Unitree surged 629% on debut, then fell 11%: the stock market got its ChatGPT moment first
Unitree rose more than sixfold intraday and closed about 460% higher before opening 11% lower the next day; DeepSeek had invested about $20.8 million and agreed to work on embodied intelligence. The stock proves enthusiasm, not mature robot commercialization.
The bottom line
Unitree Robot had just staged a "six-fold surge" on its first day of listing in Shanghai, but fell 11% the next day.
On August 19, Unitree Technology opened at about 1,100 yuan, an increase of more than 600% from the issue price; it still rose by about 460% at the close, and the company was valued at about 50 billion US dollars. By the next day, the stock price opened down 11%.
This is not proof that "robots can already do work", but a skit that simultaneously ignites AI, domestic robots, DeepSeek's investment in stocks, and capital market sentiment.
What’s even more dramatic is that Unitree founder Wang Xingxing did not continue to shout “immediately replace humans” after the surge. Instead, he said: Robots are still two to three years away from the real ChatGPT moment, and it may take two or three years to be optimistic, and it may be five to ten years to be conservative.
The stock market has already celebrated the robot, but the robot itself said that it has not graduated yet.
In one day, Unitree rode a capital-markets roller coaster
Unitree Technology is one of the most well-known manufacturers of robot dogs and humanoid robots in China. It relies on robots that can dance, run, box and perform martial arts movements to make its mark. Because of its price and mass production capabilities, it is regarded as one of the first companies in the domestic robot industry to come into the spotlight.
On the first day of listing, market sentiment almost went out of control:
- The opening price was about 1,100 yuan, an increase of more than six times from the issue price.
- The closing price was about 845 yuan, still roughly 460% above the issue price.
- The company was valued at approximately $50 billion at closing.
- It opened the next day down about 11%.
Looking at this set of numbers together, it is more interesting than saying "a surge of 629%" alone: The first day is like an IPO myth, and the next day you are immediately reminded that this is a stock, not a product launch.
This is not an ordinary robotics company
Unitree's story is explosive not just because it makes robots dance.
It hits the three hottest lines at the same time:
The first is a humanoid robot. Chinese robotics companies are vying for factories, logistics and home scenarios, and the capital market sees them as the next AI interface after large models.
The second is the domestic supply chain. According to Reuters, Unitree is one of the world's largest robot dog manufacturers and the top humanoid robot manufacturer in terms of shipments; however, there are still very few robots that have truly entered the commercial environment and continue to complete tasks for people.
The third one is DeepSeek.
DeepSeek didn’t come to “visit”, but directly invested $20.8 million
On August 6, exchange documents showed that DeepSeek invested approximately 140.8 million yuan, or approximately US$20.8 million, in Unitree’s strategic placement, acquiring 933,399 shares, accounting for approximately 2.31% of the IPO strategic placement shares. The two parties also signed a cooperation arrangement.
The direction of cooperation is very straightforward:
- DeepSeek provides model capabilities.
- Unitree provides mechanical engineering, motion control and real physical world data.
- Both parties jointly explore AI models and embodied intelligence for humanoid robots.
This explains why this IPO is not just "a robotics company going public" but rather like DeepSeek is looking for a body for its own model.
A large model can write code and do math problems, but that doesn't mean it knows how soft a towel is, or that it can judge whether a cup will slip in a strange room. If robots want to enter the real world, what they lack is not another list, but massive action data, failure records and safety feedback.
Wang Xingxing himself first poured cold water on the "ChatGPT moment"
On August 20, Wang Xingxing said at the World Robot Conference in Beijing that the robot industry is approaching the "ChatGPT moment" of embodied intelligence: robots will one day be able to enter strange homes and complete about 80% of daily tasks through voice or text commands.
But he also acknowledged that robotics software still lags behind hardware. The real great leap is optimistically estimated to take two to three years, and may take five to ten years at the latest.
This passage is very crucial.
Because today's robots can already complete a set of beautiful actions on the stage, there is a whole gap between them and "leaving the housework to it":
- Can it understand vague instructions?
- Can it reroute furniture after it has been moved?
- Will it continue to perform the wrong action after knocking over something?
- Can it work for hours on end instead of demonstrating for three minutes?
- Who bears responsibility when something goes wrong?
Being able to do somersaults is a hardware performance; being able to stably complete 80% of the tasks in a strange family is a product revolution.
Capital rushed in first, but commercialization is still waiting in line
Unitree is already profitable, which is more solid than many robotics startups. However, Reuters also reminded that a large number of robot sales in the industry are still one-time purchases, and the number that is actually deployed in the business environment and creates long-term value is not large.
This is also the most alarming aspect of this surge in stock prices: the market is pricing in advance "what may happen in the next ten years."
If humanoid robots truly become the next entry point into general computing, $50 billion may even be just the starting point; if robots remain at exhibitions, factory pilots, and short video performances, today’s valuations will appear very aggressive.
Moreover, Unitree also faces realistic geographical restrictions. Reuters reports that the U.S. FCC has banned future imports of foreign-made humanoid and quadruped robots in July, which has affected Unitree's future products' entry into the U.S. market.
What does this IPO mean?
My judgment is: It proves that capital has regarded embodied intelligence as the main battlefield of AI, but it does not prove that robots have crossed the commercialization threshold.
DeepSeek's investment shows that model companies are looking for a training ground in the physical world; Unitree's listing shows that robotics companies are beginning to compete for long-term funds and industrial resources; the decline the next day reminds everyone that demonstration capabilities, laboratory capabilities and large-scale income are still three different things.
China's robotics industry may really be approaching an inflection point, but this inflection point may not necessarily occur on the stock price curve.
The real "ChatGPT moment" should look like this: Ordinary people don't need to read tutorials, the robot can enter an unfamiliar environment, understand an incomplete sentence, complete the task, and stop when unexpected, and the cost is more reasonable than hiring a human.
It may be two or three years away from that day, maybe more.
What to watch
- Whether Unitree's post-IPO valuation can be supported by continued orders and recurring revenue.
- Will DeepSeek and Unitree collaborate to release real embodied models or robot products?
- Can robots be piloted from factories into high-frequency scenarios such as homes, hotels, logistics, etc.
- Will the growth in China’s humanoid robot shipments lead to an upgrade in reliability and safety standards?
- After the U.S. market is restricted, can Unitree achieve scale in China and other overseas markets.
The last sentence:
Unitree ran up sixfold in the market on day one. Its robots may need six years of real-world work before the market knows whether this is the next AI revolution.
Why this matters to you
If you are just watching the excitement, remember not to translate "the listing surge" into "the robot has matured". If you are paying attention to investment, employment or industry opportunities, what you should really track is the robot’s repeated orders, continuous working hours, failure rate, deployment costs and after-sales maintenance, not how beautiful the movements at the press conference are.