9.78 million retail investors scrambling for shares, an online allotment rate as low as 0.0181% — setting a new record low in STAR Market history — and the “first humanoid robot stock on the A-share market,” Unitree Technology, had already pushed market sentiment to its peak before even listing. The primary market envisions a market capitalization of RMB 150 billion to 300 billion (approx. USD 22.1 billion to 44.2 billion), and IPO subscribers dream of a paper gain of RMB 200,000 (approx. USD 29,500) per lot. But once the noisy capital shell is peeled away, a sharp paradox is laid bare: this robot company with the world’s top-tier motion performance has yet to achieve large-scale implementation of the robot brain. Does the RMB 61 billion (approx. USD 9 billion) offering market capitalization actually cash in on present-day hard-tech strength, or does it draw in advance on the far-off story of embodied intelligence?
Hardware Breakthrough: A Humanoid Robot Dark Horse That Fought Its Way Out on Its “Body”
In the height of summer 2026, Unitree Technology knocked on the door of the STAR Market, completing the process from acceptance to review in 73 days, and becoming a much-watched humanoid robot benchmark. Looking back over its growth path, Unitree’s rise was a road of hardware breakthrough that ran counter to industry consensus.
In the early years, the industry favored hydraulic solutions, and most institutions were not optimistic about the electric-drive route. Wang Xingxing chose to go all in on electric drive, self-developing the QDD quasi-direct-drive joint and planetary reducer, with the entire motor architecture redesigned. Where the supply chain was immature, they engaged in deep self-development, tackling motors, reducers, and low-level control code all on their own; under small-batch orders, the team even hand-assembled motors, keeping a firm grip on hardware costs.
On commercialization, Unitree chose to “surround the cities from the countryside,” avoiding the illusory fantasy of consumer-side household robots and rooting itself in the market of universities and scientific research institutions. It insisted on selling standardized hardware, refused heavy-customization projects, handed secondary development over to customers, avoided the trap of B-side project-based work where “every order made loses money,” and achieved profitability early on.
The human-machine dance of “Yang BOT” at the Spring Festival Gala became a turning point in Unitree’s fate. The performance, given without regard for return, ignited public awareness, order volume climbed sharply, and industrial capital such as Alibaba, Tencent, and ByteDance successively entered. The G1 humanoid robot pushed the price down to RMB 99,000 (approx. USD 14,600), with its motion capability and whole-machine reliability standing in the world’s first tier, and it also made Unitree one of the few domestic enterprises to achieve large-scale shipment of humanoid whole-machines while turning a profit. Yet behind the dazzling hardware report card, a fatal weakness has always hung overhead.
The Fatal Gap: Beneath a Powerful Body, the “Machine Brain” Has Yet to Take Shape
Hardware capability has already been proven out, but the general-purpose embodied large model — the robot’s “brain” — has yet to truly land on a real machine.
The prospectus candidly discloses that Unitree’s early R&D resources were concentrated on body structure and motion control, and it was only in 2024 that it increased investment in embodied large models; its self-developed general-purpose embodied model has not yet been applied at scale in products, and it lacks the data accumulation of real industrial scenarios. Wang Xingxing has also publicly admitted: the hardware conditions of humanoid robots are already good enough, but the turning point for generalization capability still needs 2 to 10 years, and the probability of success for a self-developed robot brain is less than 50%.
The robot can dance, do backflips, and recover on its own after falling, with stunning demonstration effects; but faced with the non-standard, ever-changing real working conditions of factories, its generalized operation capability is still insufficient.
To make up for this weakness, this IPO raises RMB 4.2 billion (approx. USD 619 million), of which RMB 2.022 billion (approx. USD 298 million) goes toward intelligent robot model R&D, with nearly half the funds used to patch the brain weakness. Another possibility also exists in the industry: if self-development progress falls short of expectations, directly acquiring mature large-model technology through mergers and acquisitions cannot be ruled out. The ceiling of hardware is already in sight, and in the war of the brain, Unitree has only just entered the field.
Industrial Marriage: DeepSeek’s Three-Year “Betrothal Gift,” a Capital Binding of Body and Brain
With self-development carrying uncertainty, industrial cooperation has become another key solution, and DeepSeek founder Liang Wenfeng’s big-handed entry into Unitree’s IPO has become a market focus.
Through DeepSeek’s strategic placement, High-Flyer Quant, and Jiuzhang Asset’s offline IPO subscription, Liang Wenfeng was allotted a combined 1.1916 million shares, with a total investment of about RMB 179.9 million (approx. USD 26.5 million). Of this, the strategic placement portion has a lock-up period of as long as 36 months; this is not an ordinary financial IPO subscription, but more like a three-year industrial “betrothal gift.”
The two sides signed a strategic cooperation memorandum, agreeing on three major cooperation directions: jointly developing an embodied intelligence large model, prioritized cooperation on robot hardware procurement, and mutual prioritized adaptation of large-model training and technical solutions. The logic is very clear: large-model companies need physical robot hardware to verify the true capabilities of embodied intelligence; robot whole-machine enterprises crave an AI brain to make up for their own biggest weakness.
The body side gains brain capability, the brain side gains a physical vehicle, and a single agreement deeply binds the futures of the two enterprises together. The three-year lock-up period is also a real-world test of this “body-brain combination”: whether, after three years, this collaboration can produce a usable product will directly determine the quality of this investment.
The Valuation Illusion: Does RMB 61 Billion Buy the Present, or the Far-Off Future?
Capital has already paid a hefty valuation premium in advance for this future that has yet to be realized.
Unitree’s offering static price-to-earnings ratio is as high as 219.23 times, while the general equipment industry average is only 38.56 times; its price-to-sales ratio is 35.89 times, substantially higher than peer companies such as UBTECH and Dobot. Huxiu’s judgment is very incisive: the RMB 61 billion market capitalization buys not today’s Unitree, but the future Unitree that, after patching the brain, achieves general-purpose operation capability.
Secondary-market IPO-subscription sentiment boiled over completely, with the online subscription multiple reaching 8,288.82 times, and 9.78 million retail investors competing for limited shares. But beneath the fervor, risk signals have already appeared. In the first quarter of 2026, Unitree’s net profit excluding non-recurring items was halved year-on-year, down 52.55%, and its first-half performance is expected to continue declining. An IPO pursued by tens of thousands still saw several thousand shares abandoned in small amounts — a tiny but noteworthy crack amid the wave of revelry.
Primary-market investors give a market-capitalization expectation of RMB 150 billion to 300 billion, and when asked about the logic, there was no shortage of blunt replies like “thought up off the top of the head.” At present, Unitree’s main customers are still universities and laboratories, and large-scale industrial commercialization has not yet been verified. The robot’s stage performances are flawless, but as for landing in real production environments, the entire industry has yet to hand in a standard answer.
Standing on the Mountaintop: After the Revelry Recedes, Technology Is the Final Judge
Capital has already paid a hefty valuation premium in advance for this future that has yet to be realized.
Unitree’s listing has long been more than the capital event of a single enterprise; it is a weathervane for the entire embodied intelligence industry.
With the support of industrial policy, financing in the embodied intelligence track is red-hot, and a large batch of robot enterprises have obtained high valuations. If Unitree stands firm at a hundred-billion-yuan market capitalization, the investment narrative of the entire primary market will be further reinforced; but if its valuation comes under pressure after listing, the valuation system of the whole industry will face turbulence.
The market has split into two completely opposing narratives: the optimists believe this is the beginning of China’s physical AI era, that humanoid robots will replicate the explosive path of the smartphone, and that market capitalizations of a hundred billion or three hundred billion can all be anticipated; the rational perspective repeatedly reminds that technology cannot be force-ripened by capital. Capital can push a valuation to the mountaintop within a few months, but large-model iteration, the accumulation of scenario data, and commercialization implementation all have to be slowly honed in units of years.
Wang Xingxing himself stays clear-headed about this: in making the body, Unitree has already reached the world’s front ranks; in making the brain, this battle will not be easy to fight.
The RMB 61 billion market capitalization is half already-landed hardware hard strength, and half a long-dated option betting on embodied intelligence. The capital revelry is only the first half; when the tide recedes, the capital market will ultimately pay only for products and performance that truly land. The major test that belongs to Unitree, and also to the humanoid robot industry, has only just begun.
[Disclaimer]: The above content reflects analysis of publicly available information, expert insights, and BCC research. It does not constitute investment advice. BCC is not responsible for any losses resulting from reliance on the views expressed herein. Investors should exercise caution.