UBTech Robotics Deep-Dive Research

Company Development History

The starting point was Shenzhen hardware entrepreneurship, not humanoid robots

UBTech was founded in March 2012, with both registration and headquarters in Shenzhen. The company began with small consumer and education robots rather than industrial humanoid robots, then gradually expanded into service robots, logistics robots, and now industrial humanoid robots. Public materials from the prospectus stage laid out this path clearly: consumer robots and other smart hardware launched in 2016, education and general service robots in 2017, smart logistics in 2020, and smart elderly-care services in 2022. In other words, although humanoid robots are the most eye-catching story in capital markets, they were not the company’s original revenue base.

This path has a strong Shenzhen character. What Shenzhen gave UBTech was the density of consumer-electronics and manufacturing supply chains, not an academic-lab environment. In its early years, UBTech secured shareholders that could provide capital, industrial resources, and channels: Qiming Venture Partners was the sole investor in its Series A round in 2015; Tencent led an USD 820 million Series C round in 2018; when the company submitted its Hong Kong IPO prospectus in 2023, Tencent, Qiming, and BYD co-founder Xia Zuoquan were still among the major shareholders. This shareholder structure matters because it means UBTech has always been a hard-tech startup that must keep proving it can push technology forward while also collecting at least some money back, rather than the type of company that can slowly do research and wait for commercialization after the technology matures.

Before listing, it tried several times to explain what it was

UBTech did not move in a straight line to the Hong Kong market. Public reports show that the company went through A-share tutoring and preparation, submitting tutoring progress updates to Shenzhen regulators several times between 2019 and 2021. In January 2021, it once terminated the process due to an adjustment in development strategy, then restarted tutoring in February of the same year. It ultimately did not list on the A-share market, and turned to Hong Kong in 2023. This process itself shows that the company long faced an identity problem: should the market understand it as an education robot vendor, an AI hardware company, or a humanoid robot platform? In the old scenarios, growth was not exciting enough. In the new humanoid story, the financial statements were not yet mature enough.

On 2023-12-29, the company listed on the Main Board of the Hong Kong Stock Exchange at an IPO price of HKD 90, issuing 11,282,000 H shares and raising net proceeds of about HKD 906 million. After partial exercise of the greenshoe in January 2024, the company received additional net proceeds of about HKD 25.30 million. The story the market gave it then was centered on being China’s first humanoid robot stock. The problem was that, at listing, revenue was still truly supported by older businesses such as education, logistics, and consumer hardware, not Walker. On the first day, the share price was almost flat, closing at HKD 90.85, with a market cap of about HKD 37.96 billion. This was a typical capital-market move: grant the company a track identity first, then wait for later financial statements to substantiate that identity.

The company’s history can be divided into four phases

The first phase was the “product validation period” from 2012 to 2017. UBTech’s core task at this stage was survival and proving that robots could be sold, not scale. It chose consumer and education scenarios because their stability requirements were relatively manageable and their commercial loops were shorter, not because it viewed them as the endgame. Looking back today, this history left UBTech with two genuinely useful assets: whole-machine experience and brand recognition formed on the education side. What it did not leave behind was high-quality profit. From the beginning, these businesses looked more like product lines and were unlikely to naturally grow into high-return platforms.

The second phase was the “scenario expansion period” from 2018 to 2023. The company began moving from single robot products to industry solutions, entering logistics, smart retail, elderly care, inspection, and other fields. In the 2023 financial statements, logistics already accounted for 36.9% of revenue, above education at 32.9%, showing that the growth center before listing had shifted from “selling one robot” to “providing factories and parks with an automation solution.” The side effects were also obvious: the business became heavier, and receivables, project acceptance cadence, and capex all began to rise. Revenue did not collapse, but it did not become high-quality cash flow either.

The third phase was “redefining itself” in 2024. This was the year UBTech truly pushed the Walker series into the center of the industrial narrative. In the 2024 annual report, management repeatedly emphasized industrial scenarios, saying Walker S and S1 had entered training in several well-known automotive factories, with tasks including handling, quality inspection, process material operation, parts installation, and SPS sorting. The company also released the BrainNet swarm-intelligence network architecture that year and served as deputy head of the national humanoid robot standards working group. Financially, 2024 revenue was RMB 1.305 billion, up 23.6%, and loss was RMB 1.160 billion, slightly narrower than in 2023. The real key was the structural change: the company kept increasing investment in industrial humanoid robots, preparing to change its profit logic from “a mix of many small business lines” to “rewriting the revenue structure with one large story.”

The fourth phase is the “mass-production validation period” from 2025 to now. In 2025, UBTech made humanoid robots the protagonist in revenue for the first time: full-size embodied-intelligence humanoid robot revenue reached RMB 821 million, accounting for 41.1%, with sales of 1,079 units. Management also wrote that “small-scale mass production and delivery at the 1,000-unit level marks the maturity and stability of the manufacturing process.” In early 2026, UBTech extended scenarios from automotive and electronics to aerospace manufacturing by signing an agreement with Airbus. The commercial logic of this phase is clear: first run efficiency, stability, and the data loop in industrial scenarios; then use factory data to feed the model; then feed that back into new scenarios. The market is willing to give it a higher valuation now because it is betting that this phase is a reusable curve, not a short-term volume push.

Several nodes that changed the company’s fate

The first node was the Hong Kong listing in 2023. Its most direct financial effect was opening up financing channels, not improving profitability. IPO net proceeds were about HKD 906 million, but what truly changed the capital structure was the continuous equity financing after listing. UBTech’s 2025 annual report disclosed that the company conducted three private placements of new H shares in February, July, and December 2025, adding 71,778,549 shares in total. Share capital expanded from 431.6 million shares at the end of 2024 to 503.4 million shares, with dilution of more than 20%. Without these financing rounds, the company’s cash table today would not look this good. The market once underestimated this node: UBTech obtained funding and a more stable stock identity as a loss-making company with high optionality.

The second node was industrial-scenario training and acceptance in 2024 and 2025. The sharp change in the 2025 revenue mix came from industrial humanoid robots entering real scenarios and beginning to recognize revenue, not from a recovery in education robots. Management was blunt in the 2025 annual report: revenue from other intelligent robot products and services declined partly because some logistics projects had not yet been delivered and accepted by year-end. This shows that UBTech’s revenue recognition is highly dependent on project delivery and acceptance, rather than pure subscription or one-off stocking. Put differently, the most important lever in its financial statements is “when it can pass acceptance and become revenue,” not “how many contracts it signed.” This is especially important for investors.

The third node was the Airbus cooperation in January 2026. It does not mean UBTech has conquered aerospace manufacturing. Airbus itself emphasized that the cooperation remains at an early concept-testing stage. But the significance of this node is that it pushed UBTech one step from “pilot supplier for domestic automotive and 3C factories” toward “global high-end manufacturing validation.” Capital markets usually react strongly to this kind of news because it tells a harder-to-disprove story: if even Airbus is willing to test it, the market will infer that the capability may have a chance to migrate across industries. The risk is also here: testing is not a long-term contract, let alone scaled revenue.

Financial and Share-Price History

Financial review

UBTech’s financial trajectory over the past four years looks like a curve extending upward to the right while still pressing hard on cash flow. Revenue was RMB 1.008 billion in 2022, RMB 1.056 billion in 2023, RMB 1.305 billion in 2024, and RMB 2.001 billion in 2025. Revenue is growing, but profit remains deeply negative: losses were RMB 987 million in 2022, RMB 1.265 billion in 2023, RMB 1.160 billion in 2024, and RMB 790 million in 2025. Losses are narrowing, but that narrowing came from an improved revenue mix, lower selling and administrative expense ratios, and a significant reduction in share-based compensation, not because the company suddenly gained mature operating leverage. In 2025, share-based payment expenses fell from RMB 245 million in 2024 to RMB 99 million, which had a material effect on reported improvement.

The change in gross margin explains more than net profit. Gross profit was RMB 333 million in 2023, RMB 374 million in 2024, and RMB 754 million in 2025. Gross margin rose from 28.7% in 2024 to 37.7% in 2025. Management explicitly attributed the improvement to high-margin full-size embodied-intelligence humanoid robots becoming the largest revenue source. This suggests that UBTech’s humanoid-robot business is at least not weak at the accounting gross-margin level, and may even be better than its traditional robotics solution businesses. The problem is that gross-margin improvement has not yet flowed through to free cash flow. In 2025, operating cash flow was still a net outflow of RMB 784 million, capex was RMB 614 million, and total cash consumption for the year was about RMB 1.398 billion. It is improving, but it is still far from a cash generator.

Receivables and credit impairment deserve closer monitoring. Net receivables were RMB 913 million at the end of 2024 and rose to about RMB 1.302 billion at the end of 2025. Management also wrote clearly in the 2025 annual report that credit impairment losses were about RMB 151 million in 2025, mainly because some government-related customers delayed payments, and the company made provisions on a prudent basis. This detail matters. Many of UBTech’s most imaginative businesses today occur in scenarios with hot industrial policy and many demonstration applications. These scenarios often bring two problems: orders are easy to sign, and payments may not be fast. For hard-tech companies, the biggest concern is not losses themselves, but “profit and revenue look like they are growing while cash is stuck in receivables.”

The balance sheet looked much more stable in 2025, but mainly as a result of financing. At the end of 2025, cash and cash equivalents were about RMB 4.888 billion, external bank borrowings were about RMB 1.123 billion, gearing under the company’s debt-ratio measure fell from 71.6% in 2024 to 16.0%, and the current ratio rose from 1.4x to 3.0x. Book safety did improve, but it came from several share placements rather than operating collections. Based on the combined consumption of 2025 operating cash outflow plus capex, current cash can roughly cover about 3.5 years of the same burn rate. This is still a static estimate that excludes large acquisitions, investments, and unexpected price wars. Of the net funds raised in the November 2025 placement, the company explicitly earmarked 75% for investing in or acquiring upstream and downstream targets, industrial integration, or establishing joint ventures. That means the cash on hand may not simply sit on the balance sheet and wait for the business to mature.

On governance, UBTech has not exposed red flags for financial fraud or major litigation in reviewed documents, but that does not mean there is no governance discount. The 2025 annual report clearly disclosed that Chairman and CEO are both held by Zhou Jian, a deviation from the role separation recommended by Hong Kong listing rules. The board explained this as necessary for management continuity and strategic execution. For a company making a sharp turn from old businesses to new humanoid businesses, this centralized setup can improve execution efficiency, but it also raises the cost of correction if capital-allocation judgment goes wrong. The more practical governance discount comes from dilution: after three placements in 2025, share capital rose by more than 20% year on year, while the market has not yet seen an operating-cash-flow inflection point that proves the company no longer needs frequent financing.

Share-price and valuation history

UBTech’s share-price history is short, so doing a strict “ten-year historical valuation percentile” is not meaningful. A more useful approach is to place it into several stages after listing. The first stage was from listing to mid-2024, when the market gave it an identity premium as the first humanoid robot stock, but still lacked confidence in financial-statement quality, so the share price did not immediately enter a one-way trend. The second stage was 2025, when Walker S2 mass production, industrial-scenario validation, and warmer policy sentiment led the market to re-rate it as one of the core embodied-intelligence names. The third stage is from early 2026 to now, when the stock has been stimulated by news such as Airbus cooperation while being pulled back to reality by dilution, intensifying competition, and efficiency questions. By 2026-06-12, the share price was HKD 106.8, down about 33.7% from the 52-week high of HKD 161, but still 45.3% above the 52-week low of HKD 73.5.

Looking only at the current valuation, UBTech is no longer cheap. Based on 2025 revenue of RMB 2.001 billion, year-end net cash, and the share price as of 2026-06-12, the company currently trades at roughly 23.4x sales and 21.5x EV/Sales. For a hard-tech company that remains deeply loss-making, has negative cash flow, and has a business model just moving from pilots to small-batch production, this valuation is not supported by current profit. It can only be supported jointly by revenue continuing to double over the next few years, further gross-margin improvement, and the capital market’s willingness to keep granting high multiples. The market’s label for UBTech has shifted from “loss-making robot company” to “industrial humanoid robot option asset.” If business growth falls to a level that cannot support that label, the valuation center will move down quickly.

Business Model and Moat

Where the revenue actually comes from

Before 2025, UBTech’s revenue mix was still relatively scattered. After 2025, the main engine had changed. By product and service category, 2025 full-size embodied-intelligence humanoid robot revenue was RMB 821 million, or 41.1% of revenue; other intelligent robot products and services were RMB 629 million, or 31.4%; other smart hardware equipment was RMB 499 million, or 24.9%; non-embodied-intelligence humanoid robot revenue was RMB 48 million, or 2.4%. Breaking this down further by application vertical, education revenue was RMB 413 million, logistics RMB 274 million, other industry customized solutions RMB 787 million, and consumer robots and other hardware RMB 523 million. The most important change is that most full-size embodied-intelligence humanoid robots have already penetrated the education, logistics, and industry customization verticals rather than being isolated in a “separate business segment,” with industry customization contributing the most. UBTech is not simply selling one robot. It is closer to selling “robot body + task definition + scenario solution + acceptance delivery.”

This also explains why it keeps emphasizing industrial scenarios rather than home scenarios. Industrial customers are willing to pay for three things: replacing highly repetitive posts, flexible renovation in complex production lines, and early access to the data and process advantages of future automation. In the 2025 annual report, UBTech described its core workstations very concretely: handling, sorting, and quality inspection. It has indeed entered “real scenarios,” but this type of revenue naturally has project characteristics. Revenue recognition depends on delivery and acceptance, and the payment cycle may not be short. For investors, the key in this model is repeat order rate, degree of standardization, and whether unit delivery cost can fall with scale, not the number of contracts signed.

Cost structure and operating leverage

UBTech’s current cost structure has clear “early-commercialization hard tech” characteristics. Fixed costs are heavy, with R&D and organization weighing more than manufacturing. R&D expenses were RMB 508 million in 2025, or 25.4% of revenue. In 2024, the ratio was even higher at 36.6%. To maintain competitiveness in industrial humanoid robots, the company must keep investing in models, motion control, dexterous hands, integrated joints, and data loops. These investments are hard to cut in the short term. At the same time, as the revenue share of full-size embodied-intelligence humanoid robots has risen rapidly, overall gross margin has already improved markedly, showing that the company does have operating leverage. That leverage has simply not yet broken through the expense wall. In other words, continued revenue scale-up is a necessary condition, but not a sufficient one. Operating leverage will truly appear only if standardization improves, scenario development costs are amortized, and subsequent price declines do not outpace declines in BOM and manufacturing costs.

The moat that is actually forming

The thing that most resembles UBTech’s moat is the data loop in real industrial scenarios, not the brand. The company disclosed that, based on real production lines, it has accumulated more than 100 million pieces of industrial-manufacturing scenario data to drive robots from single-machine intelligence toward group collaboration. For today’s humanoid robot industry, this matters more than how many patents are filed or how many demo videos are released. The hardest part of industrial humanoid robots has never been “walking”; it is continuously and stably doing work in constrained real environments. Whoever obtains real task data earlier is more likely to move models from demo to usable.

The second moat is full-stack integration capability. UBTech does not only make a humanoid robot body. The value accumulated from its earlier logistics and industry-solution businesses is only beginning to show today: the company can put robots, unmanned logistics vehicles, cloud platforms, WMS, MES, and other systems into the same factory-renovation narrative. This matters for factory customers. Customers are buying an automation unit that can be integrated into existing manufacturing processes and pass acceptance, not a walking machine. Compared with new players that only sell the “body,” UBTech looks more like a company that knows how to embed robots into production lines.

The third moat is first-mover customer validation. Names such as BYD, Foxconn, Geely, FAW-Volkswagen, and Dongfeng Liuzhou Motor matter because they correspond to China’s most complex, strictest-takt, and most automation-friendly industrial scenarios, not because the logos look good. Even if many of these collaborations remain in testing or phased introduction, they are still among the scarcest validation fields in the industry. The 2026 cooperation with Airbus pushes this validation from domestic automotive and electronics further outward into aerospace manufacturing.

But “real moat” must be separated from “promotional moat.” UBTech does not yet have network effects, nor does it have high switching costs like an operating system. Being a participant in national standards is a plus, not a settlement barrier. The consumer-robot brand provides limited help in industrial-customer procurement decisions. At present, it is closer to a combination of “first-mover advantage + scenario data + integration experience” than a structural monopoly proven over 10 years. The moat exists only if it can keep deepening these scenarios, rather than allowing competitors to erase the gap with lower prices, more testing volume, and faster iteration.

Management and governance

Zhou Jian remains the company’s key person. He is both founder and Chairman and CEO. For a company like UBTech, the benefit of this arrangement is fast strategic progress, while the drawback is weaker constraints on capital allocation. A major foundation for the 2025 financial improvement was the better cash balance after continuous financing. At the same time, most of the net proceeds from the November placement were reserved for upstream and downstream investments and acquisitions, showing that management is not satisfied with “first getting the existing business running smoothly” and wants to keep expanding the industrial-chain footprint. For a growth hard-tech company, this is not necessarily wrong. But for public-market shareholders, it means two things: continued financing and continued expansion are both likely to happen, and return realization may depend more on the capital-market environment than the most optimistic profit model suggests.

Industry and Horizontal Peer Analysis

A hot industry without a mature profit pool yet

The humanoid robot industry in China and globally is still in the introduction phase, not the mature phase. China’s Ministry of Industry and Information Technology issued guidance in 2023 calling for a preliminary humanoid robot innovation system by 2025. The 2025 Government Work Report also included “embodied intelligence” in future industries for the first time. Commercial volume is indeed starting to appear. Counterpoint Research estimates that global new humanoid robot installations were about 16,000 units in 2025, with China accounting for more than 80%. Goldman Sachs forecasts a global humanoid robot TAM of USD 38 billion in 2035, corresponding to about 1.4 million units shipped. The issue is that rising installations do not mean the profit pool is already stable. The most certain things in the industry today are still narrative, financing, and policy resources, not whole-machine profit.

Industry growth mainly comes from four things: manufacturing labor costs and hiring difficulty, policy support for “AI+” entering factories, localization of key components reducing BOM, and large models pushing robots from preprogramming toward stronger generalization. But the industry’s fragile points are just as clear: applications remain narrow, prices remain high, efficiency remains insufficient, and customer procurement still contains a fair amount of “early trial” and “data-asset reserve.” At the end of 2025, the National Development and Reform Commission publicly reminded humanoid robot companies to avoid highly repetitive products entering the market and squeezing R&D space. In investment language, this means the industry is already getting crowded, and homogenization, price wars, and ineffective capacity expansion cannot be ruled out.

UBTech faces four types of competition, not one

The first type is Tesla Optimus. Its greatest threat is not how many units it has sold today, but its own gigafactories, automotive supply chain, algorithms, and capital-market narrative. Reuters reported in January 2026 that Musk said meaningful Optimus volume production would only come after the end of 2026 and would start “excruciatingly slowly.” It was also affected in 2025 by China’s export controls on rare-earth magnetic materials. In other words, Tesla has not yet moved ahead of UBTech in external commercialization. But once it runs the internal factory scenario through, its cost and manufacturing ramp speed may far exceed most startups. Against Tesla, UBTech’s advantage is earlier landing; its disadvantages are manufacturing scale and capital depth.

The second type is Figure. Figure’s strongest asset is what the U.S. capital market is willing to imagine for it, not disclosed revenue. It began cooperating with BMW in 2024, and its latest 2025 financing valued it at USD 39 billion. Figure’s niche is more like a “U.S. AI robotics flagship project,” positioned by software narrative, top-tier capital, and validation from North American manufacturing customers. It does not directly compete with UBTech for public-market valuation, but it lifts global investors’ valuation imagination for leading humanoid robot companies and, in turn, raises market demands on UBTech’s efficiency and speed.

The third type is emerging Chinese domestic players, especially Unitree and AgiBot. Data from Reuters and Counterpoint show that Unitree and AgiBot each shipped more than 5,000 units in 2025, leading globally. Unitree is also preparing for an A-share IPO at a valuation of about USD 6.2 billion. This threat is more realistic than Tesla because these companies are closer to China’s domestic supply chain and are more likely to compete on price, capacity, and exhibition visibility. Unitree in particular has already taught the market a lesson: even if sales rise, profit can come under pressure from both R&D and price cuts. Compared with these companies, UBTech’s advantages are industrial integration and listed-company financing ability; its weakness is that cost pressure may not be any lighter.

The fourth type is adjacent listed automation names, such as Dobot. Dobot is essentially a collaborative-robot company, not a pure comparable for full-size industrial humanoid robots. But it is one of the few public Hong Kong-listed names investors can use as a “robotics company market-cap anchor.” By June 2026, Dobot’s market cap was roughly in the HKD 12.2–12.5 billion range, significantly below UBTech. This comparison shows that the market is giving UBTech an option valuation for “early positioning in humanoid robot commercialization,” rather than the valuation of an ordinary robot equipment maker. This premium is reasonable, but fragile. Once it loses the identity of “early deployer,” its valuation will quickly move closer to traditional automation.

UBTech’s niche

Putting these competitors together, UBTech’s niche today is clear: it most resembles a challenger that has first produced financial-statement validation in industrial scenarios. Its technology is not the flashiest, shipments are not the highest, and capital is not the deepest. Why would customers choose it? Because it can package “robot body + scenario landing + production-line collaboration + data loop,” and has already built cases in some of China’s most real manufacturing test fields that can be demonstrated, accepted, and recognized as revenue, rather than because it is the cheapest. Why might customers also leave it? Because if others can lower prices, deploy faster, and simplify maintenance at the same workstation, industrial-customer loyalty will not be very high. UBTech is competing for the “human-robot collaboration budget of future factories.” The companies most likely to take its future profit pool are domestic peers that can better combine general platforms, low-cost manufacturing, and industrial-customer validation, not traditional education robot companies.

Current Fundamentals and Valuation Analysis

What actually happened in the most recent four quarters

Hong Kong-listed companies do not provide complete quarterly operating disclosure like U.S.-listed companies, so the most reliable “most recent four quarters” cut is the full-year picture pieced together from the 2025 interim report and the 2025 annual report. In the first half of 2025, revenue was RMB 621 million, up 27.6% year on year; gross margin was 35.0%, down 3 percentage points from the same period last year; and loss for the period was RMB 440 million, narrowing 18.5% year on year. For full-year 2025, revenue growth reached 53.3%, gross margin instead rose to 37.7%, and loss narrowed to RMB 790 million. This shows that the real revenue and profit improvement occurred more in the second half and was mainly driven by volume growth in full-size embodied-intelligence humanoid robots. The interim report also revealed an important point the market has underappreciated: in 2024, two customers each contributed more than 10% of revenue and together accounted for 33% of full-year revenue, while in the first half of 2025 no single customer exceeded 10%. This at least suggests progress in expanding new customers and diversifying concentration.

But “real fundamentals” and “market narrative” are not the same thing. The real fundamentals are that UBTech’s humanoid-robot business has crossed from zero to one, the revenue and gross-profit structure is improving, and cash has improved significantly because of financing. The market narrative is more aggressive: it is treated as one of the Chinese embodied-intelligence companies that has realized revenue earliest, is easiest to replicate into more factories, and is most likely to achieve large-scale deployment first. The most overheated part of this narrative is extrapolating “beginning to be usable” too quickly into “soon to be widely adopted.” The Financial Times cited management in early 2026 as saying that Walker S2 still only achieves 30%–50% of human efficiency on some tasks, and the company hopes to lift that to 80% by 2027. This means customer purchases today are not entirely based on today’s ROI; they are more like buying an option on a future efficiency curve.

Bull-bear divide

The strongest evidence for bulls is the jump in humanoid-robot revenue and sales in 2025. RMB 821 million in revenue, 1,079 units sold, small-batch delivery at the 1,000-unit level, and more than 100 million pieces of industrial-scenario data are all items that have landed in financial statements and operating narratives, not PPT metrics. Together with Airbus testing, Texas Instruments cooperation, and introduction by leading domestic factories, UBTech is moving from a company with the “right concept” into one that is beginning to be selected by real customers. If this curve keeps rising, the market can justify giving it a high EV/Sales multiple.

The strongest evidence for bears comes from conversion rate and cash quality. The company’s humanoid robot orders exceeded RMB 1.4 billion in 2025, but full-size embodied-intelligence humanoid robot revenue recognized that year was RMB 821 million. Roughly putting the two together, only about 60% of order value turned into related revenue within the year. The two measures may not be perfectly comparable, but this is enough to remind investors: orders are not revenue, contracts are not collections, and revenue does not necessarily equal cash. On top of that, the company still had credit impairment losses of RMB 151 million in 2025, explicitly caused by delayed payments from some government-related customers. If this link gets stuck, the valuation is most vulnerable.

The second bear argument is financing habits. The main reason UBTech changed from “not enough money” to “a lot of money on the balance sheet” in 2025 was three share placements during the year, not operating cash improvement. It issued shares at HKD 90 in February, HKD 82 in July, and HKD 98.8 in December. For a company that is still expanding capacity, still investing in R&D, and preparing upstream and downstream integration, this financing is not surprising. But for public-market shareholders, it means equity is not a stable denominator. If the market is no longer willing to provide new capital at high prices one day, UBTech’s valuation model has to be rewritten immediately.

Historical valuation and peer valuation

UBTech’s listing history is short, so instead of discussing a “ten-year percentile,” it is better to discuss today’s odds directly. Based on the 2025 annual report, the company currently trades at about 23.4x sales and 21.5x EV/Sales. This valuation is meaningfully above traditional automation equipment companies and above ranges that many mature robotics companies can maintain over the long term. The market is willing to pay this price because it no longer puts UBTech in the same basket as education robots and service robots; it classifies it with Tesla Optimus, Figure, Unitree, and AgiBot as a “main humanoid robot narrative company.” The problem is that few companies in this group can publicly validate profit and cash flow, so the industry valuation anchor itself is unstable.

In relative terms, UBTech sits in a middle zone: clearly more expensive than adjacent listed automation names such as Dobot, and far below U.S. super-financing examples such as Figure. Compared with Unitree’s proposed IPO valuation, UBTech is not cheap, while Unitree’s 2025 shipment volume was larger. What matters most for UBTech is how long the market is willing to let it maintain a high multiple, not the fact that “peers are also expensive.” If the industry shifts over the next two years from “competing on imagination” to “competing on efficiency and unit economics,” its valuation premium will be reassessed.

Cash-flow look-through

Across the four full years from 2022 to 2025, UBTech’s operating cash flow and net profit were both consistently negative. The ratio of operating cash outflow to the absolute value of net loss was roughly between 0.55x and 0.99x, with a four-year average of about 0.76x. It is not a typical case of “pretty accounting profit but no cash collection.” It is the opposite: the books already show heavy losses, and cash still flows out. In 2025, this relationship improved slightly, with operating cash outflow of RMB 784 million, close to the RMB 790 million net loss for the period, but still not positive.

On capex, spending was RMB 541 million in 2023, RMB 400 million in 2024, and rose again to RMB 614 million in 2025, mainly going to the Shenzhen headquarters, Hangzhou Lin’an project, Jiujiang industrial park, and Wuxi UQI PARK. For UBTech, most capex is not an easily cut “nice-to-have expansion item,” because the company itself is still in a phase of capacity ramp and infrastructure construction. From an owner-earnings perspective, this company does not yet have a meaningful PE or FCF Yield. If one insists on calculating it, the result is simply a persistently negative free-cash-flow yield. For this type of company, absolute valuation is better done using scenario-based EV/Sales than profit multiples.

Absolute valuation

I use a 2027 EV/Sales scenario method rather than PE for three reasons. First, the company is still unprofitable, and profit volatility is heavily affected by share-based compensation and accounting items. Second, what the market is really trading is whether the company can move around 2027 from 1,000-unit-level small batches to larger-scale standardized industrial deployment. Third, current EV/Sales is already high, so any valuation judgment must place revenue delivery and multiple contraction on the table at the same time. The following is only valuation scenario analysis under a research framework and does not constitute investment advice. Based on the 2025 annual report, current share price, current net cash, the 2026 capacity target, and my assumptions for 2027 revenue, cash consumption, and multiples, I get three levels of implied 2027 value: conservative scenario HKD 58–66, base scenario HKD 106–117, and optimistic scenario HKD 178–194. Translating this into operating price levels, the ideal buy zone only falls to HKD 47–53 after applying about a 20% margin of safety to conservative intrinsic value.

Note†: The above valuation uses 2027 revenue and EV/Sales scenarios, translated at the 2026-06-12 exchange rate, and implicitly assumes moderate future dilution in total share capital. It lays out directly “what needs to be delivered for the share price to be reasonable,” rather than promising a target price. The calculation basis is the revenue, cash, borrowings, and placement information disclosed in the 2025 annual report, plus this report’s explicit assumptions for 2027 revenue and multiples.

Expectation gap and margin-of-safety review

Today’s price roughly means the market has already accepted a prior judgment: UBTech can grow revenue to the RMB 4.3–4.8 billion range around 2027 while continuing to enjoy a high valuation near 10x EV/Sales. If the future valuation center falls to 8x EV/Sales, the revenue required by the current EV rises to about RMB 5.37 billion. If the industry is only willing to grant 6x, about RMB 7.16 billion of revenue would be needed to justify today’s market cap. In other words, the market is pricing an ideal state two to three years out, not today’s company.

The margin-of-safety conclusion is clear. The current price of HKD 106.8 represents a premium of about 64%–84% to the conservative implied value of HKD 58–66 under the 7.3 scenario, not a discount. If the most fragile assumption, that industrial customers will steadily reorder and revenue can cross RMB 5.0 billion, is cut by 30%, the base-scenario value would roughly fall to around HKD 75–85. If revenue basically stays near the 2025 scale over the next three years and valuation multiples fall back to ordinary high-growth automation levels, the annualized return from buying today would be meaningfully negative and far below long-term government bond yields. My conclusion on margin-of-safety sufficiency is: none. This does not mean the company has no value. It means that at the current price, investors are buying high elasticity, not cheapness.

Risks, Catalysts, and Cross-Framework Summary

Risk analysis

The first risk that could truly cause permanent capital loss is repeat purchase in industrial scenarios failing to materialize. I assign medium probability and high impact. UBTech’s strongest story today comes from automotive, 3C, semiconductor, and aerospace manufacturing scenarios, but the Airbus cooperation remains at a concept-testing stage, and Walker S2 is still only 30%–50% as efficient as humans on some tasks. If customers are buying a “learning opportunity” rather than stable ROI, the first order can happen, while the second and third orders may not. Observable indicators include disclosed growth in full-size embodied-intelligence robot revenue, expansion by the same customer across workstations, the pace of single-unit ASP decline, and whether acceptance cycles lengthen. Once repeat purchases fail to materialize, the transmission path will be slower revenue growth first, pressure on gross margin next, and then rapid valuation-multiple compression.

The second risk is a price war arriving before economies of scale. Probability is medium-high and impact is high. China’s domestic track is already very crowded, and the NDRC has explicitly warned against highly repetitive products entering the market. Unitree’s profit decline in Q1 2026 also shows that sales growth does not automatically equal profit improvement. UBTech’s 2025 gross-margin improvement partly came from the revenue mix shifting toward higher-margin humanoid robots. If competitors start using lower prices to compete for the same workstations, UBTech must either cut prices to keep orders or lose speed. Observable indicators are whether overall gross margin falls back below 32%, whether full-size humanoid robot revenue growth is meaningfully slower than sales-volume growth, and whether R&D and selling expense ratios rise again.

The third risk is abundant cash masking dilution of shareholder returns. Probability is high and impact is high. UBTech had abundant cash at the end of 2025, but it mainly came from multiple placements. Meanwhile, most of the funds from the November placement were designated for upstream and downstream investments or acquisitions rather than simply replenishing liquidity. For ordinary shareholders, the worst-case scenario is that the company keeps financing during a high-narrative window and revenue also grows, but per-share value is continuously diluted rather than the balance sheet simply lacking cash. Observable indicators include the pace of share-capital expansion, placement discounts to market price, acquisition execution cadence, and whether investment projects form visible synergies with the core business within two years.

The fourth risk is receivables, impairment, and delayed payments from government-related customers slowing breakeven. Probability is medium and impact is medium-high. Credit impairment losses were RMB 151 million in 2025, and management explicitly attributed them to delayed payments from some government-related customers. Industries driven by policy and demonstration projects can scale volume easily, but the weakness of demonstration projects is also payment collection. Observable indicators are net receivables as a percentage of revenue, credit impairment losses as a percentage of revenue, and the gap between contract liabilities and receivables. If receivables continue growing faster than revenue, the market will begin to view UBTech as a project-based company with good accounting growth but average cash quality, rather than a platform-like new species.

Catalysts and tracking indicators

Positive catalysts do not need to be numerous; two are enough if they are hard. First, industrial customers move from pilot to repeat purchase, especially leading automotive, electronics, and aerospace manufacturing customers expanding validation from a single workstation to multiple workstations and multiple production lines. Second, the efficiency curve delivers. If the company can move Walker S2 from 30%–50% of human efficiency toward the 80% target over the next one to two years, the valuation premium will be easier to maintain. Airbus cooperation, more global manufacturing customers, and consumer-grade new products that can truly turn into deliverable revenue can also strengthen the narrative.

Negative catalysts are more common. The most direct are another round of financing, price war, and guidance misses. Next is revenue growth remaining high while gross margin falls, indicating that growth is being bought by price. Then there is policy emphasis on reducing homogenization and promoting rational investment, which could shift market sentiment from chasing the humanoid robot concept to questioning the profit model. For a high-valuation, short-history, strong-narrative stock, bad news often appears as “financing + slowing growth + intensifying competition” together, rather than in isolation.

Note‡: The “current/latest value” in the table comes from the 2025 annual report, 2025 interim report, and FT reporting. The “normal range” and “warning threshold” are tracking disciplines set by this report, not company guidance.

The three most important indicators are whether humanoid-robot revenue share keeps rising, whether gross margin holds, and whether cash flow stops relying entirely on financing, rather than the share price itself. UBTech is actually easy to understand. The hard part is judging whether it has moved from “first-mover validation” into “replicable deployment.” If all three indicators improve together, the share price has room to tell an upward story again. If two of them deteriorate at the same time, the market will quickly mark it down from “future platform” to “expensive equipment vendor.”

Bull and bear cases

I compress the bull case into five points. First, full-size embodied-intelligence humanoid robot revenue share rose to 41.1% in 2025, moving from concept to the main engine in the financial statements. Second, 1,079 units sold and 1,000-unit-level small-batch production prove that manufacturing and delivery are no longer confined to the lab. Third, more than 100 million pieces of data from real industrial scenarios can form a stronger iteration barrier than exhibition buzz. Fourth, Airbus, Texas Instruments, and leading domestic manufacturing customers show that scenario validation is spilling over from domestic automotive factories. Fifth, the significant 2025 gross-margin improvement shows that growth is not all being bought with price.

The bear case must also be fully stated. First, Walker S2 currently reaches only 30%–50% of human efficiency in some tasks, so industrial replacement is still far from broad replication. Second, the company has had negative operating cash flow for many years, and the 2025 improvement in financial safety mainly depended on financing rather than self-generation. Third, share capital increased by more than 20% year on year in 2025, and placements have already become part of the business model rather than an accident. Fourth, receivables and credit impairment remain heavy, and delayed payment from government-related customers is already written into the annual report. Fifth, the industry is getting crowded, domestic competitors’ shipments and financing are both accelerating, and the risk of price war is not remote.

Pre-mortem

If this investment loses 50% in three years, I think the most likely scripts are two.

The first script is “efficiency failed to rise, and prices fell first.” By 2027, UBTech still cannot meaningfully pull key workstation efficiency close to its publicly stated 80% target, while the industry has already begun cutting prices because of Unitree, AgiBot, more automaker self-development, and automation players entering the field. To defend customers, UBTech can only lower prices on multi-workstation projects. Gross margin falls from 37%–38% back to 28%–30%, while the market compresses the company from around 10x EV/Sales to 6–7x. Even if revenue is still growing then, the share price could easily halve from today’s level.

The second script is “revenue grows, but per-share value does not.” The company keeps winning orders and keeps expanding capacity and investing upstream and downstream, but operating cash flow does not turn positive. In 2026-2027, it conducts another equity financing round or a large share-swap acquisition, further diluting share capital. The market finally realizes that UBTech is a heavy-investment project requiring continuous capital-market support, not a high-quality growth stock that naturally rolls into free cash flow, and lowers its valuation center. For shareholders, this is the most uncomfortable script, because the company may appear to be “improving” on the surface while per-share value is leaking away.

Final research conclusion

What UBTech has truly proven is that it has finally sent this thing into factories and into financial statements, and in 2025 it visibly rewrote the revenue structure. It did more than merely build a walking humanoid robot. This is significant progress. Over the past few years, the market’s biggest doubt about UBTech has been whether it would forever remain a company that can make robots but cannot make decent money. The 2025 annual report at least proves that the answer is no longer a simple “yes.” The company is indeed moving forward, and more solidly than many peers that only tell stories.

But UBTech has not yet proven another equally important thing: whether this path can grow into a business that sustainably creates cash without materially sacrificing shareholder interests. Financing played too large a role behind the 2025 financial improvement. Behind industrial-customer validation, efficiency still significantly lags human labor. Behind rapid order growth, the chain of revenue recognition, collections, and repeat purchase has not been fully proven. Put differently, the hardest part for this company has just begun, and is far from complete. Buying UBTech today means buying an option on 2027 revenue, gross margin, efficiency, and capital-market sentiment all at once, not an already mature industrial leader.

Therefore, my final judgment is not pessimistic, but I am also unwilling to get overly excited at the current price. It deserves continuous tracking, and even deserves a long-term place on the key watchlist, because if China’s industrial humanoid robots truly move from pilots to scale, UBTech is likely to be one of the first names validated. But near HKD 106.8, the market has already paid a lot for that possibility. A better investment point would require three things to appear together: revenue keeps scaling, cash flow keeps improving, and the price returns to a wider margin-of-safety range, rather than trying to guess the next order headline. Without all three, investment looks more like speculation.

【Company Profile Score】

• Fundamental quality: Medium

• Growth: High

• Moat: Medium

• Financial robustness: Medium

• Management credibility: Medium

• Valuation attractiveness: Low

• Risk level: High

• Suitable investor type: Long-term growth / high-risk speculation / unsuitable for ordinary investors

【Investment Rating】

• Rating: Hold

• One-sentence investment thesis: Industrial humanoid-robot commercialization has passed “zero to one,” but the current price has already priced in much of the 2027 expansion expectation.

• Three price signals: Ideal buy price: see the standalone line below

• Holdable price: 90–135 HKD

• Clearly overvalued price: 196–213 HKD

• Current price classification: Holdable

• Worth waiting for a better price: Yes. For new capital, the better trigger is the share price returning to 47–53 HKD, or the company improving operating cash flow and gross margin together for two consecutive reporting periods without further significant share-capital dilution.

• Target holding period: 1–3 years

• Expected annualized return: Conservative about -30%, base about +3%, optimistic about +45%

• Maximum loss risk: 50%–70%; triggers are industrial-scenario repeat purchase failing to materialize, price war breaking out early, and the valuation center moving down to the ordinary automation-equipment range.

• Signals triggering reassessment: Overall gross margin falls below 32% for two consecutive reporting periods

• Operating cash outflow expands for two consecutive reporting periods, with no slowdown in capex

• Net receivables continue growing faster than revenue

• Another large equity financing or high-ratio share-swap acquisition occurs within the next year

• By 2027, key workstation efficiency still has not moved meaningfully close to the company’s publicly stated 80% target

【Ideal/Fair Buy Price】47–53 HKD

Basis: This range equals the conservative intrinsic value of HKD 58–66 in this report after applying about a 20% margin of safety. It is a buying discipline that incorporates execution, collection, and dilution risks for an early-stage hard-tech company, not a “bearish price.”

【Valuation Range】

• current: 106.8 (as of the 2026-06-12 close)

• bear (conservative · ideal buy zone): [47, 53]

• base (reasonable · acceptable holding zone): [90, 135]

• bull (optimistic · above the clearly overvalued line): [196, 213]

Key Data Table

Note§: 2022-2025 financial data all come from the company’s annual results announcements. The 2025 cash improvement was mainly driven by several share placements in 2025.

Note: Private-company figures in the table are based on the latest publicly disclosed financing or IPO valuation measures. They are not fully equivalent to public-market prices and are used only as valuation coordinates.

Research Uncertainty

First, the company’s listing history is short and public valuation samples are limited, so “historical percentiles” can only be read roughly and cannot be as precise as for mature companies. Second, many competitors in the humanoid robot industry are still private companies, with incomplete disclosure on revenue, gross margin, collections, and unit economics, so horizontal comparison naturally has blind spots. Third, UBTech’s disclosure granularity on “orders,” “delivery,” “acceptance,” and “revenue recognition” is still not high enough, making it difficult for outside investors to fully reconstruct the quarterly path from orders to revenue. Fourth, the current share-price snapshot comes from the latest closing page visible over the 2026-06-14 weekend. This report treats it as the 2026-06-12 most recent trading-day closing price, as stated in the metadata.

Reference Sources

The core basis of this report is the company’s 2023, 2024, and 2025 results announcements disclosed on the Hong Kong Stock Exchange, the 2025 interim report, the June 2026 monthly return, several 2025 placement announcements, and 2026 news on Airbus cooperation; it is supplemented by public reporting from Reuters, FT, Counterpoint Research, Goldman Sachs Research, and others on the industry, peers, and policy. The most important primary sources remain company announcements and HKEX filings.

Other Securities Mentioned in This Report

• TSLA.US: Optimus is used as UBTech’s most important global industrial humanoid-robot reference.

• 02432.HK: Dobot is the closest public Hong Kong-listed robotics automation valuation anchor, but it is not a pure comparable for full-size humanoid robots.

• 09868.HK: XPeng is pushing IRON mass production, reflecting the potential threat of automakers extending into humanoid robots.

• 005380.KS: Hyundai Motor is the parent company of Boston Dynamics, representing the potential pressure from traditional industrial-robotics leaders on the humanoid path.

• AIR.PA: Airbus is one of UBTech’s most important overseas industrial-scenario validation customers in 2026.

This report is based on public information and does not constitute investment advice. Markets carry risk; invest with caution.