Unitree is the first humanoid-robot maker to go public with a profit. At 36 times sales on a 10 percent float, the profit is the least interesting number.

When Unitree Robotics finished pricing its Shanghai listing this week, the number that travelled was the valuation: around 61 billion yuan, which the wire services rounded, depending on the day's exchange rate, to somewhere between eight and a half and nine billion dollars. Everyone will repeat it, and almost no one will interrogate it, which is the usual fate of a headline number with a lot of zeros. So let me do the boring thing and take it apart, because the valuation is not the interesting figure here. Two others are: the size of the slice Unitree actually sold, and the size of the business that slice is priced against. Put those next to the headline and the listing stops looking like a verdict on humanoid robots and starts looking like what it is — a very good company wrapped around a very demanding trade.

The facts first, because they matter and several early write-ups blurred them. Unitree priced its shares on the Shanghai Stock Exchange's STAR Market — China's Nasdaq-style board for technology names — at 150.8 yuan each. It is selling 40.45 million new shares, which come to about 10 percent of the enlarged company. That raises roughly 6.1 billion yuan, call it 850 million dollars to 900 million depending on where you take the rate; the round "904 million" you will see quoted is one bank's conversion of a yuan figure, not a number Unitree raised in dollars. Subscriptions open on the 10th of August, payment falls due on the 12th, and the shares are expected to begin trading in the days after. The founder, Wang Xingxing, and parties tied to him keep about 31 percent of the equity and, through the share structure, roughly 65 percent of the voting power. Hold that last pair of numbers; we will need them.

The company is not the argument

Let me be clear about what Unitree is, because the skeptical case here is not the lazy one that dismisses the whole thing. This is, by some distance, the most real humanoid-robot business on either side of the Pacific. In 2025 it shipped more humanoid robots than any other company on earth — north of five thousand of them — and its cumulative sales of four-legged robots, the machines that made its name, have passed thirty thousand units. Revenue last year was about 1.7 billion yuan, up more than fourfold. Net profit was roughly 288 million yuan, more than doubled. Gross margin on the core business runs above 60 percent. More than 40 percent of sales come from outside China. Those are not the numbers of a science project. They are the numbers of a real manufacturer with real customers and, rarest of all in this category, an actual profit. Unitree is going public as the first humanoid-robot maker in the world that makes money, and that claim is true.

It is also, for the trade, almost beside the point. Because a profit of 288 million yuan against a valuation of 61 billion yuan is a price-to-earnings multiple north of 200, and a revenue of 1.7 billion yuan against that same valuation is a price of roughly 36 times sales. Those are not the multiples of a manufacturer. They are the multiples of a story, and the story is not the robot dog that most of Unitree's revenue actually comes from today. The story is the humanoid — the labour-replacing machine that walks into a warehouse or a home and does the work of a person — and the price assumes that market arrives, that Unitree wins a large share of it, and that it does both fast enough to grow into a valuation that today's income statement comes nowhere close to supporting. The company you are buying makes quadruped robots for research labs and light-industrial use at a healthy margin. The company the price describes replaces human labour at global scale. They share a name and a founder. They are not yet the same business.

The company you are buying makes robot dogs at a healthy margin. The company the price describes replaces human labour at global scale. They share a name. They are not yet the same business.

How the money actually moves

Now the mechanism, because a valuation is not a fact about a company; it is an artefact of how the shares get priced and who is allowed to buy them. Unitree floated about 10 percent of itself. That is a thin free float by any standard, and on the STAR Market, where retail participation is heavy and the first days of trading can run hot, a thin float does something specific: it lets a relatively small amount of enthusiastic money set the price for the whole company. When only a tenth of the shares trade and a wall of domestic demand meets them, the clearing price reflects the most optimistic buyers, not the median one — and that price is then multiplied across every share, including the 90 percent that never came to market, to produce the 61-billion-yuan headline. The valuation is real in the sense that it is what the marginal share costs. It is not real in the sense that anyone could sell the whole company, or even a large piece of it, at that level. Concentration cuts both ways, and a thin float has no brakes on the way down.

This is where the two numbers I asked you to hold come back. Wang keeps about 31 percent of the equity but 65 percent of the votes. So the public is buying a small, richly priced minority of a company that its founder will continue to control outright regardless of what outside shareholders think. That is a common structure for a founder-led technology firm, and it is not sinister. But it changes what the buyer is actually getting: not a claim on the direction of the business, only a claim on its economics, at a price set by the most eager 10 percent of the shares. If the humanoid market takes longer than the multiple assumes — and every humanoid timeline in this industry has so far taken longer than promised — the outside holders are the ones who wear the repricing, and they cannot vote for a change of course while they wait.

The cheapest expensive thing in the room

The bullish rejoinder is comparison, and it is a fair one as far as it goes. Look west, the argument runs, and Unitree is a bargain. Figure AI, the American humanoid startup, has been valued in private rounds at something like 39 billion dollars while booking essentially no revenue at all. Against a competitor priced at forty-odd billion dollars on a story with no income statement, a profitable Chinese manufacturer at eight or nine billion looks almost sober. And it is true: if you are going to pay for the humanoid dream, paying for the version that already ships product and turns a profit is the more defensible way to do it. Unitree is the cheapest expensive thing in the room.

But being the least extreme valuation in a category of extreme valuations is not the same as being cheap, and it is worth not letting the comparison do your thinking. Thirty-six times sales is a heroic multiple whether or not someone across the ocean is paying more for less. The relevant question is not "cheaper than Figure?" It is "what has to be true for this price to be the right one?" — and the answer is the same demanding list regardless of what Figure is worth: a humanoid market that materialises on schedule, a Unitree that captures a large slice of it, margins that survive the competition rushing in behind, and a Chinese champion that does all of this while cut off from parts of the West. Each of those is plausible. All of them together, on the timeline the price implies, is a bet, and dressing it in a favourable comparison does not make it less of one.

The downside no one is pricing

Which brings us to the exposure the listing is quietest about. Days before Unitree formalised its IPO schedule, the United States moved to restrict imports of new foreign-made robots on national-security grounds — a measure written broadly but widely read as aimed at China, and at the sector's largest player in particular. Unitree drew a bit over 13 percent of its revenue from the United States in the most recent year, already down from closer to 19 percent as it diversified into Europe and Southeast Asia. So the immediate hit to the top line is real but survivable. The larger exposure is not the 13 percent of sales; it is the fact that the country doing the shutting-out is also the country most likely to be an early, high-value market for exactly the labour-replacing humanoids the valuation is built on. The price assumes a global humanoid market. The politics are quietly carving that market into a piece Unitree can serve and a piece it may not, and the multiple has not obviously marked down for the wall going up.

• What the headline says: Unitree priced at roughly 61 billion yuan (about $8.5–$9 billion), the first profitable humanoid-robot maker to go public.
• What the price assumes: a humanoid-labour market that arrives on schedule, a dominant Unitree share, durable margins, and few walls between it and its end markets.
• What the buyer actually gets: a richly priced 10 percent float of a founder-controlled company, valued at 36 times sales, whose revenue today is mostly four-legged robots, not humanoids.
• The exposure that is underpriced: the largest likely humanoid market, the United States, is moving to shut Chinese robots out.

None of this is a call on the stock, and it is emphatically not a claim that Unitree is a bad company — it is, on the evidence, the best real business in its field, which is precisely why the trade is seductive. A great company and a dangerous position can be the same instrument at the same time, and this listing is shaping up to be both. The thing to watch after it starts trading is not the valuation, which the thin float will make volatile and which everyone will quote at you anyway. It is the gap between the two companies inside this one: the profitable maker of robot dogs that exists now, and the humanoid champion the price insists is coming. For as long as that gap stays open, 61 billion yuan is not a measure of what Unitree is. It is a wager on how fast it can become something else — and the people who bought the top 10 percent are the ones holding the ticket.