Has the robot race gone cold?
First, three big surprises.
1. Unitree Technology
On the first day of listing, it opened at 1,100 yuan. Now it is only 450.40 yuan. Its peak market cap was 444.9 billion yuan, and now it is about 182.2 billion yuan.
More than 260 billion yuan in market value has vanished.
2. Mech-Mind Robotics
Its Hong Kong public offering was oversubscribed 3,835 times, with 250,000 people competing for it.
But it broke issue price on the first day, and later once fell more than 20% below the offering price.
3. Huanchuang Technology
It surged 265.68% on the first day, then plunged nearly 48% on the second day.
This roller coaster meant that those who chased the rally mostly ended up at the top.
So, has robotics gone cold?
I don't think so.
The era when robots were valued by PPT stories and concepts is getting colder.
1⃣ Whole-machine performance:
Humanoid robots have indeed started shipping, but there is still a long way to go before they can truly replace human labor and generate stable profits.
Unitree's humanoid robot revenue in 2025 was 868 million yuan, accounting for about half of the company's revenue.
UBTECH delivered 921 full-size embodied AI humanoid robots in the first half of this year.
But what does 921 units really mean?
For a sector worth hundreds of billions of yuan in market cap, it is still far from true large-scale industrialization.
And although UBTECH is leading in commercialization progress, it is still operating at a loss.
Its cumulative losses over the past six years have exceeded 5 billion yuan.
Building a robot is easy; letting it demonstrate capabilities is one thing, but turning it into a stable source of cash flow is truly hard.
2⃣ Hong Kong IPOs: the market has already started cooling this story down in advance.
In September, a total of 12 new stocks listed in Hong Kong, 7 of which broke issue price on the first day, with a break rate close to 60%.
Huanchuang Technology rose 265.68% on day one, then nearly halved again.
The most typical example is Mech-Mind.
It went public carrying the title of "the first stock in embodied AI vision-brain-hand integration."
But if you look closely, you will find:
In 2025, 93% of its revenue came from intelligent robot guidance products, which in essence are still industrial 3D vision and related software.
The Mech-GPT and Mech-Hand products that truly represent the "brain" and "hand" together accounted for less than 1.1% of revenue.
This is not to say Mech-Mind has nothing going for it. Quite the opposite.
Revenue grew 54.7% in the first half of this year, newly signed orders grew 75.3%, gross margin was 65%, and orders on hand doubled year on year.
But its current performance is still some distance away from the "embodied AI" story the market has been told.
That is exactly why the market has begun to reprice it.
3⃣ The robotics supply chain is still pushing forward
The real money makers are already shifting toward the "selling shovels" direction.
In the first half of this year, Inovance Technology reported revenue of 24.675 billion yuan and net profit attributable to shareholders of 2.81 billion yuan.
Laisai Technology's humanoid robot business has also started to see real orders:
Its frameless torque motors have more than 1 million units in hand orders, and planetary joints, harmonic joint modules, and dexterous hands have all begun batch supply, while securing orders from nearly 100 robotics customers.
SUTENG Innovation Technology also secured a nearly one-million-unit, five-year designated lidar order for a Toyota joint venture model.
These real, landed products are clearly different from backflip-style performances.
In the end, they all have to turn into orders, shipments, revenue, and profit.
At this stage, I actually think the robotics industry has entered a new phase.
👉 Before, everyone compared:
Whose humanoid robot was cooler.
Whose launch event was more explosive.
Who could tell the biggest story.
👉 Now, the comparison is starting to be:
Who can truly mass-produce.
Who can bring costs down.
Who can secure large customer orders.
Who can scale up component shipments.
Who can make factory owners do the math and realize:
Buying robots is cheaper than hiring people.
So, robotics has not gone cold. What has gone cold is the logic that says: as long as you slap on an embodied AI label, you should be worth hundreds of billions or even trillions.
The rite of passage for robots has never been a backflip.
It is the day when a boss is really willing to pay it a monthly salary.
🤖 Conclusion: what has gone cold is not robots, but the people blowing smoke.
First, three big surprises.
1. Unitree Technology
On the first day of listing, it opened at 1,100 yuan. Now it is only 450.40 yuan. Its peak market cap was 444.9 billion yuan, and now it is about 182.2 billion yuan.
More than 260 billion yuan in market value has vanished.
2. Mech-Mind Robotics
Its Hong Kong public offering was oversubscribed 3,835 times, with 250,000 people competing for it.
But it broke issue price on the first day, and later once fell more than 20% below the offering price.
3. Huanchuang Technology
It surged 265.68% on the first day, then plunged nearly 48% on the second day.
This roller coaster meant that those who chased the rally mostly ended up at the top.
So, has robotics gone cold?
I don't think so.
The era when robots were valued by PPT stories and concepts is getting colder.
1⃣ Whole-machine performance:
Humanoid robots have indeed started shipping, but there is still a long way to go before they can truly replace human labor and generate stable profits.
Unitree's humanoid robot revenue in 2025 was 868 million yuan, accounting for about half of the company's revenue.
UBTECH delivered 921 full-size embodied AI humanoid robots in the first half of this year.
But what does 921 units really mean?
For a sector worth hundreds of billions of yuan in market cap, it is still far from true large-scale industrialization.
And although UBTECH is leading in commercialization progress, it is still operating at a loss.
Its cumulative losses over the past six years have exceeded 5 billion yuan.
Building a robot is easy; letting it demonstrate capabilities is one thing, but turning it into a stable source of cash flow is truly hard.
2⃣ Hong Kong IPOs: the market has already started cooling this story down in advance.
In September, a total of 12 new stocks listed in Hong Kong, 7 of which broke issue price on the first day, with a break rate close to 60%.
Huanchuang Technology rose 265.68% on day one, then nearly halved again.
The most typical example is Mech-Mind.
It went public carrying the title of "the first stock in embodied AI vision-brain-hand integration."
But if you look closely, you will find:
In 2025, 93% of its revenue came from intelligent robot guidance products, which in essence are still industrial 3D vision and related software.
The Mech-GPT and Mech-Hand products that truly represent the "brain" and "hand" together accounted for less than 1.1% of revenue.
This is not to say Mech-Mind has nothing going for it. Quite the opposite.
Revenue grew 54.7% in the first half of this year, newly signed orders grew 75.3%, gross margin was 65%, and orders on hand doubled year on year.
But its current performance is still some distance away from the "embodied AI" story the market has been told.
That is exactly why the market has begun to reprice it.
3⃣ The robotics supply chain is still pushing forward
The real money makers are already shifting toward the "selling shovels" direction.
In the first half of this year, Inovance Technology reported revenue of 24.675 billion yuan and net profit attributable to shareholders of 2.81 billion yuan.
Laisai Technology's humanoid robot business has also started to see real orders:
Its frameless torque motors have more than 1 million units in hand orders, and planetary joints, harmonic joint modules, and dexterous hands have all begun batch supply, while securing orders from nearly 100 robotics customers.
SUTENG Innovation Technology also secured a nearly one-million-unit, five-year designated lidar order for a Toyota joint venture model.
These real, landed products are clearly different from backflip-style performances.
In the end, they all have to turn into orders, shipments, revenue, and profit.
At this stage, I actually think the robotics industry has entered a new phase.
👉 Before, everyone compared:
Whose humanoid robot was cooler.
Whose launch event was more explosive.
Who could tell the biggest story.
👉 Now, the comparison is starting to be:
Who can truly mass-produce.
Who can bring costs down.
Who can secure large customer orders.
Who can scale up component shipments.
Who can make factory owners do the math and realize:
Buying robots is cheaper than hiring people.
So, robotics has not gone cold. What has gone cold is the logic that says: as long as you slap on an embodied AI label, you should be worth hundreds of billions or even trillions.
The rite of passage for robots has never been a backflip.
It is the day when a boss is really willing to pay it a monthly salary.
🤖 Conclusion: what has gone cold is not robots, but the people blowing smoke.