The Event
On July 2, the China Securities Regulatory Commission (CSRC) approved the registration application of Unitree Robotics (宇树科技) for an initial public offering and listing on the STAR Market (科创板). The approval is valid for 12 months from the date of registration. In other words, within at most a year, this Chinese legged/humanoid robot company founded in 2016 will enter mainstream investors'视野 via the public market.
According to disclosed prospectus data: founder Wang Xingxing, born in 1990, established Unitree in Hangzhou in 2016. The company's core business is civilian legged robots and humanoid robots, with leading global quadruped robot sales. As of June 2025, total headcount exceeds 1,000.
The most notable point is the "full-stack in-house" supply chain. Unitree has achieved fully domestic self-development of core components—joint motors, reducers, and controllers—meaning the cost of a joint unit is only about one-third of imported products. For a hardware company, this cost structure is both the foundation of pricing power and the key to resisting supply chain geopolitical risk.
Deep Analysis
First, this is the first milestone for "Chinese embodied AI" moving from private fundraising into public markets.
In recent years, domestic embodied AI companies have raised funds mainly in private markets—Unitree itself went through several rounds. But private-market valuations circulate only among institutional LPs, with poor liquidity, opaque pricing, and weak cycle resistance. An IPO is the event that truly hands pricing power and oversight to the public market. Once listed, Chinese embodied AI will have its first "public yardstick": investors can evaluate the intrinsic value of embodied AI companies using the same financial models, short-selling mechanisms, and brokerage research systems.
Second, the STAR Market's tolerance for "hard tech + sustained losses" is the practical basis for Unitree's path.
Embodied AI companies are generally in a state of "high R&D investment, unprofitable, unstable orders"—vastly different from traditional manufacturing financial models. The STAR Market's fifth listing standard (allowing unprofitable companies to list) and hard requirements on R&D investment ratio are tailor-made for such companies. Unitree getting approval is, to some degree, regulatory endorsement that "embodied intelligence is a national-level hard-tech direction."
Third, "full-stack in-house core components + 1/3 cost" carries strategic significance—amid the geopolitical backdrop—far greater than its commercial significance.
Joint motors, reducers, and controllers are the "muscles" of robots, traditionally monopolized by Japan's Harmonic Drive and the US's Moog. If export controls tighten (as with 2022 restrictions on Category A robot exports), China's embodied AI industry would immediately face chokepoints. By localizing these components at one-third the cost, Unitree has effectively bought a "decoupling insurance policy" at the supply chain level.
Fourth, Wang Xingxing's personal shareholding deserves attention.
The founder's personal stake reflects governance stability and long-term vision. In a赛道 with 5–10 year product cycles, founder control directly determines R&D cadence. Wang's technical background (robot mechanical structures, control systems, hardware engineering) is a scarce profile among domestic peers—technology-savvy embodied AI founders are rare even globally.
Why It Matters
Unitree's IPO approval is the first complete closed loop for "Chinese embodied AI"—from concept to engineering to public market validation. It sends at least three signals to the industry:
- Chinese regulators consider embodied intelligence a strategic tech direction and are willing to back it with the strictest capital market (STAR Market registration-based IPO system)
- Investors finally have a publicly comparable embodied AI company for due diligence; pricing of later companies will be more transparent
- Full-stack in-house upstream components has been proven a "listable" business model; domestic substitution on the supply side will accelerate
Risks and What to Watch
First, "fifth-standard approval + IPO registration approval" ≠ immediate listing—roadshows, pricing, subscription, and other steps remain. The 12-month approval window is a "window," not a "promise"; the bell-ringing is not guaranteed.
Second, embodied AI's "commercialization inflection point" has not arrived. Unitree leads global quadruped sales, but quadruped applications remain mainly performance, education, and research—not "essential consumer goods." Humanoid robots are still at PoC stage. How a hardware-plus-algorithm company explains "three-year growth" in its prospectus will determine what P/E the market is willing to pay.
Third, even if the IPO succeeds, embodied AI valuation models remain immature—traditional manufacturing uses P/E, software companies use P/S; what multiple applies to embodied AI has no global consensus. Unitree's listing will, in a sense, force primary and secondary markets to jointly explore a new valuation framework.
Source: CSRC announcement of July 2, IT之家 compilation, Unitree prospectus disclosures.