Unitree Robotics IPO: How China's $610M Robot Startup Turned Profitable Before the Industry Did
March 20, 2026 marked something genuinely unusual. The Shanghai Stock Exchange accepted Unitree Robotics' STAR Market application, set to raise RMB 4.202 billion — roughly $610M USD — through at least 40.45 million new shares. If it clears, Unitree becomes China's first pure-play humanoid robot listed on the A-share market. That's a milestone worth pausing on.
But forget the milestone for a moment. Look at the financials.
Full-year 2025 revenue hit RMB 1.708 billion — a 335% jump year-on-year. Adjusted net profit surged approximately 674% to RMB 600 million. Gross margins expanded from roughly 44% to nearly 59.5%. In a sector where burning cash is practically a badge of honor, Unitree did something almost heretical: it turned a real profit, at scale, before the category even matured. That's the actual story.
The company didn't start as a humanoid robot startup. Its foundation is quadrupeds — four-legged robots. Over 30,000 units shipped cumulatively, commanding roughly 70% of the global market, with gross margins holding steady between 50–55%. That installed base gave Unitree something most competitors desperately lack: supplier leverage, field data, distributor relationships, and a genuine profit engine to fund the pivot to humanoids.
The pivot happened fast. Humanoid revenue jumped from just 1.9% of total sales in 2023 to over 51% in the first nine months of 2025. Full-year 2025 humanoid shipments exceeded 5,500 units — enough for Unitree to claim the global top spot, though Shanghai-based AgiBot reported 5,168 units over a comparable period. Definitions matter here. What counts as a humanoid? Delivered versus produced? Bipedal versus wheeled? Either way, the bigger signal is this: China now has at least two humanoid manufacturers shipping in the low-thousands. That manufacturing threshold has been crossed.
The pricing strategy is the real competitive weapon. Unitree lists its humanoid lineup publicly — R1 at ~$4,900, G1 at ~$13,500, H1 approaching ~$90,000 — with no invite-only mystique, no prototype theater. Think about how DJI conquered the drone market, or how Foxconn built electronics dominance. Neither won by having the flashiest product demo. They won through relentless pricing, availability, and iteration speed. Unitree is running that exact playbook.
The customer base reflects a market still finding its footing. Roughly 30% of buyers are universities and research institutes — Stanford, MIT, Tsinghua among them. About 45% come from Chinese state-owned enterprises using robots for inspection and logistics. The remaining 25% spans leasing and performance, including high-profile appearances at the CCTV Spring Festival Gala. Overseas revenue sits around 50%. The critical question for investors isn't first-sale volume — it's whether those SOE clients reorder after pilots wrap up. Repeat density within a single vertical would confirm genuine industrial substitution. Scattered demand across subsidized experiments suggests the "funded curiosity" phase still has legs.
Where does the $610M actually go? Primarily toward AI model R&D (~RMB 2B), hardware and robot body development (~RMB 1.1B), manufacturing base construction (~RMB 600M), and new product lines (~RMB 450M). That weighting says something important. This is a supply-chain and capital-expenditure race — not a software multiple story. At least not yet.
Unitree today is the strongest "body company" in humanoids. Full stop. Whether it becomes the strongest "brain company" is a different question entirely — and general-purpose autonomy in unstructured environments is still years away. Meanwhile, Xiaomi, BYD, and Midea are all wading into the hardware layer, which compresses the commoditization timeline considerably.
There's also a geopolitical ceiling that bullish takes tend to soft-pedal. A documented backdoor vulnerability in an earlier product line, plus U.S. congressional scrutiny over national-security concerns, creates real friction in Western enterprise and defense-adjacent markets — precisely where premium deployments concentrate. With half of revenue coming from overseas, that risk deserves a hard look.
Unitree may well become the category's defining infrastructure company before it ever becomes its smartest one. The bull case is structural: vast low-cost installed base, China's supply-chain density, and public capital ready to scale ahead of what many estimate as an "embodied AI inflection" roughly two to three years out.
The bear case is equally structural: hardware margins erode, the software layer capturing terminal value may accrue elsewhere, and public markets have a long history of overpaying for the gap between "category exists" and "category scales."
Unitree is profitable, credible, and an early industrial leader. Own it as that.
not investment advice