Unitree, a Chinese maker of legged humanoid robots, sold 5,215 units in 2025 with 70% going to research labs, and a new U.S. import cap on certain foreign humanoids landed the same week.
Unitree Robotics closed its first day on Shanghai's STAR Market up roughly 460%, capping an IPO that raised about 6.1 billion yuan (~$905 million) and made the Hangzhou-based legged-robot maker the first major legged-robotics company to go public (The Robot Report).
Unitree sold 5,215 humanoid robots in 2025, up from 5 in 2023, and roughly 70% of those went to universities and research labs. Ghost Robotics CEO Gavin Kenneally estimates that fewer than 10% have reached genuine commercial deployment (The Robot Report). The IPO is the first time a humanoid maker has had to defend that gap on a quarterly disclosure cadence.
The listing itself was unusually clean. Unitree sold 40.45 million new shares, equal to 10% of its enlarged share capital, and reported 2025 revenue near 1.7 billion yuan (~$250 million). Overseas sales of 731.66 million yuan (~$108 million at the same implied rate as the IPO raise), about 44% of main-business revenue, made the export side of the business nearly as large as the domestic one (The Robot Report). Two valuation figures are circulating: a roughly 42 billion yuan (~$5.9 billion) implied mark on the raise, and a higher $9 billion figure reported at IPO pricing by CNBC. The two are not contradictory; they reflect different stages of the same listing (Forbes).
Other humanoid makers, including Figure AI at a cited $39 billion private valuation, have raised at figures chosen by their own investors (Forbes). Unitree's post-IPO price gives the rest of the field a number it did not get to pick, and a quarterly disclosure cadence that private rounds never required (Seeking Alpha).
Kenneally argues that the open question is whether any of these machines run a full shift without a human babysitter. Robot Report's filing analysis describes Unitree as evidence of a real hardware business with manufacturing leverage, but stops short of calling the industrial-humanoid case proven (The Robot Report). The company unveiled a new humanoid, branded "Superman," in the same IPO week: a standing jump past two meters, a top speed near 12.6 m/s, and a development cycle of roughly three months (The Robot Report). The demonstration answers a question about agility. It does not answer the one Kenneally is asking.
A U.S. import cap landed the same week. The Federal Communications Commission's restrictions, effective July 28, 2026, bar new imports of certain foreign-made humanoids, quadrupeds (four-legged robots), and autonomous mobile robots heavier than 4.4 pounds that ship with sensing, networking, or control software (The Robot Report). The rule is narrower than its reputation: it covers new devices, not every Chinese humanoid already in the U.S., and it grandfathers models authorized before the cutoff. China's commerce ministry has warned of countermeasures. For Unitree and its rivals, the practical effect is a hard U.S. addressable market, defined device by device, instead of an open one.
Four questions will read through the next four quarterly disclosures. Can the next cohort of humanoid makers reach a public-market bar without the listing collapsing under Kenneally's durability critique? Can Unitree shift the unit mix from research to factory, where the revenue is recurring and the warranty claims are real? Can the FCC carveout hold for pre-2026-authorized models, and what new device categories will regulators add next? Can the overseas revenue line, already 44% of main-business revenue, hold up if Beijing escalates the countermeasures already on the table?
Unitree's first quarterly filing will land in November. That is the next data point the field will read.