Successful test flights of the “Iron Man” prototype, robotic arms accurately picking tea, and assistive robots aiding the elderly in walking… Scenes once depicted in science fiction are now rapidly becoming a reality. In the 2025 government work report, “embodied intelligence” was explicitly included in the national strategic framework, injecting strong momentum into the development of this emerging technology field. As the industry is widely regarded as the “year of large-scale implementation,” the gold rush at the capital level is also heating up: the valuation of Zhiyuan Robotics has surpassed 10 billion, and Yushu Technology has officially initiated the IPO guidance process, with over a dozen companies in the embodied intelligence sector intensively submitting applications to the stock exchange. Meanwhile, internet giants like JD.com, Tencent, and Alibaba are entering the fray, accelerating their layout through strategic investments and acquisitions, which not only helps related robotics companies break free from dependence on a single capital source but also significantly raises the competitive entry barriers in the industry.
However, beneath the lively surface of this capital feast lies a reality that cannot be ignored: the vast majority of the highly publicized star companies are still mired in losses. Behind the halo of technology stocks is the commercialization dilemma and profit anxiety faced by the entire industry. The fervent pursuit of capital and the practical difficulties of industrial development together constitute the true picture of the embodied intelligence track in 2025.
Losses Become the Industry Norm: Even Leading Companies Struggle to Break Through
In 2025, the humanoid robot IRON launched by Xiaopeng marked the entry of the global humanoid robot competition into a white-hot stage through a fully public core technology disassembly demonstration, and Musk’s public praise further peaked this capital frenzy. However, behind the spotlight, financial data disclosed by several companies rushing for IPO reveal that net losses have become a common phenomenon in the industry. According to statistics from Phoenix Finance’s “IPO Observation”, several embodied intelligence companies, including Jizhijia-W, Yunjijia Technology, Yuejiang Technology, Stand Robot, Meijia Technology, Woan Robot, and Ledong Robot, have not achieved profitability in the 2024 fiscal year.
Even the company known as the “first stock of humanoid robots,” UBTECH, has also failed to escape the loss cycle. The company achieved a high revenue of 1.305 billion yuan in 2024, but its net loss still reached 1.124 billion yuan. Reflected in the capital market, its stock price has shown extreme volatility: it once fell below the issue price at the beginning of its listing, and then experienced a nearly 90% surge in a single day, with a total market value once exceeding 60 billion Hong Kong dollars. Meijia Technology, focused on intelligent robot solutions, saw its revenue grow from 455 million yuan in 2022 to 930 million yuan in 2024, doubling in scale, but its adjusted net loss has remained at a high level of 700-800 million yuan for three consecutive years, with no signs of narrowing.
Many companies are trapped in the dilemma of “the faster the revenue growth, the greater the loss.” Stand Robot achieved a revenue of 251 million yuan in 2024, while its cumulative loss from 2022 to 2024 has reached 273 million yuan; Ledong Robot achieved rapid expansion with a compound annual growth rate of over 40%, reaching a revenue of 467 million yuan in 2024, but its net profit loss still amounts to 56.48 million yuan; Woan Robot set a revenue peak of 610 million yuan in 2024, but its adjusted net profit showed a loss of about 3.07 million yuan, with a cumulative loss exceeding 100 million yuan over three years.
A long-time investor in the technology sector explained to Phoenix Finance’s “IPO Observation” that the embodied intelligence industry is characterized by high investment, long cycles, heavy R&D, and strong scene dependence, leading to a very high overall entry barrier for the industry. Currently, the industry is still in the early stages of commercialization exploration and has not yet reached the profitability inflection point. High R&D investment is one of the direct reasons for the losses: Stand Robot’s R&D expenses have consistently accounted for over 18% of its revenue, and UBTECH’s Chief Brand Officer Tan Min has also revealed that the company’s annual R&D investment accounts for about 40% of its revenue.
Of course, there are also profitable cases in the track. According to exclusive news from Wall Street Insights, Yushu Technology achieved a net profit of about 100 million yuan in 2024; Bomingwei Robot, focused on pipeline inspection and repair robots, disclosed continuous profitable financial data when submitting its Hong Kong stock listing application in August 2025, although the company also faces the risk of high customer concentration.
IPO “Single Plank Bridge”: The Capital Logic Behind Passive Sprinting
As embodied intelligence technology gradually emerges from the laboratory and enters family scenarios as “smart housekeepers,” an industrial revolution reshaping human-machine interaction has already begun. However, behind the glamorous transformation lies a core question worth pondering: why are these companies, despite continuous losses, rushing to go public?
The aforementioned technology sector investor revealed that as the financing amounts required for enterprise development continue to increase, the difficulty of financing in the primary market has been rising, making it an unavoidable choice for most companies to turn to the secondary market for fundraising through IPOs. More critically, the stringent betting agreements attached to early capital investments usually require companies to complete their IPO within five years, or else they will trigger share repurchase clauses. The repurchase agreements often require founders to bear joint liability and personal unlimited guarantees, and if they fail to go public on time, founders may face the situation of “selling their homes to pay off debts.” Under such pressure, regardless of whether the company has mature listing conditions, they can only be forced to sprint for an IPO.
“This essentially turns high-risk venture capital into a principal-protected investment product, a situation not uncommon in the domestic market,” the investor bluntly stated, noting that the embodied intelligence industry still needs three years to reach the commercial profitability inflection point, and achieving a level of “one robot per household” will require at least ten years.
Long-term growth and investment incubation partner Lao Yang, who has been deeply involved in the embodied intelligence field and also participates in the operation of the GEIA Global Embodied Intelligence Accelerator, analyzed for Phoenix Finance’s “IPO Observation” that the current IPO frenzy in the embodied intelligence sector is closely related to the relaxation of Hong Kong’s Chapter 18C, which allows unprofitable technology companies to go public. The R&D of embodied intelligence requires continuous large-scale funding, and the gradual decrease of domestic RMB capital investment in this field makes the Hong Kong market, which can connect with diversified international capital, a natural choice for companies.
It is worth noting that in 2025, “embodied intelligence” was first written into the government work report and elevated to a national strategy, with supporting policies being introduced in various regions. From the perspective of enterprises, as long as they can rush to the capital market to obtain funding support and seize the first-mover advantage in the industry, the current short-term losses will eventually be covered by the future trillion-level market scale. The capital market is also cautiously optimistic, looking forward to the arrival of the “ChatGPT moment” in the embodied intelligence field.
Lao Yang predicts that this wave of IPOs will bring dual industry impacts: on the one hand, it will accelerate the technological iteration and industrial evolution of the entire embodied intelligence track; on the other hand, it will also squeeze industry bubbles, pushing resources to concentrate on leading companies. For small and medium-sized innovative enterprises lacking financial support and not yet finding a clear business model, survival pressure will further increase.
Red Sea Game: Supply Chain Bottlenecks and Commercialization Dilemmas Await Resolution
As various players enter the market, the embodied intelligence track has rapidly shifted from a blue ocean to a red ocean of competition. Established companies like UBTECH and Yushu Technology occupy the first-mover advantage due to their early technological accumulation, while new entrants like Zhiyuan Robotics quickly penetrate the market by leveraging the technological dividends of AI large models. Against the backdrop of intensifying homogeneous competition, price wars have already erupted in the mid-to-low-end market, with some companies launching low-priced models priced at 80,000 to 100,000 yuan to compete with high-end models priced at 250,000 yuan.
More severe than the price war are the two core issues of supply chain fluctuations and high costs. In 2022, Stand Robot faced a backlog of semi-finished products worth tens of millions due to a shortage of imported core components like high-precision harmonic reducers, leading to cash flow tensions and even instances of delayed payments to suppliers. Although the crisis was alleviated through capital injection later, the uncertainty of the global supply chain still poses hidden dangers for enterprise production and operations.
Cost structure data shows that in the total cost of humanoid robots, core components such as servos, reducers, and controllers account for over 70%. How to break through core technologies through independent R&D and reduce the procurement costs of key components has become a challenge that all companies must tackle.
The popularity of the track has also given rise to investment chaos. According to Blue Whale News, some robotics companies have staged technical scams during roadshows by manipulating the perceived maturity of their technology. The aforementioned technology sector investor analyzed that the high heat of the embodied intelligence track has led many third- and fourth-tier capital sources, unable to participate in leading projects, to chase inferior projects, ultimately giving rise to industry chaos such as roadshow fraud, forming a distorted ecology of “mediocre teams financing and low-level capital investment.”
At a deeper level, the dilemma of commercialization landing presents a “chicken or egg” paradox: humanoid robots need to obtain data in real application scenarios like Tesla to continuously optimize algorithm models, but the prerequisite for data acquisition is the realization of large-scale market deployment of products. Currently, most robot products remain at the technical demonstration stage, facing issues such as insufficient load in industrial scenarios, and the complexity of home environments makes adaptation even more challenging, leaving a long way to go before truly achieving large-scale commercial use.
Path to Breakthrough: Scene-Driven Development and Ecological Layout are Key
Faced with profitability challenges and red ocean competition, industry participants urgently need to find sustainable development paths. Lao Yang suggests that companies should adopt a “scene-driven product” R&D strategy: first clarify specific application scenarios and core customer needs, and then develop targeted products and technical solutions. In his view, at the current stage, consumer-level scenarios such as companionship and entertainment, as well as some standardized industrial and service scenarios, are more feasible entry points.
Lao Yang particularly points out that utilizing existing embodied intelligence technology to provide “intelligent upgrade” services for traditional robotics and automation companies will be a market with huge potential, likely to generate numerous investment and acquisition opportunities. In terms of specific companies, he is optimistic about Yushu Technology’s product R&D capabilities and, from a capital operation perspective, sees Zhiyuan Robotics as having long-term competitive advantages due to its ecological layout capabilities.
Looking back at the history of the industry, Tesla also experienced a decade of losses before reaching the profitability inflection point. The current IPO frenzy in the embodied intelligence field is essentially a capital layout for the trillion-level market prospects. As leading companies like Yushu and Zhiyuan advance their IPO processes, the deep integration of capital markets and industrial development will enter a new stage. For all industry participants, the real test is just beginning: those who can first cross the breakeven point and establish replicable business models will ultimately gain the upper hand in this future-oriented industrial competition.
