Unitree's $904M IPO — The Humanoid Robot Race Is Going Public. And China Is Already Profitable.

For most of the last decade, humanoid robots have been a demonstration product. They walk across a stage, pick up a box, climb some stairs, and get a standing ovation from a conference audience. Then the presentation ends, the robots go back to the lab, and the question of when any of this turns into a real business goes unanswered. That question now has an answer — at least from one company. And the answer is coming from China.

Unitree Robotics, the Hangzhou-based company behind the G1, H1, and R1 humanoid robot platforms, is pushing toward a high-profile Shanghai Stock Exchange IPO seeking to raise approximately 6.1 billion yuan — roughly $904 million. What separates Unitree from the dozens of robotics startups that have announced similar ambitions and produced nothing but press releases is a number that almost no humanoid robotics company in the world can claim: it is profitable. Unitree generated nearly 1.7 billion yuan in revenue in 2025, with more than 40% coming from overseas markets, and it turned that revenue into actual profit. The humanoid robot race is no longer just a venture capital bet on a distant future. It is a business. And it is about to be a public one.

Who Is Unitree — And Why It Is Winning

Unitree was founded in 2016 by Wang Xingxing, a robotics engineer from Zhejiang University who started building quadruped robots in his spare time as a graduate student. The company's first product — a dog-like quadruped called AlienGo — cost a fraction of what Boston Dynamics was charging for Spot, its equivalent platform. That price gap was not a compromise. It was a strategy. Wang built Unitree around the idea that robotics would only achieve real scale if the hardware became affordable enough to deploy outside government labs and prestige research institutions.

That philosophy produced the Go2 — a consumer-accessible quadruped that retails for under $2,000 and became the most widely deployed four-legged robot in history across universities, research labs, and industrial inspection operations. It also produced the H1 humanoid, the G1 — a more compact and affordable humanoid released in 2024 — and the R1, Unitree's latest platform designed specifically for commercial deployment in manufacturing and logistics environments.

The company's competitive advantage is not that it builds the most capable robot in any category. Boston Dynamics' Atlas is more capable than the H1. Figure's 02 robot has more sophisticated manipulation. Tesla's Optimus has more vertically integrated supply chain ambitions. Unitree's advantage is that it builds robots that are good enough to be useful at a price point that allows real customers to place real purchase orders rather than signing memorandums of understanding for machines that do not yet exist. That is a fundamentally different business model from most of its competitors — and it is why Unitree is profitable while they are not.

The IPO — What the Money Is For

The 6.1 billion yuan raise targets the Shanghai Stock Exchange's STAR Market — China's technology-focused exchange designed to list high-growth companies under more flexible listing criteria. The backing Unitree brings to this listing is credible: investors include Tencent, Alibaba, Meituan, and multiple state-linked funds that China's government has been directing toward its "embodied intelligence" strategic priority — the government's term for AI systems that operate in the physical world through robotic bodies rather than existing purely in software.

The proceeds target three areas. First, manufacturing scale: Unitree currently produces robots at a rate that serves research and early commercial customers, but the R1 platform targets factory floor deployment at volumes that require dedicated high-throughput production lines. Second, R&D acceleration: manipulation — the ability to handle objects with the dexterity and reliability a factory environment demands — remains the hardest unsolved problem in humanoid robotics, and Unitree's current platforms are capable but not yet at the precision level that replaces skilled human workers on complex assembly tasks. Third, international expansion: with 40% of revenue already coming from overseas, Unitree is building direct commercial relationships in Europe, Southeast Asia, and cautiously in North America, where regulatory and export-control scrutiny of Chinese robotics companies is increasing.

The Competition — How Unitree Sits in the Global Race

The humanoid robotics market in 2026 is simultaneously overcrowded and underdeveloped. Overcrowded in the sense that hundreds of companies globally are developing bipedal robots and claiming proximity to commercial deployment. Underdeveloped in the sense that none of them — including Unitree — have yet demonstrated the reliability, manipulation capability, and total cost of ownership that would justify replacing human workers at industrial scale for general-purpose tasks.

That gap between demonstrated capability and commercial viability maps onto the competitive landscape in a revealing way. Tesla's Optimus is the most watched platform globally, but CEO Elon Musk's most recent timeline places the Fremont production line starting at the end of 2026 — and that production line is intended to produce Optimus robots for Tesla's own factories first. Boston Dynamics Atlas, now owned by Hyundai, is the most capable platform in terms of whole-body movement and dynamic balance, but it remains extremely expensive and targets narrow high-value industrial applications. Figure, Physical Intelligence, Apptronik, and Agility Robotics are all well-funded US startups chasing the same commercial opportunity with platforms at varying stages of readiness.

China's "Embodied Intelligence" Strategy — Why the Government Is All In

Unitree's IPO does not happen in isolation. It is one piece of a deliberate national industrial strategy. Beijing has designated "embodied intelligence" — AI systems that operate through physical robotic bodies — as a strategic technology priority alongside semiconductors, quantum computing, and electric vehicles. The rationale is consistent with China's broader technology policy: identify the categories that will define economic and military power in the next 20 years and build domestic leadership before dependence on foreign technology becomes entrenched.

The manufacturing supply chain argument is particularly compelling for China. Humanoid robots require actuators, sensors, cameras, batteries, motor controllers, and structural components — almost all of which China already produces at scale for electric vehicles and consumer electronics. The reduction in robot costs that Unitree has demonstrated is not purely an engineering achievement. It is partly a supply chain achievement: access to the world's deepest manufacturing base for the components a robot requires. A US or European company building a humanoid robot sources its actuators from a smaller set of suppliers at higher prices. Unitree sources from the same ecosystem that builds EV motors for BYD.

The state-linked investment in Unitree's cap table — alongside Tencent and Alibaba — reflects how seriously Beijing is taking the sector. State-linked funds do not invest in robotics companies for financial returns alone. They invest to ensure that China's national champions in a strategic sector have the capital they need to scale before foreign competitors can establish dominant positions.

What Robots Are Actually Being Used For Right Now

The gap between the marketing imagery of humanoid robots — gracefully assembling cars, folding laundry, stocking shelves — and what they actually do commercially in 2026 is still significant. The current commercially viable use cases are narrower but real.

The gap between current deployment and the long-promised general-purpose robot that can do anything a human can do remains large. Manipulation — the fine motor control required to handle irregular objects, assemble complex components, or perform the dexterous tasks humans handle routinely — is the core unsolved problem. Every major robotics company is working on it. None has solved it at the reliability and cost required for broad industrial deployment. The question Unitree's IPO implicitly answers is: you do not need to solve general-purpose manipulation to build a billion-dollar business. You need to solve enough specific tasks well enough to generate repeatable revenue. That is what Unitree has done.

The Broader Signal — Why This Matters Beyond One Company

Unitree's IPO is the clearest signal yet that humanoid robotics has crossed from science project to industry. When a company can take a complex, expensive, technically demanding product category to public markets on the strength of real profitability rather than projected revenue from theoretical future products, it means the category has arrived at a stage of commercial maturity that attracts a fundamentally different class of investor and customer. Pension funds and institutional investors who cannot touch venture-stage companies can own shares in a profitable public robotics company. Industrial customers who cannot justify the procurement risk of buying from a pre-revenue startup can sign contracts with a company that has a trading history and audited financials.

That shift in who can participate — in investment and in procurement — is what accelerates the pace of deployment. Unitree going public does not solve manipulation or make general-purpose robots commercially viable overnight. But it puts significantly more capital to work on the problem and significantly reduces the friction for the customers who are ready to deploy what the robots can already do.