When Unitree Technology rang the bell on August 19th, its shares opened at 1,100 yuan, surging 629% and propelling its market capitalization to 444.9 billion yuan, prompting many to hail the arrival of the "Year of Humanoid Robots." Just four trading sessions later, the stock has retreated 45.1% from its opening price, erasing approximately 200 billion yuan in market value. While capital has voted with real money for the future of "embodied intelligence," it now faces an awkward reality—Unitree's revenue still primarily relies on selling research and educational equipment, a truly profitable commercial scenario has yet to be established, and the industry's eagerly anticipated "ChatGPT moment" remains elusive.
What exactly happened to trigger the loss of roughly 200 billion yuan in just four days? At the moment the bell rang, paper wealth peaked. Based on the opening price, founder Wang Xingxing held approximately 121.4 million shares, corresponding to a personal stake exceeding 130 billion yuan, instantly making him the richest person born in the 1990s. The wealth-creation effect extended far beyond the founder. Some Unitree employees also became early beneficiaries, with 45 core staff members joining the millionaire ranks through strategic placement alone. In the secondary market, investors lucky enough to secure one allotment saw peak floating profits of nearly 480,000 yuan, setting a new single-lot profit record since the A-share market adopted its registration system.
No one anticipated the celebration would last less than a day. By the first day's close, the stock had fallen from its 1,100 yuan peak to 845 yuan, trapping those who chased highs during intraday trading. On August 20th, Unitree Technology opened down 6.51% and continued its slide, hitting an intraday low of 685.01 yuan before closing down 18.70%. Investors who had been celebrating their allotments just a day earlier suddenly watched their paper gains evaporate at an alarming pace. By August 24th, the fourth trading day after listing, Unitree's shares fell more than 10% intraday, closing at 603.08 yuan, down 10.3%, with a single-day market value loss exceeding 20 billion yuan. Total market capitalization shrank to 243.9 billion yuan, down 45.1% from the opening price—nearly halved. In just four trading sessions, roughly 200 billion yuan in market value had vanished.
Retail investor forums were filled with lamentations; some reported losing nearly half their investment, while others decried the situation as a trap. Economist Fu Peng, in an interview, stated that Unitree's valuation before listing had already far exceeded market consensus, and the post-listing price surge and volatility were extraordinarily exaggerated. The robotics industry remains in its very early stages, and despite completing its IPO, he believes it is fundamentally a venture capital investment typical of the primary market. Under the fervent expectations of the technology wave, it does not yet qualify as a long-term value or growth stock. He noted that ordinary investors could use small amounts of capital to speculate on the future, but for those seeking stable investment returns, it is not a suitable target.
Can Unitree's fundamentals support a market value of hundreds of billions? The sharp decline ultimately stems from a deep chasm between valuation and fundamentals. In the first half of 2026, Unitree reported revenue of 1.152 billion yuan, up 48.54% year-on-year—respectable numbers, but growth has slowed dramatically from the 332.64% recorded for the full year of 2025. The profit picture is even more concerning: non-GAAP net profit attributable to shareholders was 244 million yuan, down 19.34% year-on-year. Nomura Securities issued a target price of only 370 yuan on the first day of listing, nearly 40% below the August 24th closing price of 603.08 yuan. It's not that institutions lack faith in the future of humanoid robots; rather, the current price has already priced in too many events that haven't yet occurred.
Fu Peng pointed out several practical issues in Unitree's financials. First, the company's end customers are primarily research institutions and universities, with purchases largely funded by research grants—there is almost no repeat purchasing, as the cycle ends once funding is secured and procurement is completed. Second, similar robot assembly has become commonplace in the Shenzhen region, and as an industry benchmark company, Unitree faces considerable competitive pressure. In its response to the second round of inquiry letters, Unitree disclosed that from January to September 2025, humanoid robot revenue reached 595 million yuan, with 73.6% coming from research and education customers including universities and research institutions. Revenue from commercial consumer scenarios accounted for 17.39% of humanoid robot revenue, while industry applications such as enterprise tours, smart manufacturing, and intelligent inspection contributed 9.01%. Previously, the share from research and education customers was even higher.
In other words, humanoid robot revenue is currently propped up by research education, mall tours, and stage performances—selling technology products to research-oriented buyers, essentially "selling teaching tools" rather than "selling productivity." As early as 2025, Zhu Xiaohu, managing partner at Jinshajiang Capital, revealed in an interview that his angel fund had invested in several early embodied intelligence projects over the past few years but had been exiting them in recent months. He had asked the CEOs of these companies the same question: where exactly are your commercialization customers? "I feel like they're all describing customers they've imagined," Zhu said. "Who would spend over a hundred thousand yuan to buy a robot to do these tasks?"
Financial commentator Liu Ge noted that Unitree's founder Wang Xingxing has himself publicly stated that embodied intelligence is still far from its ChatGPT moment. According to Wang's own standard, robots need to complete 80% of tasks in 80% of unfamiliar scenarios, a level that is currently far from being achieved. The current high valuations for humanoid robots borrow directly from Wall Street narrative logic—benchmarking against the global automotive market and imagining humanoid robots entering millions of households in the future, where capturing even a few percent market share would represent an astronomical market. "This valuation model presupposes that humanoid robots can enter homes at scale. But the industry reality is: robot hardware is advancing rapidly, while the AI brain remains stagnant. Without a breakthrough in the brain, robots cannot enter homes and will be confined to specialized scenarios like industry, research, and performance—degrading into ordinary equipment manufacturing, no different in essence from machine tool companies," Liu Ge remarked.
In his view, Unitree's reasonable valuation range is currently 600-100 billion yuan, and general-purpose humanoid robots cannot enter households within the next decade. Wang Xingxing himself acknowledges that the "ChatGPT moment" for humanoid robots could arrive in as little as 2-3 years or as long as 5-10 years before the industry truly hits its critical inflection point. But the capital markets cannot wait that long. Unitree's stock volatility is not a rejection of the humanoid robot direction itself, but rather a deep breath taken by the market after a bout of frenzy. The capital markets have already provided ample premium; now it's up to Unitree to prove itself—not just as an internet-famous robot, but as a machine that can genuinely create economic value.
Why did the stock surge so high initially?
Liu Ge attributes the initial surge to multiple factors stacking together. First, Unitree Technology is one of the few companies among hundreds of humanoid robot enterprises that has successfully closed the commercial loop—selling products to research institutions and tourist attractions with real shipment volumes, real sales revenue, and actual profitability. This is its biggest fundamental difference from peers. However, he admits this level of profit provides very weak support for the post-listing high stock price. The real drivers pushing up market value were subsequent layers: it's the first pure-play humanoid robot stock on the A-share market, carrying inherent scarcity value. If the same quality company were listed in Hong Kong or the US, its valuation might be cut by five to ten times. But the A-share market needs an embodied intelligence benchmark, and after capital captured this signal, funds rushed in, further strengthening market expectations. Finally, after listing, retail investors' enthusiasm for chasing highs pushed the stock even higher above institutional pricing.
Why has the stock fallen so dramatically?
Liu Ge believes the core issue is limited application scenarios. Wang Xingxing has publicly stated that embodied intelligence is far from its ChatGPT moment, setting a standard that robots must complete 80% of tasks in 80% of unfamiliar scenarios—currently far from being achieved. The high valuations for humanoid robots today borrow from Wall Street's narrative logic, benchmarking against the global automotive market and fantasizing about humanoid robots entering millions of households, where capturing even a few percent of market share would represent an enormous market. This valuation model presupposes humanoid robots can enter homes at scale. But the industry reality is that robot hardware is advancing rapidly while the AI brain remains stagnant. Without brain breakthroughs, robots cannot enter homes and will be confined to industrial, research, and performance scenarios—degenerating into ordinary equipment manufacturing, no different from machine tool companies. Once the valuation logic collapses, the stock price naturally follows.
Is "brain breakthrough" the same as the ChatGPT moment for humanoid robots?
Liu Ge confirms they are the same thing—the ChatGPT moment refers to a paradigm breakthrough in the robot's AI brain. Looking at AI history shows how difficult such breakthroughs are. Early chess-playing Deep Blue and later AlphaGo led everyone to believe they had found the ultimate path for artificial intelligence, but neither proved to be it; it wasn't until ChatGPT emerged that a true large-model paradigm shift was completed. The same applies to humanoid robots—only when the brain achieves a breakthrough of similar magnitude can robots truly enter homes. Why is this so difficult? The models currently used in humanoid robots derive their underlying logic from autonomous driving. The data volume needed for autonomous driving is comparable to a small town; achieving general-purpose humanoid robots capable of various household tasks would require data on the scale of the solar system. The existing technology route can go from a small town to a county seat, but reaching solar-system-level capability involves orders-of-magnitude differences that cannot be bridged in the short term simply by throwing money and people at the problem.
Can general-purpose humanoid robots ultimately succeed, and when might they truly enter homes?
Liu Ge believes that long-term technological success is possible, but his judgment is that general-purpose humanoid robots cannot enter households within a decade. This assessment is based on the previously mentioned data scale gap and the uncertainty of brain breakthroughs—paradigm shifts are never linear, so you cannot project current progress to determine when the destination will be reached. This doesn't mean embodied intelligence isn't progressing. On the contrary, the deployment of specialized scenarios may happen faster than people expect. Just as food-delivery robots were already widespread before the so-called ChatGPT moment, outdoor inspection and specific care scenarios—such as specialized humanoid robots helping bathe disabled elderly individuals—might also appear relatively quickly. China has advantages in engineering and cost control for specialized robots.
Can Unitree become a track giant like early CATL or BYD?
Liu Ge acknowledges that the track has the potential to birth giants, but the giant may not necessarily be Unitree Technology. His reasoning has several layers. First, Unitree's current advantage lies in hardware and dynamic balance, not the brain—and it's the brain that will determine the endgame for humanoid robots. Second, the brain belongs to the information domain, where technological breakthroughs are easily replicated, and first-mover advantages may not be defensible. Third, the entire track is still in its very early stages, with the competitive landscape far from settled. Referring to internet history, many star companies that once attracted widespread attention eventually disappeared, with later entrants overtaking them. Humanoid robots will likely follow a similar path—today's leader is not necessarily the final winner.
Is there further downside from the current near-halved stock price?
Liu Ge believes the stock will continue to fall. He places the reasonable valuation range at approximately 600-100 billion yuan. This valuation corresponds not to a "general-purpose humanoid robot platform company" but rather to a "specialized robot and component supplier with stable shipments"—essentially an equipment manufacturing valuation logic, not a tech growth stock logic. The company has real customers and stable shipments, including robot dogs, tourist attractions, and university research procurement—these businesses are solid but have low ceilings. Each university purchasing a few units creates a limited market; tourist attractions and performance scenarios cannot support a trillion-yuan company. Having a real business and having enormous imagination space are two entirely different things.
Is breaking through the "brain" the key to maintaining track leadership?
Liu Ge confirms that the core variable is indeed the brain. But the problem is that the brain belongs to the information domain, where technology is easily replicated and cannot be monopolized by Unitree alone behind closed doors. If the industry achieves a major brain-level breakthrough, combined with Unitree's existing excellent body and dynamic balance hardware capabilities, it would indeed gain wings. To use an analogy, it's like having already built a personal computer with excellent hardware, waiting for the Windows operating system to be born. But Windows may not be something you create yourself, and once it's created, all hardware manufacturers can install it. The market remains highly competitive—Unitree has hardware advantages but has not yet formed an absolute moat. Whether it transforms from "current leader" to "final winner" depends on whether it can catch the breakthrough the moment it arrives and convert its hardware advantages into ecosystem advantages.
Unitree is essentially a primary market venture capital investment
Fu Peng views the dramatic volatility as unsurprising. Unitree's pre-listing valuation already far exceeded market consensus, and the post-listing surge and fluctuation were extraordinarily exaggerated. The robotics industry is still in its very early stages—despite completing its IPO, it is fundamentally a venture capital investment typical of the primary market. Under the fervent expectations of the technology wave, it does not yet belong to long-term value or growth targets. Ordinary investors can use small amounts of capital to speculate on the future, but for those seeking stable investment returns, it is not a suitable target. Reviewing the financials reveals several practical issues: end customers are primarily research institutions, with purchases funded by research grants that end once procurement is completed; similar robot assembly is becoming widespread in Shenzhen, creating competitive pressure for the industry benchmark company; and notably, as a benchmark for the humanoid robot industry, the funds raised from listing should be invested in R&D, yet its R&D spending is even lower than the research investment of Muyuan Foods' pig farming operations—a signal in itself that the industry is still in its early stages.
When asked whether Unitree's listing exemplifies the "absorbing star technique" market phenomenon he previously described, Fu Peng confirmed it does. The day Unitree listed was a textbook example of the absorbing star technique—massive amounts of market capital were siphoned away by it. This is not an isolated case; looking at US stock market trends over the past six months reveals similar characteristics: the index appears resilient, but the number of stocks truly providing effective support to the index is rapidly declining. Equal-weighted volatility remains elevated while index volatility is suppressed to low levels—a classic capital concentration structure where funds continuously converge toward a few core targets while liquidity is drained from peripheral assets.
Where to begin your assessment
For investors evaluating Unitree Technology, the fundamental question is whether the current valuation reflects realistic near-term fundamentals or speculative future expectations. The company's revenue concentration in research and education customers, combined with intense competition in the Shenzhen region, suggests near-term growth may remain constrained. The gap between the current stock price and institutional target prices, such as Nomura's 370 yuan figure, indicates significant downside risk remains if the market shifts toward fundamental valuation. The eventual trajectory of Unitree's stock will likely depend on whether the company can demonstrate meaningful progress in commercial applications beyond research institutions and achieve the brain-level breakthroughs that would justify its premium valuation.