, marking up prices layer by layer, with none ever tested by the public market.

After Unitree listed, its 219x issue P/E ratio and 35.89x price-to-sales ratio handed the whole track a public ruler — or rather, a set of examination standards.

The market's frenzy is undeniable. Data shows 9.7846 million valid online subscription accounts, breaking the previous record of 9.42 million set by CXMT (Changxin Technology); the lottery win rate was 0.018%; 313 institutions and 11,052 placement targets participated, with effective intended subscription multiple of 2,618x.

Huatai Securities believes that behind Unitree's surge, the robotics industry is shifting from "theme-based pricing" to "volume and demand-verification pricing," and Unitree's listing valuation could become the sector's new "valuation anchor."

Nankai University finance professor Tian Lijun's judgment is more direct: Unitree's listing marks China's humanoid robotics industry moving from the "technical validation" and "storytelling" phase into the commercialization-value verification era of "profitability proving the mettle."

In the first half of 2026, global humanoid robot shipments totaled about 19,100 units, up roughly 275% year-on-year, with Chinese companies accounting for about 97% of the share.

Unitree's own growth curve is also bright enough. From 2023 to 2025, revenue grew from 159 million yuan to 1.699 billion yuan; over the same period, net profit swung from a loss of 11 million yuan to a profit of 278 million yuan.

In 2025, it shipped over 5,500 humanoid robots (excluding wheeled dual-arm robots), ranking first globally. The market is willing to price it because it is one of the few humanoid robotics companies that can deliver both growth and profitability simultaneously.

But this also means all of the valuation and premium rests on whether this curve can continue.

On the second day after listing, Unitree's stock plunged 18.7%, with market value evaporating over 160 billion yuan from its peak.

The market voiced strong doubts about whether the ruler Unitree handed out was calibrated correctly.

02 The Cards in Hand and the Weaknesses

On August 19, after opening at the sky-high price of 1,100 yuan, the stock quickly began to cool, with gains retreating from a peak of 629% to around 485%.

As the track's undisputed focus, the moment Unitree braked, the supply-chain "followers" making reducers, motors, and other components also stumbled collectively, and the entire robotics sector's relevant index was even dragged down 7.5% intraday.

By the close of August 24, Unitree's market value had shrunk sharply back to around 243.9 billion yuan. But even after squeezing out some of the bubble, calculated on the basis of its prior-year core profit, the P/E ratio still stood at as high as 413x.

That means, at its current pace of profit-making, it would take more than 400 years to earn back its current market value.

As of now (August 25), Unitree's stock has basically leveled off around 602.80 yuan (down a marginal 0.05% for the day) — the market has clearly entered a phase of scrutiny and wait-and-see.

Although the stock has fallen from its peak in the clouds, this stabilized tape still sends a message: the secondary market is genuinely willing to pay real money for a "sci-fi future" that has not yet been fully validated.

For the peers still queuing up to list, the clearly priced figure Unitree has carved out is both a tremendous temptation and a brutally cruel yardstick.

After all, the wind has changed. Two years ago, in the embodied-AI circle, founders only needed to make a prototype walk a few steps or do a backflip at a launch event to easily secure large funding rounds.

But now, capital, having seen it all, is hard-pressed to pay for mere "acrobatics" and demo videos. Whether one can secure small-batch commercialization orders, whether yield rates can keep improving, and whether costs have room to fall — these have become the core assessment criteria before investment.

The whole track is destined to keep diverging: head players can still attract heavy bets from local state capital and large industrial capital, while those PPT projects that only churn out flashy videos but can never land in a factory will find the door to fundraising basically shut.

Returning to Unitree's market value, the root of its drawdown lies in its revenue structure.

In the first three quarters of 2025, research-and-education revenue accounted for 73.60% of Unitree's humanoid robotics revenue, while industrial-application revenue was only 9.01%.

Judging from the revenue structure, Unitree is still a hardware company that turns a profit through the research-and-education market — not a robotics company that has proven itself in industrial scenarios.

The secondary market has no way to price it entirely by the imagination of a general-purpose robotics company.

On the other hand, Unitree's profit is also under pressure. In the first half of 2026, while revenue grew 48.54%, non-GAAP net profit declined 19.34% year-on-year. Scale expansion is eroding profit margins.

In production, Unitree only procures basic components such as gears for core parts, self-assembling joint modules and reducers in-house.

This self-development-and-integration strategy keeps control over its supply chain, but it also means heavier cost pressure. In 2025, the cooperation order between supplier Liangzhi Joint (良质关节) and Unitree was about 6 million yuan — not a large scale.

Yet Liangzhi Joint has already landed a new factory in Pinghu, which formally began production in October 2025.

Unitree needs to prove it can expand that 6-million-yuan order to the point of filling the entire factory's capacity.

03 The Showroom and the Assembly Line

In response to media interviews, Wang Xingxing stated that to achieve large-scale commercialization of high-performance general-purpose robots in industrial and home settings, breakthroughs are still needed in the "brain" level — embodied large-model capabilities — and the "dexterous hand" in terms of fine precision and durability.

The biggest technical challenge remains that embodied large models are still in early development overall, with insufficient generalization capabilities.

The founder's claims are backed by regulatory filings. Unitree disclosed in its filings that, given that global embodied large-model technology is still overall in the R&D and testing stage, during the reporting period the company had not yet applied its self-developed general-purpose embodied large model at scale to its robot products.

A pilot is underway. In early 2026, the company's self-developed industrial-grade embodied large model UnifoLM-X1-0 began pilot deployment and testing at its own factory, capable of executing tasks such as joint-motor assembly.

Unitree is not alone — the entire industry is trying to cross the gap from "pilot deployment" to "scaled repeat purchases."

Guojin Securities believes this year is an important node for humanoid robots to deliver on their "0-to-1" promise. Head domestic-brand unit shipments could jump from several thousand to tens of thousands, and leading companies' supply chains and technologies will tend to converge, making global robotics competition even fiercer.

Every player on the track has entered a validation period; Unitree is simply one of the first companies to receive the exam paper.

Industrial manufacturing scenarios, with their relatively controllable environments and highly standardized tasks, are considered the first core domain for commercial landing.

The industry chain's numbers simultaneously show bottlenecks and progress. The localization rate for planetary roller screws — the core of linear actuators — is only 20%, yet Sanhua Intelligent Controls (三花智控) plans to deliver actuator orders corresponding to roughly 5,000 humanoid robot bodies in 2025.

Shortcomings and orders coexist; the industry as a whole is in a semi-mature state.

The market is waiting for Unitree to be the first to walk this path through: letting robots work continuously in factories for hours or even thousands of hours, with stable task success rates, controllable maintenance costs, and total customer cost lower than manual labor or traditional automation — thereby securing sustained repeat orders, rather than just selling a few thousand units.

At the 2026 World Robot Conference, Wang Xingxing acknowledged that the biggest bottleneck to robots truly entering factories and homes at scale remains insufficient efficiency and embodied-AI generalization — new tasks often require retraining.

Showcase value has been repeatedly validated; production value has not yet been proven.

In this sense, Unitree's valuation premium is more like a patience health bar for investors. Product delays, gross-margin declines, and deployment hiccups all drain the bar.

Listed Unitree has become the sector's bellwether — and has placed itself at the center of the storm.

Capital can, overnight, use hundreds of billions to smash out a utopia. But the gears and code in a factory can only be tuned and tempered by time.

From the display case to the assembly line, from the stage to the trading floor, Unitree wants to prove it can carry the banner of embodied AI — which means crossing not only the technology gap, but also overcoming the real-world gravity of humanity's over-inflated expectations of AI.

If you'd like this trimmed to a specific character/word limit (e.g., 100/150 words), split into sections, or translated into German/Italian/Korean/Russian as well, just let me know.