Introduction: During this National Day holiday, robots appeared in official consumption data for the first time in a manner closer to that of "consumer goods".
On October 7, big business data released by the Ministry of Commerce (MOFCOM) showed that from October 1 to 6, sales of embodied intelligent robots on key platforms increased by 130% year-on-year. For reference, during the same period, sales of smart glasses grew by 100%, exoskeleton assistive devices by 12.8%, and outdoor sports equipment by 10.4%.
Looking solely at the 130% growth rate, it is easy to interpret this as just another beautiful story about robots.
However, what truly deserves the attention of the capital market is not that robots are selling better, but rather the hardest question surrounding this industry over the past two years—whether there is genuine demand for robots.
Judging from this National Day holiday, this question has begun to receive data validation from the consumer end for the first time.
Moreover, this is not an isolated National Day phenomenon.
MOFCOM previously disclosed that in July this year, sales of embodied intelligent robots on key platforms increased by 95.1% year-on-year. In other words, even before the National Day holiday, robot consumption had already shown signs of continuous growth.
This makes the humanoid robot market in 2026 start to look quite different from before.
In the past, investors discussed when Tesla would start mass production, which companies would enter the supply chain, how much a ball screw was worth, and how many reducers a single robot would use.
Now, a more realistic question has been placed on the table—if robots are truly starting to be bought, where exactly will the first batch of money flow?
The Sample of Beijing E-Town
Over the past few years, the humanoid robot industry has never lacked demonstrations.
Robots can run, do somersaults, dance, and practice martial arts, and they are increasingly attracting crowds at exhibitions.
China has become one of the most concentrated markets for humanoid robot products globally. The Ministry of Industry and Information Technology (MIIT) disclosed in July this year that in the first half of the year, the number of complete humanoid robot products in China had reached over 400, accounting for more than half of the global total.
The quantity is growing rapidly.
However, "making" and "selling" have never been the same thing.
For a new hardware industry, the truly difficult parts usually occur after the laboratory stage.
Technical personnel need to solve reliability issues, factories need to solve yield rate problems, suppliers need to reduce costs, and end-users still need to answer a simpler question: Why do I need it?
This is also the valuation paradox that humanoid robots have long faced in the past.
The capital market can price the robot supply chain based on the market space several or even more than ten years later, but today's financial statements of enterprises can hardly immediately provide revenues that match such valuations.
The fact that a robot can run a marathon does not mean someone is willing to pay hundreds of thousands of CNY for it.
The fact that a robot can fold clothes does not mean it is already more economical than a traditional robotic arm or a human worker.
Therefore, the most important change in the robot industry this year may not be how much motion control has advanced, but that more and more robots are starting to enter a previously unfamiliar place: the cashier counter.
Beijing E-Town is a good microcosm.
During the first E-Town Robot Consumption Festival in 2025, sales of robots and related products exceeded 190,000 units, with total sales surpassing CNY 330 million; robot consumption vouchers directly leveraged sales of over CNY 70 million. Relevant disclosed data also showed that within 16 days of operation, the 4S store for embodied intelligent robots saw multiple million-level orders, with orders over CNY 1 million accounting for 25%.
This year, it continues to expand.
In 2026, the special fund for promoting consumption at the E-Town Robot Consumption Festival increased from CNY 15 million to CNY 18 million. A new 9,600-square-meter robot consumption block was added, and over 50 robot experience devices covering education, elderly care, and home scenarios were deployed.
Behind this is actually a very typical logic for cultivating new consumer goods: first let consumers see it, then let them experience it, and finally make a purchase.
New energy vehicles (NEVs) have gone through a similar process, and so have smartphones.
It is just that for robots, this process is much more difficult.
Because today, the buyers of robots are still largely enterprises, schools, research institutions, scenic spots, shopping malls, and government scenarios, rather than ordinary households.
Even if the so-called "robot consumption" is growing rapidly, one must pay attention to an easily overlooked statistical caliber issue: what MOFCOM disclosed is a 130% increase in sales of "embodied intelligent robots," which is not equivalent to "a 130% increase in humanoid robot sales," and certainly does not mean that humanoid robots costing hundreds of thousands of CNY have begun to enter households on a large scale like smartphones and cars.
In fact, what is truly moving faster at present are still those robots with clear tasks and relatively standardized scenarios.
For example, Xinhua News Agency reported during the National Day holiday that a wheeled humanoid robot store in a cultural and creative park in Hefei can independently complete product identification, grasping, and delivery. During the holiday, the daily average number of orders was about 100, with a daily turnover of approximately CNY 1,500.
This number is not large, but it may be more important than a robot backflip, because it means that robots are starting to shift from "technical demonstrations" to "production tools," and they are beginning to generate revenue.
For the robot industry, what lies between these two is the true commercialization chasm.
The "Disenchantment" of Humanoid Robots
When a new hardware industry begins to scale up, the most common mistake the capital market makes is treating all supply chain companies as the same story.
In fact, from smartphones to NEVs, the companies that truly make stable profits in the early stages of industrial development are often not necessarily the most watched terminal brands.
Sometimes they are chip companies, sometimes battery companies, and sometimes connector, optical component, or precision manufacturers.
Robots are very likely to be the same.
The reason is not complicated.
Today's complete humanoid robots are still in a stage of rapid iteration.
The appearance is changing, the number of joints is changing, the transmission solutions are changing, the dexterous hand solutions are also changing, and even the question of "whether it must be made in a humanoid form" itself has not yet reached a final answer.
At this stage, it is still difficult to judge the winner among complete machine brands.
But some demands have become relatively certain.
If robots need to move, they need actuators; if they need high-precision control, they need reducers, ball screws, and motors; if they need to interact with the real world, they need visual, force, and tactile sensors; if they want to truly enter factories, they must solve issues of lifespan, cost, and reliability.
This is also why, as the robot market has developed to today, what is increasingly worth looking at is no longer whether a certain company has announced "deploying humanoid robots," but rather three more specific questions: Have they entered the supply chain? Are they making batch deliveries? How much money has the robot business actually contributed?
Leaderdrive is a sample of this change.
The company has achieved batch deliveries to top international robot customers, with overseas revenue increasing by 78% year-on-year in the first half of the year.
This is a different stage from simply obtaining "sample submission," "designated supplier status," or signing strategic cooperation agreements.
Because after reaching batch delivery, the logic of a robot component enterprise truly shifts from thematic investment to volume investment: for every additional complete machine produced, it means more demand for reducers, motors, sensors, or actuators.
The more robots are sold, the more obvious this economies of scale effect becomes.
Ecovacs provides another sample worth referencing.
Although robot vacuums and service robots are not equivalent to the most popular humanoid robots today, it at least proves one thing—only after completing product definition, cost control, and channel deployment can robots become consumer electronics in the true sense.
According to Ecovacs' 2026 semi-annual report, the global shipment of Ecovacs brand service robots increased by 44% year-on-year, and the proportion of overseas revenue exceeded 50% for the first time in the second quarter.
The difference between such figures and "robots can dance" is the business model.
When a product truly goes to market, the ultimate questions to answer are not technical parameters, but: How many people buy it? How much does it sell for? What is the gross profit margin? How often is it replaced? Do consumers actually use it after buying it?
And it is precisely here that an increasingly obvious divergence will occur within humanoid robot concept stocks.
Hongxin Technology is a very typical negative observation sample.
The company has signed a framework agreement for component procurement with a leading domestic embodied intelligent robot company and has received small-batch procurement orders for humanoid robot structural parts.
If looking only at this sentence, it almost possesses all the elements that make robot concept stocks most popular in the market: top customers, embodied intelligence, humanoid robots, and orders.
But continuing to look at the financial data below, the story will quickly calm down.
The company disclosed that the relevant business only began to generate sales revenue in June 2026. In the first half of the year, robot component revenue was only CNY 164,000, accounting for 0.03% of operating revenue during the same period. The company itself also explicitly warned that the business is still in its infancy, and there is significant uncertainty regarding the pace of customer designation and mass production.
CNY 164,000—this figure may be one of the most valuable data points for understanding the current robot market.
It tells investors one thing: there is still a long way to go between entering the robot supply chain and robots truly becoming a source of profit.
And this distance is very likely to become the biggest valuation watershed in the robot sector in the next stage.
Previously, the market was willing to give a high premium to "0 to 1."
As long as a company has robot technology reserves or obtains customer validation, the market will start trading on future large-scale mass production.
But when the entire industry truly begins to enter the scale of tens of thousands of units or even larger, the evaluation criteria will change: "whether there is" will gradually become "how much there is"; samples will become orders; orders will become deliveries; and deliveries must ultimately become revenue and profit.
By then, whether a company's robot business accounts for 0.03%, 3%, or 30% of its revenue, the valuation system will be completely different.
Countdown to the "Ten-Thousand-Unit Level" Landing
In June this year, the MIIT and the State-owned Assets Supervision and Administration Commission of the State Council (SASAC) jointly launched a special action for real-scene practical training of humanoid robots and embodied intelligence.
The document proposed that by the end of 2026, key products such as humanoid robots will complete application verification and regular deployment in a batch of representative scenarios, and condense and form more than 100 high-value application scenarios, driving the formation of a landing capacity at the scale of ten thousand units.
"Ten-thousand-unit level" is a very important term.
Because from the perspective of the capital market, the truly dangerous stage for the robot industry may precisely not be the lack of mass production, but the beginning of mass production.
When there is no mass production, every company can talk about its own technical route.
Once mass production begins, price, cost, yield rate, reliability, and delivery capacity will start to speak.
This is an exam that all manufacturing industries ultimately cannot escape.
CCTV News, citing the "2026 Humanoid Robot Industry Development Report," stated that in the first half of this year, China's humanoid robot shipments exceeded 40,000 units, with the global share increasing to 97% (the statistical caliber of the industry report includes products such as wheeled humanoids, and there are differences in the statistical scope of different institutions). In 2025, according to market research data released by the China Institute of Electronics (CIE) and others, China's humanoid robot shipments were about 14,400 units.
Regardless of how large the differences in product caliber are among different statistical institutions, one direction is already quite clear: the number of robots is increasing rapidly.
The next real question is where these robots are being sent.
If most robots still remain in laboratories, schools, exhibition halls, and data collection sites, then the industry is still mainly solving the problem of "how many to produce."
Only when robots begin to continuously enter automobile factories, 3C production lines, warehouses, shopping malls, elderly care institutions, and even households, will the industry truly enter the stage of "creating value."
This is also why the MIIT, in the latest special action, focuses on real scenarios such as production manufacturing, testing and analysis, maintenance, warehousing and logistics, catering and retail, medical and elderly care, work safety, and emergency rescue.
The policy goals have actually undergone a subtle but important change.
In the past, the emphasis was on what robots can do; now, the emphasis is on whether robots can do it in the long term.
This means that the robot industry is moving from an engineering problem to an economics problem.
Assuming a robot is sold for hundreds of thousands of CNY and can work continuously for several years, is it cheaper than manual labor? What is the maintenance frequency? How often does the battery need to be replaced? If a failure occurs in a complex scenario, will it cause the entire production line to halt? For the same action, is the success rate 99% or 99.99%?
These questions, which may not seem as eye-catching as doing somersaults, ultimately determine whether robots have real orders.
They also determine the next round of reshuffling in the A-share robot supply chain.
Over the past two years, the two most valuable words in the robot sector were "entering."
Entering the Tesla supply chain; entering top robot manufacturers; entering sample submission; entering validation.
But in the coming years, the two most valuable words may become: "scaling up."
Because only by scaling up can the profit assumptions behind market capitalizations of billions or tens of billions of CNY be cashed in.
This is also why the significance of "a 130% increase in sales of embodied intelligent robots" during the National Day holiday cannot be understood merely as a holiday consumption data point; it is more like a signal for the switching of the valuation system of the entire industry.
In the past, the market bought technical possibilities; next, the market will start to buy commercial results.
In this process, the robot sector will not become easier to invest in because of increased sales; instead, it may become more difficult.
Complete machine enterprises need to prove that their products are continuously used by people, not just purchased for display; component enterprises need to prove that their orders can grow from hundreds of sets to tens of thousands of sets; traditional manufacturing enterprises must prove that robots are not just a new story accounting for 0.03% of revenue in the annual report.
The companies that ultimately survive are very likely not those that release the most robots, nor necessarily those with the sexiest stories in the capital market.
Rather, they are the enterprises that can achieve three things: robots are truly sold out; customers truly buy repeatedly; and for every additional robot sold, the company truly makes money.
During the 2026 National Day holiday, robots seem to have finally taken the first step—people are starting to buy robots.
But for the capital market, the real story is just beginning now.
Because the next question to answer is: Who has sold the robots, and who has only sold a story about robots.