On Monday, the Hangzhou robotics company Unitree unveiled a humanoid it calls Superman. The claims were specific: a standing vertical jump of 2 metres, and a top running speed of 12.66 metres per second. The company says both figures pass the equivalent human records, and on the sprint number it is right — World Athletics biomechanists clocked Usain Bolt’s peak velocity during his 9.58-second world record in Berlin at 44.72 km/h, which is 12.42 metres per second. Unitree says the whole machine came together in a little over three months.

On Wednesday, two days later, Unitree listed on Shanghai’s STAR Market. The stock priced at 150.80 yuan. It opened as much as 629% higher, touching 1,100 yuan — roughly $163 a share — before settling to close up 460.34%, at a market capitalisation of 341.77 billion yuan, about $50.7 billion.

Unitree’s revenue for all of 2025 was 1.699 billion yuan. That is about $240 million.

The Robot

Superman is the fourth generation of a line that has been closing on human athletic benchmarks for two years. Unitree’s earlier H1 model reached 10 metres per second, described at the time as approaching Bolt’s peak. Superman has a leg length of 0.85 metres — shorter than an adult sprinter’s — which means the speed is coming from stride frequency and joint torque rather than reach.

Chinese analysts quoted on the launch were careful to frame the numbers as engineering demonstrations rather than product specifications. “Such extreme performance tests essentially serve as stress tests for a robot’s underlying hardware and software coordination,” said tech analyst Ma Jihua. Industry director Xiang Ligang made the same point mechanically: a machine doing this “must coordinate multiple joints in real time and remain stable upon landing.”

Why a Sprint Record Is Not a Product
Nobody is buying a humanoid to run the 100 metres. What a 12.66 m/s sprint and a 2-metre standing jump actually prove is that the actuators, the balance controller and the landing logic survive loads far past anything a warehouse or a factory floor will ask of them. It is a headroom demonstration — the robotics equivalent of a benchmark score. The commercial question is a different one entirely: how many of these get sold, to whom, and for what job.

What the Market Bought

The IPO itself was conventional in size and extraordinary in reception. Unitree sold roughly 40.4 million shares — about 10% of its enlarged share capital — raising 6.1 billion yuan, or around $905 million, at a pricing valuation near 60.9 billion yuan. Then the market repriced it by a factor of five and a half in the first minutes of trading.

Unitree’s DebutFigure
ListingShanghai STAR Market, August 19, 2026
IPO price150.80 yuan (~$22)
Opening move+629% (to 1,100 yuan)
Close+460.34%
Closing market cap341.77bn yuan (~$50.7bn)
Raised6.1bn yuan (~$905m), ~40.4m shares, 10% of capital
2025 revenue1.699bn yuan (~$240m)
2023 revenue159m yuan
2025 net profit~278m yuan (~$41m)
Core gross margin60.13%
Price / sales~210×
Price / earnings~1,200×

Two things in that table are worth separating, because they point in opposite directions.

The business is real. Revenue went from 159 million yuan in 2023 to 1.699 billion in 2025 — a tenfold increase in two years — at a core gross margin above 60%. Unitree is profitable, which almost no humanoid robotics company anywhere is. Its 2025 sales split roughly evenly between quadrupeds (697 million yuan) and humanoids (868 million yuan), and the humanoid half is the half that is growing.

The valuation is a different animal. At the close, investors were paying about 210 times revenue and roughly 1,200 times earnings for that business.

9.8 Million Accounts, 9.7 Million Shares

The demand figures explain the open better than the fundamentals do. The offering was oversubscribed by more than 8,000 times overall — a record for the STAR market. In the online retail tranche, roughly 9.8 million retail investment accounts competed for about 9.7 million shares. The final allocation rate was 0.018%.

Put plainly: for every hundred thousand shares people asked for, eighteen were handed out. When a float that small meets demand that large, the opening print is not a valuation. It is an auction among the few people who got any.

“Whether a company thrives cannot rely solely on short-term enthusiasm. For the industry, the real long-term test is whether the capital feast can be transformed into technological breakthroughs and commercial implementation.” Chen Jing, Technology and Strategy Research Institute

Chen also made the bullish case, and it is the one Beijing will be repeating: “Unitree’s capitalization path has proven that China’s hard-tech route is viable.” The company is the first humanoid robot maker to list on a mainland Chinese exchange. Founder Wang Xingxing, who holds around 121.4 million shares, ended the day worth roughly 107 billion yuan — about $15.8 billion.

The Comparison Nobody in Silicon Valley Wants

Unitree is not a speculative pre-revenue robotics startup that got lucky with a listing window. By shipment volume it is the largest humanoid maker on earth after one competitor. Market researcher Smart Analytics Global put its first-half 2026 humanoid shipments at about 5,900 units — up 170% year on year — giving it 31% of the global market and second place worldwide. Cumulatively the company says it has built around 18,000 humanoid robots.

Hold that against the American field. Agility Robotics, one of the most advanced US humanoid companies, is going public via SPAC at a pre-money valuation of $2.5 billion and an implied value near $4 billion. It is unprofitable, does not publicly disclose revenue, and points to more than $300 million in committed multi-year orders as evidence of traction.

So the scoreboard reads: the Chinese company ships thousands of units a year, earns a 60% gross margin, turns a profit, and is valued at $50.7 billion. The American company has a strong order book, no disclosed revenue, and is valued at about $4 billion. One of those two numbers is wrong, and the interesting argument is over which.

What Actually Has To Happen Next

A 210× sales multiple is a bet that this is 1999 for robots — that unit volumes go up by two orders of magnitude and the current revenue line is noise. For that bet to pay, several specific things have to be true, and none of them are yet.

  • A job, not a demo. Sprint records and kung-fu routines have carried Unitree’s brand for two years. The revenue that justifies this valuation comes from humanoids doing repetitive industrial work reliably for years, and that deployment data does not exist publicly for anyone.
  • Margin survival at scale. A 60% gross margin on 18,000 units is a research-and-prototype margin. Mass production usually compresses it hard.
  • Export access. Unitree’s growth story assumes global markets. Its robots are already the subject of security scrutiny in Western countries, and the geopolitics of Chinese hardware in warehouses and public buildings is not settled.
  • Float normalisation. Only 10% of the company trades. Lock-up expiries will introduce a supply of shares the opening auction never had to absorb.

What To Watch

  • Where the price settles in 90 days. The open was a liquidity event; the level a month or a quarter out is the market’s actual opinion.
  • Second-half shipment numbers. 5,900 units at +170% is the single most important operating figure Unitree publishes. Watch whether that rate holds.
  • Superman’s price and ship date. Neither has been disclosed. A record-setting robot nobody can buy is marketing; a record-setting robot with a price is a product.
  • Who follows it public. Unitree is the first humanoid maker on a mainland exchange, and it just showed every Chinese robotics firm what the listing window looks like.
  • The American response. A profitable competitor at 12× Agility’s valuation changes how US robotics companies raise, price and pitch.

The Takeaway

Unitree unveiled a humanoid on Monday that runs at 12.66 metres per second — faster than the 12.42 m/s peak Usain Bolt hit in his world-record 100 metres — and jumps 2 metres from standing, built in a little over three months. On Wednesday the company listed in Shanghai, opened up 629%, and closed up 460% at a $50.7 billion valuation.

The underlying business did $240 million of revenue in 2025 at a 60% gross margin and about $41 million of profit. That is a genuinely good hardware company. It is also roughly 210 times sales, in a company where 9.8 million retail accounts chased 9.7 million shares and 0.018% of the orders were filled.

The number to keep is not $50.7 billion. It is 5,900 — the humanoid robots Unitree shipped in the first half of this year. Every dollar of that valuation is a wager on how fast that number grows. The robot can already outrun Usain Bolt. The company now has to outrun a multiple.