Unitree’s IPO is being priced as something larger than a single company coming to market.
In a typical listing, a business hands itself over to the secondary market and lets trading and future earnings reports do the work. In Unitree’s case, the sequence looked reversed. Before the bell has even rung, primary-market investors, public-market institutions, industry peers and strategic partners have already placed their expectations, valuation assumptions and next-stage bargaining chips on the company.
Online subscriptions opened on Aug. 10. The offer price was set at RMB 150.80 per share, implying a price-to-earnings ratio of 219 times. At the time, the average P/E ratio for listed peers in the same sector was about 38 times. In the offline bookbuilding process, the effective subscription multiple reached 2,618 times. The public online tranche was even tighter: the initial allotment was only 6.47 million shares, and with 500 shares per lot, the market had fewer than 13,000 winning lots in total. The winning rate fell to roughly 0.02%.
Across stock forums, investors were posting the same message: hoping to win a single lot. The market did not look like it was waiting to test Unitree after listing. It looked like participants were trying to make sure the company got off to a strong start.
This was not a routine IPO
Calling Unitree simply the first robot stock would miss what was happening around the deal.
The prospectus, as described in the source article, shows a sharp shift in the company’s business trajectory over the past three years. Revenue rose from RMB 159 million in 2023 to RMB 1.699 billion in 2025, a compound annual growth rate of 226%. Profit attributable to the parent after deducting non-recurring items moved from a loss of more than RMB 18 million to a profit of RMB 591 million. Net cash flow from operating activities in 2025 reached RMB 670 million. Cash and cash equivalents at period-end stood at RMB 1.419 billion, while the debt-to-asset ratio remained at a relatively low level.
The biggest change came from product mix.
In 2025, Unitree’s humanoid robot products, including H1 and G1, generated RMB 868 million in revenue, accounting for 51.78% of total revenue. Three years earlier, that business contributed less than 2%. Over the period, the company increased the share of self-developed core components such as joint motors, reducers and controllers, cutting the full-machine cost of humanoid robots and shifting its business from quadruped consumer hardware to general-purpose humanoid hardware.
For founder Wang Xingxing, the listing carries a separate burden. The company needs to show the market that a privately run robotics business can generate its own cash and produce large-scale profits by selling hardware.
Unitree’s valuation reaches beyond Unitree
The paper gains for early investors are already substantial.
Based on the offer price, Variable Capital’s RMB 2.09 million investment in 2018 translates into a paper return of more than 174 times, according to the source article. Sequoia China invested a cumulative RMB 102 million, and the market value of its holding is close to RMB 3 billion. Funds linked to Meituan hold 9.65%, with paper gains of more than RMB 3.6 billion.
But the central issue is not the paper profit. It is the pricing benchmark this IPO creates.
Unitree is presented in the article as the first humanoid robot full-system company to reach China’s A-share public market. Robotics companies such as Dobot and DEEP Robotics that are preparing listings, along with industrial groups including XPeng, Lenovo and GAC that are building embodied intelligence businesses, are watching the case closely. For years, embodied AI hardware companies lacked a public-market reference point. Unitree’s issuance valuation of RMB 61 billion now serves as the first open market anchor for China’s humanoid robotics sector.
If the stock trades steadily after listing, RMB 61 billion could become the baseline for companies that follow. If it breaks below the issue price, those waiting in line may have to reassess their own pricing logic. The market is finally being forced to answer a question it had not answered before: what is a real robotics company worth?
Retail investors are noisy, but they are not setting the game
Retail sentiment was amplified during the online subscription stage.
On investor forums, many posts focused on the payoff from winning one lot. Broker estimates cited in the article show that the average first-day gain for A-share IPOs since the start of 2026 was 276%. If Unitree matches that average, the paper profit on one lot could exceed RMB 200,000. If it trades in line with the 466% average first-day gain for STAR Market IPOs this year, the potential profit per lot could move above RMB 350,000.
Yet the loudest participants are not the core players in this deal.
Unitree is offering 40.44 million shares in total, but only 16% of the initial allocation is reserved for online retail investors. Most of the rest went to strategic placement and offline institutional bookbuilding. Three portfolios under the National Social Security Fund subscribed for a combined roughly RMB 141 million. DeepSeek also received an allocation of about RMB 141 million and agreed to the longest lock-up period of 36 months.
On the first day of trading, only about 29.77 million shares will actually be available for circulation, or around 7.36% of total share capital. More than 90% of the shares will still be locked up when the company lists.
That creates a clear contrast. The investors who most want to buy Unitree are getting very little stock, while the investors holding meaningful positions did not enter through a lottery. By that measure, this is not a full public pricing exercise. Retail investors are involved, but they are more like the noisiest audience than the decisive holders of chips.
With float supply scarce and expectations tightly clustered, first-day trading becomes especially sensitive.
Investors are not paying for 2025 earnings alone
That helps explain why the market did not back away from a 219-times earnings multiple. Offline subscriptions still came in at 2,618 times.
The article’s argument is that buyers are not really paying for Unitree’s RMB 591 million in 2025 profit after deducting non-recurring items. They are paying for a future narrative.
There are, in effect, two versions of Unitree. One is the company in the prospectus: revenue of RMB 1.699 billion in 2025, but expected first-half 2026 profit after deducting non-recurring items is projected to decline 6% to 22% year over year, while industry competition is intensifying. The other is the company in the market’s valuation model: one that reaches large-scale production, lands in industrial settings, moves embodied intelligence into production systems, ships globally and eventually stands as a flagship of China’s robotics industry.
The 219-times earnings multiple reflects a premium being paid in advance for that longer-term version. That is also where the risk sits. The more appealing the future story becomes, the easier it is for the current price to pull future gains forward.
Growth comes with unresolved questions
At an online investor meeting on Aug. 7 that lasted three hours, someone asked Wang Xingxing directly whether a P/E ratio above 200 times was too high and whether it had drifted away from fundamentals.
Wang’s answer, according to the article, was straightforward: while many peers are still struggling with losses, Unitree has already achieved profitability.
That may be true, but the company’s fundamentals still contain several issues that are harder to settle.
The first is customer mix.
The prospectus says that in the first three quarters of 2025, revenue from humanoid robots was split 73.60% to research and education, 17.39% to commercial consumption and only 9.01% to industry applications. Within that 9%, lighter use cases such as showroom guidance accounted for a large share. Revenue tied to harder applications like intelligent manufacturing and industrial inspection represented less than 30% of the industry-application segment.
That suggests Unitree has shown that customers are willing to buy robots, but it has not fully shown that industrial customers cannot operate without them. There is still a wide gap between those two stages.
The second question is less obvious and centers on pricing pressure.
Humanoid robot prices are falling quickly. The average industry price was about RMB 590,000 in 2023, dropped to around RMB 160,000 in 2025, and moved closer to RMB 100,000 in the first quarter of 2026. Cost declines have not kept up. From 2023 to the first three quarters of 2025, Unitree’s unit cost fell only from RMB 73,200 to RMB 62,200, or roughly 15%.
In other words, selling prices are falling much faster than costs. The bigger test ahead may not be whether Unitree can keep growing volume. It may be whether the company can defend gross margins of around 60% once the business scales further. Industrial robotics is still a manufacturing business. As customers shift from research institutions to factories, price sensitivity will rise. As more competitors enter, price wars become harder to avoid.
Strong body, weaker brain
Another detail in the prospectus may matter even more than the lofty multiple.
Unitree states that it has not yet deployed its self-developed general embodied large model to robot terminals at scale. In the humanoid robot market, that line carries weight. The industry’s imagination over the last few years has rested on two things: the body and the brain.
Unitree’s body is moving fast. Its brain is still catching up.
Of the roughly RMB 6.1 billion to be raised, more than RMB 2 billion is earmarked for intelligent robot model research and development, with a focus on the “brain” and “cerebellum” of embodied intelligence. DeepSeek’s appearance as a strategic investor fits that picture. Robot companies need stronger model capabilities, while large-model companies need more realistic physical carriers.
What has not yet been proven is whether the two sides have actually fused in a meaningful way.
Wang also defended the decision to keep a remote controller in the product design during the roadshow. When investors questioned whether that made the robot look like a remote-controlled toy, he said the controller serves as the highest level of safety redundancy, providing a final physical safeguard if the AI model makes a wrong judgment. That is a cautious engineering choice. It also implies that today’s Unitree is still some distance away from robots making decisions fully on their own.
A model project that few participants can afford to see fail
Before listing, Wang Xingxing could still treat failure as the failure of a startup. After listing, Unitree is no longer carrying only its own story.
It now stands at once as the first humanoid robot stock on the STAR Market, a benchmark exit case for venture capital and a target for strategic capital allocation. Its market performance will directly affect the valuation room available to a group of robot makers that may list later.
That explains the unusual collective mood around the IPO. Few important participants want Unitree to become a failed case. Wang does not. Early investors certainly do not. Strategic investors with three-year lock-ups do not. Even an AI company like DeepSeek has already placed chips on the company.
If Unitree performs steadily, industry capital and supply-chain resources may flow more easily into the sector. If earnings fail to meet expectations, the sector’s valuation logic could be forced into a reset. In that sense, the IPO has raised the sunk cost of the embodied intelligence supply chain as a whole.
After the celebration, the market will turn back to earnings
Broad support for the company does not mean the stock cannot correct. If earnings growth slows or industrial deployment falls short of expectations, a pullback under such a high valuation would be hard to avoid. The rise and retreat in humanoid robot concept stocks around the Lunar New Year period this year already offered a preview.
There are external variables as well. The article notes that the U.S. Federal Communications Commission recently put advanced Chinese robots on a restriction list, while more than 43% of Unitree’s revenue depends on overseas markets. Capital can award a robotics company a rich valuation, but it cannot solve orders, costs, technology gaps or geopolitics on the company’s behalf.
First-day trading will show short-term positioning. What decides whether Unitree can support a RMB 61 billion valuation is whether future earnings reports can clear three commercial hurdles.
The first is whether the business-line shift continues. More than half of revenue now comes from humanoid robots, but that is only a starting point. The share has to keep rising if Unitree is to prove it has truly transformed from a quadruped hardware company into a humanoid robot platform company.
The second is customer expansion. Unitree needs to extend sales from research institutions and showroom-display use cases into factory production lines, warehousing logistics and hazardous inspection, moving from opportunistic purchases to a point where industrial customers actually depend on the machines.
The third is the margin gap under scale. Falling industry prices are not the real threat by themselves. The real question is whether cost reductions can outpace price cuts. Unitree still has to prove that supply-chain localization and scale effects can drive costs down fast enough to keep up with competitive pricing.
If those three tests are passed, RMB 61 billion may turn out to be only the opening valuation of an industry run. If they are not, the money rushing for allocation today may have bought little more than a very expensive ticket to a discounted future.
After Aug. 10, humanoid robots in China have, for the first time, a price that the market can recalculate every day. The industry narrative now gives way to something more concrete. From here, earnings reports do the talking.