Unitree gave back $30bn and still sits near five times its pre-float valuation

Annualise the quarter in Unitree's own prospectus and the roughly $36bn the market still assigns the company after the slide works out near 1,500 times earnings ^[1][3]. At the intraday peak, the same crude sum gave about 2,800 ^[4]. Neither figure is a sensible read of a research-stage business, and that is the useful part: nothing in the filing sets a price, so the price came from somewhere else.

Somewhere else was the offer, and the offer was low. Analysts marked the company at $7bn or more in the weeks before the float ^[8], the listing price on 19 August valued it near $9bn ^[6], and the first-day peak was about seven times that ^[7]. Even after $30bn of value came off, the stock sits around five times the floor of those pre-float estimates ^[5]. A 45% fall sounds like a reckoning. It has not yet carried the price back to where the sell side began.

The plumbing explains most of the round trip. Chinese exchanges vet listing candidates and also guide IPO pricing, which caps how far bankers can lift an offer to meet feverish demand; analysts told Reuters this distorts prices ^[10]. Rules restrict short-selling, so an overpriced open meets nobody betting against it, bankers told Reuters ^[11]. The first day was therefore not price discovery so much as the release of a queue. Yuan Yuwei of Trinity Synergy Investments put the general case to Reuters bluntly: a stock worth 10 yuan can open at 100 and slide for years, and he called that a rip-off ^[14].

What the machines can do was never the disputed variable. One Unitree robot beat Usain Bolt's 100m record at the World Humanoid Robot Games in Beijing this month, a result roboticists were quick to qualify ^[15]. Broader commercial applications have been harder to come by ^[16]. Gao Xingkun, who manages a fund at China Southern Asset Management, told an online roadshow that judging these companies on profit alone is unfair, because many robot makers spend heavily on research while commercial orders are not yet in sight, and he likened the moment to the early years of China's electric vehicle industry ^[17]. That is the bearish case delivered in a bullish voice. The item he identifies as absent is orders, and China's EV makers earned their re-rating on volume they eventually shipped.

Wang Xingxing, who holds about a third of the shares, said days after the listing that robots are approaching a "ChatGPT moment" ^[18][20]. Possibly. But the three sessions of decline did not test whether a biped can sprint; they tested what the sprint sells. Between the vetting, the guided offer price and the restriction on shorts, the only investors positioned to register a view were the ones who had to buy on the tape after the open. They have now registered it, and the multiple they are left holding still assumes a revenue line the prospectus does not yet show.