The dexterous hand track has become one of the most fiercely competitive segmented battlefields in the humanoid robot industry chain. Looking at the 2026 semi-annual reports of related concept stocks, the industry's capability narrative is highly consistent: first, multiple technical routes are running in parallel, and second, it is entering the mass supply stage. However, returning to the financial statements, the contribution of the dexterous hand business to listed companies is still limited.
Behind this narrative consistency is the dispersion of scenes and demands for dexterous hands. "The market has not yet been finalized, and everyone is exploring the optimal solution (for products)." Recently, a listed company engaged in the upstream motor business admitted to a reporter from the "Daily Economic News" that customer demand is very fluid and has not yet reached the standard for companies to start fully automated mass production.
The industry is still being explored, and the popularity of the track remains high. Since the beginning of this year, capital has been pouring into the dexterous hand track non-stop. On average, dexterous hand manufacturers have completed a new round of financing almost every week. At the same time, companies are publishing prices, parameters, and demonstrations. This seems to be the standard narrative for high-tech industries to take off.
However, behind the booth is another picture: the three-finger claw or low-degree-of-freedom hand is still "sales-driven"; most customers are still in the proof-of-concept stage and are still far away from mass purchase; the model's control capabilities are still limited.
When the competition for capital, production capacity and parameters starts first, the real test question of the industry has just emerged: who can find the scalable "battlefield" in the scattered scene, and only with dexterous hands can cross the key intersection from "showing off skills" to "commercial use".
“Customers are highly willing to purchase three-fingered dexterous hands”
At the "2026 World Robot Conference" held not long ago, the display of dexterous hands with high degree of freedom was dazzling. There were spectators queuing up to play guessing games with dexterous hands, manufacturers demonstrating the use of dexterous hands to assemble dexterous hands on site, and robots equipped with dexterous hands to play guitar and piano.
As the most delicate "end effector" of human beings, the hand is usually considered to have 27 degrees of freedom, and the dexterous hand assembled by machinery has already surpassed the human hand of flesh and blood in terms of degrees of freedom. The new generation of bionic tensile and compression body dexterous hand developed by Yuequan Bionic - Yingshou Y-Hand M2 has 38 degrees of freedom. According to the company, it is the world's highest degree of freedom bionic dexterous hand. There are also many other manufacturers that have launched dexterous hands with 30+ degrees of freedom.
The higher the degree of freedom, the more flexible and precise the operation ability of dexterous hands will be. However, the dexterity hand industry is showing a set of intriguing contradictions. On the booth, various companies compete to display high-degree-of-freedom five-finger dexterous hands, but the real hot sellers are more three-finger hands and low-degree-of-freedom products.
"As a company specializing in robot end effectors, it would be unconvincing if we didn't reveal some high-end products with high degrees of freedom." A dexterous hand practitioner said frankly, "But whether dexterous hands with high degrees of freedom can open molds for mass production, in fact, no."
According to the "Daily Economic News" reporter, the dexterous hand shipments of the company where this practitioner works are still dominated by products with six active degrees of freedom.
"We believe that it is unrealistic to make money by relying on high-degree-of-freedom dexterous hands." The above-mentioned practitioners further stated that the core bottleneck of high-degree-of-freedom dexterous hands making it difficult to achieve large-scale supply is not technical capabilities, but the demand side. Currently, there are no bulk purchase orders in the market that are sufficient to support large-scale mass production. From the perspective of application scenarios, at present, high-degree-of-freedom dexterous hands are mainly concentrated in the scientific research market, and downstream companies purchase a small amount and install them on cutting-edge or high-end robot platforms for product display, etc.
Zhongke Silicon Century also promotes a dexterous hand with 11 joints and 6 drives. According to Zhang Tianyi, CEO of Zhongke Silicon Ji, customers are very willing to purchase three-fingered dexterous hands. "The third finger is between the second finger and the fifth finger. It not only takes into account the pointing ability of the second finger, but also has a certain general grasping ability of the five-fingered hand. It is quick for customers to get started and use."
Performance does not necessarily increase linearly with greater degrees of freedom in dexterous hands. Zhang Tianyi pointed out that a high degree of freedom also brings problems such as high failure rate, high cost and poor consistency. Dexterous hands that can truly continue to work in the industry and work uninterrupted are truly useful hands. Compared with the degree of freedom, "we pay more attention to key data such as battery life, load, number of movements, and temperature control."
“The industry has entered the stage where software controls hardware”
So, why are high-degree-of-freedom dexterous hands currently “not useful”?
In terms of cost, a reporter from "Daily Economic News" learned that the market price of a high-degree-of-freedom dexterous hand ranges from hundreds of thousands to hundreds of thousands of yuan, and there are even products worth millions of yuan.
Fu Yuancheng, head of Xinuo Future Capital Markets, said that after the degree of freedom exceeds 25, the marginal effect of dexterous hands is diminishing, while the marginal cost gradually increases.
Motors account for about 20% of the cost of a dexterous hand. The higher the degree of freedom, the greater the number of motors, sensors, etc. required. However, adding too many motors to a dexterous hand with limited space not only increases the pressure on hardware costs, but also faces greater challenges in system integration, heat dissipation and other issues.
A dexterous hand moving towards commercialization must strike a balance between cost, reliability and scene requirements. Fu Yuancheng believes that, taking all things into consideration, a dexterous hand with about 20 active degrees of freedom and about 25 total degrees of freedom is enough to realize more than 90% of the functions of the human hand; more degrees of freedom will increase structural instability, and there will be more variables in cost.
On the other hand, Fu Yuancheng pointed out that the pre-training of the dexterous hand operation model requires the use of a large amount of human hand movement data. If the dexterous hand configuration has additional degrees of freedom beyond the scope of the human hand, it will be difficult to collect high-quality training data, and errors will also occur during the remote operation of the real machine.
Compared with hardware parameters, the shortcomings of the dexterous hand model may be more prominent.
"So far, all VLA (Visual-Language-Action) models cannot drive a hand with 20 to 30 degrees of freedom, and there is no good 'brain' that can drive it. It can move, but lacks flexibility." Zhang Zhengtao, founder and chairman of Zhongke Huiling and chairman of Zhongke Huisi, said recently.
According to Fu Yuancheng, the existing VLA model can only output about 10 motion control vectors at the same time stamp, which means that it can only control hardware with about 10 degrees of freedom at the same time point. "The industry has entered a stage where software controls hardware," Fu Yuancheng said. Currently, the degree of freedom of dexterous hands has been raised to a high level, and some precise movements can be achieved through remote control. However, the control capabilities of the model have not yet kept pace with the development of hardware.
Zhang Tianyi also pointed out that from a hardware perspective, it is difficult to say that there is a significant gap between the products of various manufacturers, including dexterous hands. The real focus of competition lies in the software layer, how to make dexterous hands smarter through algorithm upgrades, thereby maximizing the release of their dexterous operation capabilities. At present, software systems such as the "big brain" of dexterous hands have not yet reached a mature stage, which is one reason why some manufacturers have not attempted high-complexity scene demonstrations.
"The important task this year is to implement application scenarios"
Despite the challenges, the dexterous hand industry is accelerating from the laboratory to the factory.
According to a report from the Advanced Industrial Research Institute (GGII), China's dexterous hand market sales in 2025 will be approximately 19,200 units, a year-on-year increase of 236.84%; it is expected to reach 70,200 units in 2026, a year-on-year increase of 265.63%; and is expected to exceed 430,000 units by 2030.
In the future, Xinuo expects to achieve a delivery capacity of 10,000 units of dexterous hands by the end of 2026. Zhongke Silicone plans to have an annual production capacity of 30,000 to 40,000 units, with a target of exceeding 100,000 units in 2027.
Orders are increasing in volume, but the structure is clearly differentiated. Ni Hualiang, CEO of Aoyi Technology, once revealed that the company's recent orders have exceeded 1,000 units. Currently, there are nearly 300 dexterous hand-related partners, of which 90% are in the proof-of-concept stage. Looking at the customer income structure, 10% of customers contribute more than 90% of the income.
"Currently, the demand (for dexterous hands) is relatively scattered," according to Fu Yuancheng, about one-third is the demand for scientific research, data collection, and model training, and another one-third is the product deployment and implementation demand of robot body manufacturers, and the rest is for cutting-edge exploration and specific extreme scenarios.
However, there is no consistent solution in the industry for diversified scene needs. From motors to transmissions, from materials to sensors, domestic manufacturers have numerous product differences. For example, the tendon hand is suitable for basic operation scenarios such as simple general grabbing, the link hand performs better in industrial scenarios such as logistics sorting, loading and unloading, and the direct drive hand is usually used in scientific research, education, and special industries.
The aforementioned dexterous hand practitioners bluntly stated that customization of dexterous hand products will not be avoided in the short term. Although as a manufacturing enterprise, it prefers to provide standardized products because standard products mean that they can be mass-produced on a large scale. However, in the context of the current technical roadmap that has not yet converged, trial and error and adaptation are still issues that the industry must face.
"This year's entire embodied intelligence market, a very important and even life-and-death task is to implement application scenarios." said the dexterous hand practitioner.
Looking at different scenarios, in the next five years, Zhang Tianyi is cautious about the entry of dexterous hands into the home market, but he can take the lead in entering factories and special industries. For example, logistics sorting and handling, loading and unloading and other links have the potential for batch replication; in special industries, although the overall demand is limited, there are actual deployment conditions. In some specific processes, the proportion of production assisted by dexterous hands is expected to reach 40% to 50% within five years.
In the long term, Fu Yuancheng believes that the dexterous hand market will show a polarization trend along the degree of freedom dimension, and there will be almost no overlap in the application scenarios of clamping claws and high-degree-of-freedom dexterous hands. The low-degree-of-freedom dexterous hand solves the pain points of limited model capabilities and cost during the current transitional period. However, in the future, simple repetitive work will be completed by the gripper, and complex and refined operations will be undertaken by the high-degree-of-freedom dexterous hand. There will be almost no intermediate products with a few to a dozen degrees of freedom.
Daily reporter Yang Yu Daily editor Yang Yi
(Daily Economic News)
