On this page
The story you've been told for two years is very tidy: humanoid robots are about to walk into factories, replace labor, and whoever owns the smartest robot owns the future. On August 19, 2026, China's stock market seemed to confirm that story with an enormous number.
That story is wrong in the place that matters most. Not wrong that robots are coming — they are. Wrong because it makes you look at the robot's intelligence, when what actually just changed is its price. And if you watch the wrong variable, you'll draw the wrong lesson for your own business.
Read the balance sheet, not the headline.
What happened on August 19
Unitree Robotics — based in Hangzhou, founded in 2016 — listed on the Shanghai Stock Exchange's STAR Market. It's the first pure-play humanoid robot maker to go public in mainland China.
According to the South China Morning Post, the first session went like this:
| Point | Share price | Market cap |
|---|---|---|
| IPO price | 150.80 yuan | |
| Opening price | 1,100 yuan (+629%) | |
| Closing price | 845 yuan (+460%) |
The company sold 40.45 million shares — 10% of post-offering capital — raising about 6.1 billion yuan. Founder Wang Xingxing, 36, holds roughly 121.4 million shares, worth about 103 billion yuan at the close. Meituan, a major shareholder with 8.7% after the IPO, is sitting on roughly 70 times its original investment.
Demand beforehand was off the charts too. Bloomberg reported on August 10 that the retail tranche was 5,526 times oversubscribed.
That's the news part. Now for the part almost no outlet put in a headline.
The number almost nobody read: 73.6% of revenue comes from schools
Unitree isn't a cash-burning startup. According to the Shanghai Stock Exchange's official release, revenue went from 159.13 million yuan (2023) to 392.77 million yuan (2024) to 1.70 billion yuan in 2025 — nearly 11x in two years. Net profit in 2025 was 278.21 million yuan; excluding non-recurring items, profit attributable to the parent was 590.75 million yuan. Overseas revenue accounted for 43.65%.
This is a real business with real profits. That's rare in this industry and deserves credit.
But where the money comes from is the question that matters. An analysis by 36Kr published March 23, 2026, based on the first three quarters of 2025 in the IPO filing, broke revenue down like this:
- 73.6% — scientific research and education
- 17.39% — consumer/commercial
- ~9% — actual industrial applications
And within that thin 9%, the share going to manufacturing, inspection and logistics — robots actually doing work — is less than a third.
Let that sink in. The market just put a $48–66 billion price tag on a company that earns nearly three-quarters of its revenue from universities and research institutes buying lab equipment. 36Kr describes the remaining business customers bluntly: most buy robots to display in showrooms, not to raise productivity.
That's not an accusation of fraud. Unitree doesn't hide any of this — it's in the IPO filing. The issue is the gap between the story the valuation is telling (robots replacing labor worldwide) and the story the financials are telling (the world's best supplier of research hardware).
Do the simple division: 342 billion yuan of market cap over 1.70 billion yuan of revenue is roughly 200x revenue. Over the adjusted profit of 590.75 million yuan, it's nearly 580x earnings. At the opening price, those multiples are more than 30% higher.
One telling signal from the same day: per SCMP, during the August 19 session the STAR Market Composite index fell 7.2% and the Shanghai Composite fell 2.4%. Unitree's debut didn't lift the robotics sector — money piled into exactly one ticker. That's the signature of a concentrated frenzy, not a repricing of the whole industry.
This is the real news: prices fell 72% in two years
If you take one fact away from this piece, make it this one.
According to Unitree's IPO filing, the average selling price of its humanoid
robots fell from about 593,400 yuan ($85,000) in 2023 to 167,600 yuan ($25,000)
in 2025 — down nearly 72% in two years. The product line makes it even
clearer:
- 2023 — H1: from 650,000 yuan
- 2024 — G1: from 99,000 yuan
- July 25, 2025 — R1: from 39,999 yuan, about $5,900 — 121cm tall, 25kg, 26 joints
From 650,000 yuan to 39,999 yuan in two years. That's a 94% drop at the low end of the range.
And here's the detail that makes it interesting: while prices collapsed, gross margin actually rose to 62.91% — per 36Kr, each humanoid robot still brings in more than 100,000 yuan of gross profit. For comparison, Tesla's gross margin over the same period was around 18% and Xiaomi's around 15%.
Prices down three-quarters, and profit per unit got fatter. That only happens when you make your own core components and control your own supply chain — exactly as Unitree describes itself in the IPO filing.
So the news on August 19 wasn't "robots are smart enough now." The news is: someone just turned the humanoid robot from an $85,000 instrument into a $5,900 commodity, without sacrificing margin. The 2025 robot isn't ten times smarter than the 2023 robot. It's sixteen times cheaper.
Why the market rewards the cost curve, not the intelligence curve
If this sounds familiar, it's because you've seen it before.
We've written about how Chinese AI models came to account for as much as 46% of US enterprise tokens — not because they're the smartest, but because they're 60–90% cheaper. The same mechanism is now repeating at the hardware layer.
The pattern goes like this: at every layer of the AI stack, the winner isn't whoever demos the most impressive thing, but whoever first collapses the cost of the whole category. Superior capability gets copied within months; a cost advantage compounds. Low prices unlock volume, volume creates scale, scale drives prices down again.
That's why investors pay 200x revenue for Unitree even though today's customers are universities. They're not buying today's revenue. They're betting that when a robot's price falls from $85,000 to $5,900, real-world applications will come looking for it — and whoever sits at the bottom of the cost curve at that point takes it all.
It's a logical bet. It could also be wrong — what keeps robots out of factories today may not be price, but reliability, safety and orchestration software. As one analyst told Al Jazeera on listing day, Unitree's main challenge is "turning hardware and manufacturing leadership into application and workflow leadership." That's an entirely different problem, and nobody has solved it yet.
The key point isn't whether the bet is right. It's that the market is pricing on cost, not on intelligence — and you should read every AI headline through that lens.
What this means for your business
You're not buying humanoid robots. But you are buying technology, and the pattern above applies directly.
First: don't pay a premium for capability that's about to become a commodity. Every time a vendor sells you "the most advanced technology" at the most advanced price, ask: how much cheaper will this capability get over the next 18 months? If the answer is "a lot" — as it is for most AI capabilities today — don't sign a multi-year lock-in. Choose an architecture that lets you swap vendors when the market reprices.
Second: tell a demo from a deployment. 73.6% of Unitree's revenue comes from customers buying robots for research and display. A lot of corporate AI projects are the same: a pretty chatbot on the homepage, an "AI-powered" feature in the pitch deck, and nothing changes in the business numbers. The test is simple — if you switched this feature off tomorrow, would any company metric get worse? If not, you own a showpiece.
Third: your advantage isn't the tool. If tool prices fall 94% in two years, the tool can't be your moat. Your competitor can buy the exact same thing, cheaper, next year. What doesn't get cheaper over time is the customer data you accumulate, the processes you design, and the brand trust you build.
That's the principle I apply when integrating AI into websites and workflows: pick technology for the problem, not the headline; design it to be replaceable; and measure it by business results, not by how impressive the demo looks.
On August 19, the world saw a $66 billion robot company and concluded the robots have arrived. The balance sheet tells a much calmer story: what arrived isn't a robot that can do the job, but a price that makes trying them cheap. That's still big news — just big in a different way than the headlines say.
Sources
- Unitree Robotics surges 629% to US$66 billion valuation in Shanghai share debut — South China Morning Post, August 19, 2026
- China's Unitree soars in market debut as investors bet on humanoid robots — Al Jazeera, August 19, 2026
- Unitree IPO draws spotlight to China's fast-growing humanoid robot sector — Shanghai Stock Exchange / China Daily, August 11, 2026
- Unitree Profits, but Humanoid Robots Fail to Make Money — 36Kr, March 23, 2026
- Unitree's Shanghai IPO 5,526 Times Subscribed by Retail Buyers — Bloomberg, August 10, 2026
- China's Unitree debuts US$5,900 humanoid robot in race to make cheaper products — South China Morning Post, July 25, 2025
