Yushu Technology strives to become the "First Stock in Humanoid Robots," with only 30% of its applications in commercial and industrial sectors.

Source: Emerald King Capital

By | Emerald King Capital

From completing the stockholding-company restructuring, to initiating filing for listing advisory services, and then formally submitting its prospectus to take aim at the STAR Market, Unitree Robotics’ capital-exization process has entered a critical stage.

On March 20, the Shanghai Stock Exchange officially accepted the IPO application of this robot company listed among “Hangzhou’s six little dragons.” Unitree plans to raise RMB 4.2B for R&D upgrades and capacity expansion, launching a bid for “China’s first A-share humanoid robot stock.”

Compared with topics that the market previously focused on—such as capital entering the picture and equity-structure arrangements—this time the financial data disclosed in the prospectus, for the first time, clearly depicts the reality of Unitree Robotics’ profitability.

This company, which has appeared on the CCTV Spring Festival Gala stage twice, has delivered an impressive performance with non-recurring items deducted net profit exceeding RMB 400 million. However, beneath the glittering façade, the core concern that commercialization is taking a slow pace is still a hurdle that cannot be avoided for its IPO review and subsequent capital development. Whether technological dividends and market heat can be converted into sustainable commercial value still needs long-term verification by the market.

Non-recurring profit surge

When the humanoid robot industry has generally fallen into the dilemma of “high R&D investment and low revenue realization,” and market stereotypes about robot companies—“heavy on technology, light on profits”—are deeply rooted, Unitree Robotics’ profitability performance is truly an exception in the industry, completely breaking the industry’s profit curse.

According to the prospectus, in January–September 2025, Unitree achieved net profit attributable to owners of the parent company after deducting non-recurring gains and losses of RMB 431 million. In FY2025, after review, that figure reached RMB 600 million, a year-on-year increase of 674.29%. After excluding distortions from items such as share-based payments and other non-recurring items, Unitree’s core operating profit still maintained explosive growth, and its profit base is relatively solid.

Looking back at the performance trajectory, Unitree Robotics was still loss-making in 2022–2023. In 2024, it successfully turned profitable, recording non-recurring net profit of RMB 77.5036 million. In just one year, its profitability scale underwent a multiple-fold leap, with a growth speed that leads the industry. Revenue growth was equally strong: in 2025, Unitree’s full-year revenue reached RMB 1.71B, up 335.36% year on year; in January–September 2025 alone, revenue had already reached RMB 1.17B, far exceeding the RMB 392 million revenue level for all of 2024. The growth momentum has been fully unleashed.

Behind the performance surge are two factors: the brand breaking through the mainstream driven by its CCTV Spring Festival Gala exposure, and a fundamental leap in revenue structure. The Unitree H2 humanoid robot, standing 1.8 meters tall, appeared in consecutive Spring Festival Galas in 2025 and 2026. Its stage performance sparked widespread attention online, directly propelling the humanoid robot business to become Unitree’s largest revenue source.

Data comparison shows that in FY2023, Unitree’s humanoid robot revenue was only RMB 2.9671 million, accounting for less than 2% of total revenue. In the first three quarters of 2025, revenue from this business had already approached RMB 600 million, with its share surpassing 51%. Meanwhile, as a traditional core business, the quadruped robot line has continued to grow steadily; in the first nine months of 2025, its revenue reached RMB 488 million, accounting for 42.25% of total revenue. The new and old businesses complement each other and provide support.

As revenue scale expands, Unitree’s main business gross margin rose from 44.18% in 2022 to 59.45% in the first three quarters of 2025. After review, the full-year 2025 gross margin further increased to 60.27%. This advantage is attributable to cost reduction from core components being developed and produced in-house, along with large-scale mass production, as well as brand premiums formed when high value-for-money products seize the market—further strengthening its earnings resilience.

For overseas and domestic market deployment, from 2022 to 2024 Unitree’s share of overseas revenue stayed above 55%, becoming a stable base for profitability. In the first three quarters of 2025, driven by improved brand awareness domestically and domestic revenue growth far outpacing overseas impacts, the overseas revenue share declined to 39.2%. However, overseas market recognition of Unitree’s products has not weakened; the foundation for globalization remains solid.

R&D investment and revenue growth are not matching

However, although Unitree Robotics has achieved full-stack in-house development, its technology roadmap still shows clear shortcomings. Issues such as the allocation ratio of R&D spending and the implementation of AI large models have become the core focus of both regulators and the market, and they also leave hidden risks for subsequent commercialization rollout.

First, R&D investment and revenue growth do not match, and the expense ratio has been falling continuously.

In January–September 2025, Unitree’s R&D expenses were RMB 90.2094 million, and the R&D expense ratio was only 7.73%. While the absolute value of R&D expenses increased slightly, the portion of revenue it represents dropped noticeably compared with prior years. In the humanoid robot industry, where technological iteration is accelerating—and where core technologies such as motion control and embodied large models determine long-term competitiveness—R&D investment has not been increased in step with revenue surging. Instead, it needs to be supplemented using IPO proceeds to make up for R&D shortcomings. This arrangement has led the market to question the sustainability of its technology.

Second, the “brain” layout for AI is lagging, and core competitiveness is concentrated on the body hardware.

Both rounds of regulatory inquiries focused on the humanoid robot’s AI integration capabilities. Unitree’s core strength, however, is concentrated in “body capabilities”—using high-precision joint module components, advanced drive technologies, and real-time motion control algorithms to build hardware barriers. In the stage where AI large models have not yet matured, it relies on the top-tier “body” performance of robots to seize the market.

But in terms of its technology roadmap, Unitree’s core technologies and in-development projects are mainly concentrated in the “small brain” motion control domain. The two embodied large-model projects are still at the basic research stage. The two existing large models are only in R&D testing; they have not yet been applied in volume to products, nor have any technology shortcomings been addressed through external procurement. Relevant progress remains on par with the industry’s average level.

In response to regulatory inquiries, Unitree stated that its embodied large models are still in a technical exploration phase. Globally, a mature roadmap has not yet formed. The company has already achieved phased results in the two major mainstream model tracks: WMA and VLA, placing it in the first tier of embodied intelligence. But it cannot be denied that hardware advantages are difficult to form long-term barriers. The technical lag of the AI “brain” directly constrains robot scenario adaptation and large-scale applications, becoming the core obstacle to breaking through commercialization.

Commercial and industrial application share is too low

The technical shortcomings at the product level directly carry over to the commercialization rollout process. Even with impressive financial statements in hand, Unitree still struggles to solve the industry problem of “good reviews but weak sales.” The first-mover advantage is being continuously diluted by intense market competition.

From the perspective of application scenarios, Unitree’s humanoid robots still remain at the initial demonstration stage and have not achieved large-scale industrial deployment.

Although in 2025 its shipment volume exceeded 5,500 units to rank first in the industry, its revenue structure is extremely concentrated. In January–September 2025, revenue from the research and education sector accounted for 73.6%, while commercial consumption accounted for 17.39%. Applications in core industries such as industrial production and domestic services accounted for less than 10%. The products are still mainly focused on technical showcases and research experiments, not yet transformed into real industrial productivity. In contrast, peers such as Tesla and Boston Dynamics have product planning centered on large-scale industrial-scenario applications, and their development paths have already diverged clearly.

Although Unitree claims that this layout follows the evolution pattern of “R&D—pilot—deployment” for frontier technologies, the industry’s competitive landscape has quietly changed. On one hand, international giants and tech companies are accelerating entry: Tesla’s Optimus mass production is entering the countdown, Xiaomi’s CyberOne has opened for presales. These companies, leveraging advantages in capital, supply chains, and ecosystems, directly challenge Unitree’s market share and pricing power. On the other hand, domestic startups such as Yunshenchu and Juzhi Dongli focus on segmented tracks. By using differentiated technologies to divert customers, competition in the industry has shifted from a single-tech battle to a full-scale contest of “commercialization speed.”

More importantly, the logic of capital-market investment has fundamentally changed.

Since 2026, companies focusing on robot “brain” technology such as Zhipingfang and Zivariable have entered the ranks of hundred-billion valuation firms one after another. Figure AI achieved a valuation of $39 billion by leveraging its VLA large model. Capital has shifted its bet from “cool motion capabilities” to “production power that lands in real scenarios.” This means that without the ability to deliver long-term performance recognition from capital solely supported by hardware advantages and market heat.

As of June 2025, Unitree Robotics’ asset-liability ratio was 18.42%, and the financial fundamentals were relatively stable. This RMB 4.2 billion fundraising provides it with an opportunity to resolve its commercialization dilemma. If the company can truly direct the raised funds into core technology breakthroughs, AI large-model deployment, and adaptation to vertical scenarios—making up for technology shortcomings and connecting the commercialization chain—its long-term growth space is likely to be fully opened. But if the company continues to remain at the level of technology demonstrations and capital-driven hype, even if it ultimately succeeds in listing on the STAR Market, it will be difficult to gain sustained recognition from the capital market.

What the capital market ultimately recognizes has never been the flattering numbers from financial statements in the moment. It is the real commercial value that can be deployed, can generate profits, and can sustain itself. For Unitree Robotics, the IPO is only the starting point of capitalization. Solving commercialization issues is the core key to standing firm in the track.

A massive amount of news and precise analysis are available on Sina Finance App

Responsible editor: Yang Hongbo