This article is reprinted from Sohu, all rights reserved. Click to view the original link.

---

After a "Lightning" 104-Day Approval: Unitree Robotics and the "Mass Production Test" of an Industry

On July 2, a single approval on the CSRC website thrust Unitree Robotics into the spotlight of China's capital market. From the acceptance of its STAR Market IPO application on March 20 to registration taking effect, it took only 104 days—the fastest review record since the implementation of the STAR Market's pre-review mechanism. As the world's first company to achieve scaled profitability in humanoid robots before an IPO, it has officially obtained its "ticket" to the A-share market and will become the "first humanoid robot stock" on the A-shares.

This is not just a capital milestone for one company. Behind Unitree Robotics lies a hundred-billion-yuan industry undergoing a profound transformation from lab to production line, and from concept to order. The year 2026 is widely recognized by the industry as the "Year One of Mass Production" for humanoid robots—not the "Year One of Demos" or "Year One of Financing," but the year when "real machines roll off the line, customers sign for delivery, and production lines run." While Tesla's Optimus moves materials at the Fremont factory, Figure AI completes a new funding round with a $39 billion valuation, and Chinese manufacturers capture nearly 80% of the global market share, a grand narrative involving technology, capital, and industrial landscape is unfolding.

From Quadruped to Bipedal

Unitree Robotics' story began in 2016. At that time, humanoid robots were still concepts in sci-fi movies, and quadruped robots were the starting point for this startup. Relying on full-stack self-developed servo motors, motion control algorithms, and cost-control capabilities, Unitree accumulated over 30,000 cumulative sales units in the quadruped robot sector.

The real turning point occurred in 2023 and 2024—Unitree successively launched its first humanoid robot H1 and mid-sized humanoid robot G1. In 2025, the company shipped over 5,500 humanoid robots (pure humanoids, excluding wheeled dual-arm robots), ranking first globally in shipment volume. According to third-party statistics, this figure accounts for approximately 25% of the global market share.

Accompanying this was a structural leap in performance. From 2023 to 2025, Unitree Robotics' operating revenue soared from 159 million yuan to 1.699 billion yuan, while net profit after deducting non-recurring items turned from -18 million yuan to 591 million yuan. Even more noteworthy is its gross margin level—from 2023 to 2025, they were 44.22%, 56.41%, and 60.27% respectively. Although the gross margin for humanoid robots fell from 87.7% to 62.9% as mass production expanded, it remained at an extremely high level.

However, behind the rapid growth lie concerns. In Q1 2026, Unitree Robotics achieved revenue of 423 million yuan, up 68% year-on-year, but net profit after deducting non-recurring items was only 40.2536 million yuan, down 52.55% year-on-year. The company explained that it had gradually increased investment in embodied large model R&D since 2024, leading to short-term pressure on profits.

This IPO plans to raise 4.202 billion yuan, corresponding to an issuance valuation of approximately 42 billion yuan. The largest single investment—2 billion yuan concentrated on large model R&D—accounts for 48% of the total raised funds, focusing on filling the "brain" gaps in robotic intelligent decision-making, autonomous interaction, and environmental perception. Unitree Robotics is transitioning from "hardware self-development" to full-stack competition of "Brain + Body."

A "National-Level" Industrial Mobilization

Unitree Robotics' lightning-fast IPO is not an isolated event. On June 9, 2026, just eight days after Unitree passed the review, the Ministry of Industry and Information Technology (MIIT) and the State-owned Assets Supervision and Administration Commission (SASAC) jointly issued a notice, officially launching the 2026 Special Action for Real-Scene Training of Humanoid Robots and Embodied Intelligence.

The action goals are clear and aggressive: By the end of 2026, key products like humanoid robots will take the lead in completing application verification and routine deployment in a batch of representative scenarios, initiating "work mode"; condense and form over 100 high-value application scenarios, driving the formation of ten-thousand-unit scale deployment capability. The action deploys six key tasks: creating real-scene training spaces, forming innovation application consortia, tackling practical work skills, strengthening real-scene application verification and routine deployment, reinforcing key element guarantees, and distilling mature experiences.

This is not the first supportive document. In 2025, embodied intelligence was included in the Government Work Report for the first time; in February 2026, China released its first full-industry-chain standard system, the Humanoid Robot and Embodied Intelligence Standard System (2026 Edition). From national strategy to industry standards, from fiscal support to scenario opening, the policy side is building a complete institutional framework for this industry.

Market data confirms the strength of the policies. Global humanoid robot shipments in 2025 were approximately 18,000 units, a year-on-year increase of 508%; TrendForce estimates global shipments will exceed 50,000 units in 2026; Morgan Stanley raised its 2026 China market sales forecast to 28,000 units and expects them to reach 262,000 and 2.6 million units in 2030 and 2035 respectively. Goldman Sachs predicts the global humanoid robot market size will reach $38 billion by 2035.

The "Dual-Track" Global Competitive Landscape

In this global race, a clear pattern is emerging: Chinese companies dominate mass production, while US companies focus on AI. Of the humanoid robots delivered in 2025, 90% came from China. Among global humanoid robot suppliers, Chinese companies account for 51% of the number, compared to about 23% for the US. Specifically regarding market share, Unitree Robotics ranks first with 25%, followed closely by Agibot with 24%. Agibot delivered over 5,000 units in 2025, and UBTECH Robotics delivered about 1,000 units. These three companies collectively control nearly 80% of the global market.

The advantage of Chinese companies is rooted in industrial cluster effects—the unit cost of actuators, sensors, and drivers is only one-third that of the West. An entry-level humanoid robot provided by Unitree costs only $6,000, whereas Tesla's Optimus targets a price of $20,000 to $30,000.

But the US catch-up cannot be underestimated. Figure AI completed a new funding round with a $39 billion valuation, second only to Tesla's Optimus. This startup, jointly invested in by Microsoft, OpenAI, and NVIDIA, is deeply integrating large model capabilities, enabling robots to understand natural language instructions and autonomously plan action sequences. Tesla is converting the assembly lines for Model S and Model X into Optimus production lines, with Elon Musk confirming production will start in late July or August 2026.

The difference between the two paths is essentially a difference in strategic choice: Chinese companies emphasize mass production landing and cost optimization, while top US companies focus on software iteration and "brain" models. In the short term, China leads significantly in shipment volume; in the long term, the gap in "brain" capabilities may become the decisive variable.

From "Walking" to "Working"

Despite broad prospects, the humanoid robot industry is far from a moment to "rest easy." Multiple industry experts define 2026 as the "Scale Verification Period"—capacity ramp-up, scenario verification, and technical convergence advancing simultaneously. Between "stunning reveals at launch events" and "stable work on production lines," at least three challenges stand in the way.

First Challenge: The Data "Hunger Games." Industry consensus holds that embodied large models capable of general autonomous ability require at least tens of millions of hours of high-quality real interaction data. However, as of early 2026, the global supply of compliant, usable effective data from real machines plus no-body sources is only 500,000 hours, with a gap exceeding 99%. Compared to automotive autonomous driving, which possesses hundreds of millions of hours of data, robot data remains stuck in manual collection and material annotation stages. Unitree Robotics' investment of 2 billion yuan in large model R&D is precisely an attempt to fill this gap.

Second Challenge: The Gap from "Lab" to "Real World." Currently landable scenarios are highly concentrated in three types of structured environments: industrial handling, warehouse logistics, and hazardous environment operations—where environmental rules are clear, task boundaries are defined, and error tolerance is higher. For home-service humanoid robots to work reliably in unstructured environments (messy living rooms, children and pets roaming, vague user instructions), longer technical accumulation is still needed. In Unitree Robotics' humanoid robot sales revenue from January to September 2025, scientific research and education accounted for as high as 73.60%, commercial consumption for 17.39%, and true industry applications for only 9.01%—the expansion of "working" scenarios has just begun.

Third Challenge: The Commercial Loop Has Not Been Closed. Some industry experts point out that the embodied intelligence supply chain is extremely long, covering hardware mass production, engineering delivery, scenario adaptation, and data iteration. "Teams with one-way technical leadership but no experience in engineering landing find it difficult to close the commercial loop." If hidden expenses such as equipment depreciation, annual maintenance, part replacement, and software O&M are further added, the total cost of ownership over the robot's lifecycle could severely squeeze profit margins. Moving from "single-scenario pilots" to "cross-scenario scaled commercialization" still requires multi-party collaboration to break through.

The Future Watershed

Regarding the main competitive thread for the next 2-3 years, industry insiders' judgments are highly consistent: The key lies not in the height of gross margins, but in the data closed-loop capability of embodied intelligence large models and the depth of scenario landing. Market competition will shift from "whose robot looks more like a human" to "whose robot can work stably."

This means that the real test for Unitree Robotics after its IPO has only just begun. Can the 2 billion yuan investment in large models translate into true "brain" capabilities? Can the shipment of 5,500 units generate enough high-quality real data? Can industry applications beyond scientific research and education truly open up? The answers to these questions will determine whether Unitree Robotics can grow from being the "First Humanoid Robot Stock" to becoming the "Strongest Humanoid Robot Company."

For the entire industry, 2026 is both a watershed and a starting line. Policies are in place, capital has entered, and production lines are running—but as Elon Musk frankly admitted when inspecting the Optimus production line, initial production will be "extremely slow because everything is new." From "walking" to "working," from "ten-thousand units" to "million units," the Chinese humanoid robot industry still needs to cross one threshold of the real world after another.