In the first half of 2025, almost every humanoid robot company will showcase their muscles: punching, dancing, running, folding clothes, wiping tables… Various demos (demonstration videos) will appear one after another, becoming the latest synonym for high technology in the eyes of the public.
Entering the second half of the year, the wind direction has changed. Demo gradually fades out, and orders become the new focus.
In the past two months alone, multiple companies have successively entered the “Thousand Platform Club”. According to incomplete statistics from the Economic Observer, 8 companies including Ubiquitous, Yushu Technology, Zhiyuan Robotics, Songyan Power, Xingchen Intelligence, Zhifang, Zhongqing Robotics, and Accelerated Evolution have all disclosed orders for thousands of units.
According to publicly available data, the selling price of educational and exhibition humanoid robots is generally between 20000 and 100000 yuan, while industrial products are generally priced at 200000 yuan or more. Therefore, thousands of orders often correspond to contracts worth billions of yuan.
For example, in early October, Zhiyuan Robotics signed a cooperation agreement worth hundreds of millions of yuan with Longqi Technology (603341. SH), planning to deploy nearly a thousand robots. Since September, Uber has won the bid for a 250 million yuan project; The collaboration between Xingchen Intelligent Disclosure and Xiangong Intelligent at the level of thousands of units; Zhifang and Huike signed a 500 million yuan agreement to deliver over 1000 units within 3 years; Zhongqing Robotics has signed a cooperation agreement with Duolun Technology (603528. SH), and Duolun Technology plans to purchase no less than 2000 units within 3 years. Earlier, Songyan Power announced that it had received orders for over 2500 units, with a contract amount exceeding 100 million yuan.
This wave of signing orders is the result of multiple factors combined. Since its debut on this year’s Spring Festival Gala, humanoid robots have frequently appeared in the public eye, quickly establishing a cognitive foundation. At the same time, the actual demands of industries such as manufacturing, education, and cultural tourism are emerging, creating a batch of scenarios that can be implemented in bulk. In the context of tight financing, many companies urgently need “real sales volume” to leverage the next round of financing and valuation.
Thousands of orders are becoming a watershed for the industrialization of humanoid robots. It means that robots are no longer just demonstration prototypes in the laboratory, but are beginning to enter real industrial scenarios.
At this stage, it can be compared to the situation ten years ago when new energy vehicles were first introduced into logistics, ride hailing, and government demonstration projects – verifying their usability first, and then considering their popularization. However, for humanoid robots to truly enter households and become mass consumer products, they still need to go through product optimization, cost reduction, and the establishment of user habits.
When humanoid robots take the first step towards industrialization, some key issues also emerge: who is buying, what is buying for; Who is selling and how can it really be sold?
A sales war shifting from “demo” to “delivery” is quietly rewriting the business logic of humanoid robots.
Major clients from the manufacturing and education industries
Behind the current wave of “signing orders”, there are not many customers who truly have the ability to deploy thousands of units, and the manufacturing and education industries are the two outstanding performers.
Manufacturing demand is more rigid. Yao Maoqing, partner of Zhiyuan Robotics and president of the Tailored Business Department, said that manufacturing companies have the ability to undertake thousands of units, and the key lies in the common pain points in the industry: difficulty in recruiting workers, rising labor costs, insufficient automation of production lines, and sustained high pressure on production tasks. Once robots can replace human labor in positions with three shifts, night shifts, and repetitive labor, they have the feasibility of mass deployment.
In the past, factories generally adopted non-standard automation solutions. Once the product model is updated, the production line needs to be dismantled and rebuilt. Especially for OEM companies with complex SKUs and frequent product iterations, this model has high costs and poor flexibility. However, humanoid robots are more like “trainable operators” who can switch positions by updating models or task instructions, without the need to replace hardware. This flexible deployment method is the biggest advantage of humanoid robots compared to robotic arms.
In cooperation with shareholder company Longqi Technology, the Zhiyuan G2 robot is deployed on its PC and tablet production lines, and will be expanded to high iteration product lines such as mobile phones in the future.
From the perspective of cost structure, in regions such as the Yangtze River Delta, the annual comprehensive cost of a frontline worker is about 150000 yuan. If robots are used to replace manual labor in dual shift positions, the cost can be basically balanced in two years. With the continuous optimization of structural modules and core components, this account is close to being accounted for in the Chinese market and more attractive in the European and American markets.
Yao Maoqing said that the current return on investment (ROI) of humanoid robots is not as good as mature robotic arm systems, and the real value lies in filling the gaps that cannot be standardized and still require a large amount of manpower investment.
Manufacturing customers are not only the biggest orderers, but also the invisible “product managers” driving robot iteration.
Taking Stardust Intelligence as an example, founder Lai Jie told Economic Observer that in the past year, he led a team to visit hundreds of potential customers, from chemical laboratories to high-end manufacturing workshops, with extremely diverse scene demands: some hope that robots can replace chemical protective suits to enter experimental spaces, some propose automatic inspection of cell culture equipment every two hours, some teams test cross-border and inter provincial remote control capabilities, and even imagine applying robots to high-risk scenarios such as bomb disposal.
These frontline feedbacks have prompted Stardust Intelligence to make adjustments in its product plan: on the one hand, Stardust Intelligence has made “subtraction” and launched the half body product S1-U. For example, in cooperation with industrial robot manufacturer Xian Gong Intelligence, the two parties jointly launched a modular combination of “upper body machine+AGV (automatic guided vehicle) chassis” to promote the robot to enter the factory for testing, adapt to scenarios such as warehousing, manufacturing, logistics sorting, loading and unloading, and collaborative handling.
On the other hand, on the flagship product S1, Stardust Intelligence has made an “addition”, equipped with 5 sets of depth cameras, 2 laser radars, and 1 blind correction camera, combined with an edge computing platform, for high-precision path planning and task execution, better responding to complex working conditions such as stacking materials and narrow passages in factories. Although this configuration increases the cost of a single machine, it helps to verify the capability boundaries in extreme scenarios and provides standards for the establishment of future product systems.
At the same time, leading manufacturing companies are also considering how to embrace humanoid robots. In mid October, at a embodied intelligence academic conference held in Hangzhou, Zeng Fangqin, the chairman of “Guolian” leader Zhizhi Zao (002600. SZ), spent three hours recording technical points. She told the Economic Observer that once humanoid robots enter the factory, the entire production line and process will be restructured, and “the company is learning from top to bottom”.
Although Zeng Fangqin did not disclose the specific deployment pace, Lingzhi Robotics disclosed on October 27th that it provides complete machine manufacturing and OEM services for humanoid robots to multiple customers. Recently, it has also received orders for more than 100 complete machines from a top customer and deepened cooperation with multiple companies such as QiangBrain Technology, Beijing Humanoid Robot Innovation Center, and Zhiyuan Robotics.
However, compared to the high-profile announcement of orders by sellers, the voice of manufacturing buyers still appears weak.
Several frontline engineers told Economic Observer that they have not yet seen humanoid robots in factories. Even within companies that announce large purchases, frontline employees lack a clear understanding of “why buy” and “why buy”.
In addition, there are also phenomena similar to those in the early new energy vehicle industry behind some signed orders: some orders come from affiliated parties of shareholder enterprises or group enterprises, or are signed in the form of framework agreements, lacking specific delivery plans and rhythms. These signs collectively point to a reality: the current industrialization process of humanoid robots is still mainly driven by supply, and real demand has not yet been widely established.
Unlike the manufacturing industry, the procurement path of humanoid robots in the education industry usually starts with “small-scale pilot projects”, which may not have the ultimate requirements for technical performance, but have higher expectations for teaching adaptation, course resources, and on-site demonstration effects. If the manufacturing industry is looking forward to an efficiency revolution, the education market offers another possibility: can humanoid robots become a popular teaching tool like personal computers did back then?
Several education industry robot dealers have stated that the procurement process is generally driven by the “POC” (Proof of Concept) model: initially, 1-3 units are deployed in small batches on campus for cultural performances, teaching demonstrations, or programming competitions. Subsequently, the decision to expand the procurement scale is made based on student feedback and the degree of integration with the curriculum. Due to budget approval and procurement processes, the education market is more sensitive to “low threshold, high availability” products. Local education commissions, university laboratories, and other institutions have introduced humanoid robots in bulk through the “competition to promote teaching” approach, becoming another force driving order release. This type of order demand is more evident in first tier cities.
Some small orders also went to companies that do secondary development and exhibitions. These enterprises actively use robots to “create scenes” – leasing, exhibitions, cultural tourism, and other parallel promotion.
A company engaged in robot entertainment has purchased over 50 robots and robot dogs from various brands since last year, and has completed more than 100 performances this year. Robots are more like “mobile advertising spaces”.
The Economic Observer learned that the daily rental price of a single robot ranges from 4000 to 7000 yuan, a decrease of nearly 30% compared to the beginning of the year, reflecting the downward trend of robot popularity this year.
Delivery is the touchstone
Signing an order is just the starting point of industrialization. The real test lies in the delivery phase.
On October 26th, Songyan Power announced the completion of nearly 300 million yuan in Pre-B round financing. Seven months ago, this company launched N2 robots for developers, and has since signed over 2500 orders with a total contract amount exceeding 100 million yuan.
Jiang Zheyuan, the founder of Songyan Power, told the Economic Observer that the team had found that all the ankle joints of a batch of products were missing pins in the quality inspection at an early stage. They fined the production line at the first time, and then adjusted the process to preload the pins with the incoming materials and launch them as a whole.
This is an important transformation. 99% of delivery problems are not caused by workers, but by product design, “Jiang Zheyuan said. Compared to the technical polishing in the laboratory, what is truly tested after entering the delivery stage is the product design ability, which relies on a production system that can self correct and iterate quickly.
Another heavy pressure on delivery comes from the supply chain. The core components of Songyan Power’s robot are self-developed, with some parts sourced from external sources. As the order volume increases, some core suppliers are experiencing tight production capacity. Songyan Power responds to uncertainty by promoting multi-source stocking and other means.
After accumulating delivery experience, Songyan Power has also become one of the few companies in the industry to publicly disclose detailed expansion plans. The Economic Observer learned that its Beijing factory currently has a monthly delivery capacity of over 200 units, the first phase of the Changzhou factory is expected to reach a production capacity of 300 units/month in December, and the third factory with a planned production capacity of 500 units/month is also in preparation.
In Jiang Zheyuan’s view, the most important thing is not “posting orders”, but the closed-loop ability of delivery. This includes whether the entire business process, from obtaining customer intentions to collecting deposits, delivering products, and collecting the final payment, truly runs smoothly. Songyan Power regards “confirmed revenue” and “healthy cash flow” as key indicators, gradually moving away from the early pursuit of a single signing amount.
This shift in industry pace from signing contracts to delivery is also reflected in the capital market.
Several investors told the Economic Observer that the financing window for 2026 is gradually tightening. The top team has basically completed angel and Series A financing. In the next stage, who can enter Series B and higher stages no longer depends on the team’s background and storytelling ability, but on whether the product is truly implemented and delivered stably.
Li Yipei, director of Beijing Robot Financial Leasing Co., Ltd., told the Economic Observer that this wave of enthusiasm is quietly changing the evaluation logic of investors. At the beginning of the year, the industry could only see demos and promotions, but now we are looking at real sales and payment capabilities. The recent surge in “thousand level orders” has indeed helped with valuation, but ultimately the competition is still about who can truly deliver robots to customers, receive real payments, and continue to repurchase.
Zhao Weichen, Vice President of Accelerated Evolution, holds a similar view. He told the Economic Observer that in the first half of 2025, the industry experienced a period of traffic dividends, and there were also situations where financing could be obtained with just one viral video, but this model quickly became ineffective. It is becoming increasingly clear that relying solely on demos is meaningless, as the industry is returning to the fundamentals. ”
Unlike many companies betting on industrial scenarios and attempting to quickly increase volume by signing large orders, Accelerated Evolution has chosen a slower path by entering the education market. They package the robot body, graphical development platform, and supporting events into a complete solution and sell it to institutions such as universities, primary and secondary schools. Just like the first computer installed in the classroom back then, we also want to make robots a standard feature, “Zhao Weichen said.
In order to avoid “signing too many contracts and not being able to deliver”, the accelerated evolution adopts a phased delivery strategy of “10-100-1000”, selling one unit first, followed by ten or a hundred units. Founder Cheng Hao revealed that the team had voluntarily declined a large order of 500 units just to stabilize the delivery pace. He said that robots, as highly integrated hardware, have no tricks and must rely on solid engineering, design, and supply chain to solve problems one by one. In the process of serving education scenarios, Accelerated Evolution has also established a set of user operation indicators system, which not only calculates shipment volume, but also tracks indicators such as robot daily activity and repurchase rate to evaluate the real usage of robots.
As of the end of October, Accelerated Evolution has shipped over 700 humanoid robots, with overseas market shipments accounting for over 50%.
Price drops, urgent shortage of sales
When delivery capability becomes the touchstone, who will sell robots has become a new competition.
Entering any robot conference or industry matchmaking event, the most active participants are not only engineers, but also a group of salespeople wearing work badges and holding business cards: key account managers, regional directors, channel business, product consultants… Although their titles are different, their tasks are the same: to sell robots.
The pressure of shipment is permeating into the organizational structure. The Economic Observer learned that several companies have recently offered annual salaries ranging from 300000 to 500000 yuan to recruit sales teams. Even some top companies directly poach people from the retail system of investors and quickly build a humanoid robot version of the 4S network through existing channels.
Channel construction is also being promoted synchronously. Some companies have set the goal of “100 cities in three years”, while others have built independent sales systems around the three major application scenarios of education, manufacturing, and cultural tourism. The goal is very clear, to truly push robots into campuses, scenic spots, and workshops.
The price decline is another battle without gunpowder.
On October 23rd, Songyan Power launched the “Little Bumi” with a pre-sale price of 9998 yuan, becoming the first humanoid robot in the industry to enter the price range of less than 10000 yuan. Two days after the product was released, the first batch of 500 Little Bumi sold out on the JD platform.
Jiang Zheyuan does not shy away from external doubts about “price wars”. He said that Xiaobumi’s goal is to open up a brand new price range. “If I sell it for 40000 yuan and others sell it for 35000 yuan, it’s an involution; but we directly set the price to 9998 yuan to explore the possibility of new scenarios.” He explained that Xiaobumi uses lightweight composite materials, self-developed controllers, and core components to reduce overall manufacturing costs.
The outside world has also speculated that Xiaobumi is a “KPI product” temporarily launched by Songyan Power to sprint for fourth quarter revenue. Jiang Zheyuan explicitly denies this. He stated that a positive cash flow target had indeed been set before, but it has not yet been achieved due to exceeding expectations in research and development investment and team expansion speed. The launch of Xiaobumi is an active exploration towards the “demand side”, rather than a sprint for cash flow.
One day later, Accelerated Evolution released a new generation of entry-level product BoosterK1, positioned as a tailor-made development platform, priced at 29900 yuan. In contrast, it is more like a “lightweight platform” for developers, used for AI control, simulation, and teaching experiments. On that day, the product signed a distribution order for thousands of units.
Cheng Hao also downplayed the color of “price war” and pointed out that low prices aim to lower the development threshold. The current industry is more like the early PC industry, and each company is making differentiated layouts based on their own positioning. He believes that humanoid robots are still far from being truly consumer grade products, and in the next 1-2 years, they may provide more accompanying emotional value.
In terms of pricing strategy, Stardust Intelligence has taken a different path. Lai Jie compared the flagship S1 series, priced at 500000 yuan, to an “S-Class luxury car”. This configuration includes ontology, teleoperation, and model system, which are used to verify the performance limit and environmental adaptability of the product. He also revealed that the next generation of more cost-effective products is already under development, with plans to remove multiple modules from S1 and develop sub series covering different budgets and application scenarios, such as C-class and A-class models, in order to scale up and meet the actual needs of mid to long tail customers.
Jiang Zheyuan does not agree with the statement that the industry will cool down next year. He judged that 2026 is actually a crucial year for the humanoid robot industry to accelerate and increase production. On the one hand, Yushu Technology is advancing its IPO, which will once again increase industry attention; On the other hand, various enterprise projects are still being intensively promoted, and the delivery pace is significantly accelerating. From both an emotional and business perspective, next year may be better than this year
At the same time, going global has become one of the key directions for robot companies to expand. Yao Maoqing stated that Zhiyuan is building a distribution system globally and working with local partners in key markets to promote localized product design and delivery. He expects that by next year, Zhiyuan’s overseas business revenue is expected to exceed 30%.
The humanoid robot industry is moving away from the lively technological narrative and towards an industrialization stage centered on order fulfillment and payment collection capabilities. At this stage, the competition is no longer about who is more eye-catching, but about who can run the closed loop of products, industry scenarios, and delivery.
暂无评论内容