The humanoid robot sector has reached a pivotal moment, according to recent statements from leading robotics firms in China and the United States. Over the past three months, multiple companies have publicly committed to producing humanoid robots at scale, signaling a shift from laboratory curiosities toward commercially viable industrial tools.
Among the most concrete claims comes from UBTECH, the Shenzhen-based robotics manufacturer. Yu Zheng, a roboticist and vice-dean of the UBTECH Research Institute, stated that more than 1,000 units of the company's Walker S2 model were deployed to factories during 2025. That figure, if accurate, represents a significant volume for a product category that has long struggled to move beyond pilot projects and demonstration videos.
The Walker S2 is not a speculative concept. It is a working humanoid designed for industrial environments, and UBTECH's stated deployment numbers suggest the company has moved past the prototype phase. However, the source material does not specify which factories received these robots, what tasks they performed, or whether the deployments were permanent installations or temporary trials. Those details remain undisclosed.
The broader industry context is equally important. Yu Zheng told the publication that humanoid robots are "much closer to this dream than a decade ago." He attributes this progress to three specific technological developments: denser batteries that allow robots to operate for hours rather than minutes, cheaper and more precise actuators that convert electrical energy into movement, and AI learning algorithms integrated into robot control systems.
These three advances are interconnected. Battery density determines how long a robot can work before recharging. Actuator quality determines how smoothly and accurately the robot can move. AI algorithms determine how effectively the robot can learn new tasks and adapt to changing conditions. Together, they address the fundamental limitations that have kept humanoids out of commercial deployment for decades.
The automotive industry has emerged as a particularly promising application area. Carolina Parada, who leads the robotics team at Google DeepMind and is based in Boulder, Colorado, described the automotive sector as "an ideal setting" for humanoid robots. Her team at Google DeepMind recently announced a partnership with Boston Dynamics, the Massachusetts-based robotics company known for its advanced mobility platforms.
The partnership between Google DeepMind and Boston Dynamics is notable for several reasons. It brings together two organisations with complementary strengths: Google DeepMind's expertise in artificial intelligence and machine learning, and Boston Dynamics' track record in physical robot design and control. The collaboration suggests that the industry recognises the need to combine software intelligence with mechanical capability.
Perhaps most significantly, both UBTECH and Boston Dynamics are applying the same fundamental technique in what the source material describes as "vast data-collection centres." In these facilities, humans remotely operate humanoid robots to teach them how to perform a range of tasks. This approach, known as teleoperation-based learning, allows robots to acquire skills through demonstration rather than explicit programming.
The logic behind this approach is straightforward. By having human operators guide robots through tasks remotely, the robots can collect large amounts of data about how those tasks are performed. This data can then be used to train AI models that allow the robots to perform the tasks autonomously. The more data collected, the more capable the robots become.
This convergence on data-collection strategies is a notable development. It suggests that the industry has reached a consensus on how to address one of the hardest problems in robotics: teaching machines to handle the complexity and variability of real-world tasks. Rather than trying to program every possible scenario, the industry is moving toward a learn-by-example model.
The source material does not provide specific timelines for when these robots will achieve full commercial viability, nor does it disclose the costs involved in the data-collection centres or the scale of the remote-operation workforce. These are significant unknowns that will affect the economics of humanoid deployment.
Why it matters for European robot service
For European businesses and service providers, the developments described in the source material carry implications that extend well beyond the factory floor. The humanoid robot market has historically been dominated by North American and Asian players, and the current announcements reinforce that pattern. European companies will need to consider how they position themselves in a market where the leading suppliers are increasingly confident about scaling production.
The automotive industry focus is particularly relevant for Europe. The continent is home to some of the world's largest automotive manufacturers, and many of these companies operate extensive factory networks across multiple countries. If humanoid robots prove effective in automotive applications, European plants could become early adopters. However, the source material does not indicate whether any European automotive companies are currently involved in the UBTECH or Boston Dynamics deployments.
The remote-operation training model also raises questions about where the value in humanoid robotics will ultimately reside. If robots are trained through vast data-collection centres, then the companies that control those centres and the associated data will hold significant competitive advantages. European robot service providers may need to consider whether they should develop their own data-collection capabilities or partner with companies that already have them.
There is also the question of workforce implications. The source material describes humans remotely operating robots to teach them tasks. This suggests that humanoid deployment will not necessarily eliminate human involvement in industrial processes. Instead, it may shift the nature of that involvement, with workers moving from physical tasks to supervisory and training roles. European companies will need to plan for these workforce transitions.
The technological advances described in the source material — denser batteries, better actuators, and improved AI algorithms — are not specific to any particular geographic region. European robotics companies could potentially benefit from the same technological trends. However, the source material does not provide information about European firms' progress in these areas, so it is not possible to assess their competitive position from this information alone.
For European service providers, the key takeaway is that humanoid robots are moving from the realm of research demonstrations toward practical deployment. The pace of this transition will depend on factors that are not fully disclosed in the source material, including costs, reliability, and the availability of trained personnel to operate the data-collection infrastructure.
European buyers should also note that the source material does not address regulatory or safety considerations. Humanoid robots operating alongside human workers will raise questions about workplace safety standards, liability, and insurance. These issues are likely to be addressed at the national and European Union levels, but no information about such regulatory developments is provided in the source material.
What buyers and operators should know
For organisations considering whether to invest in humanoid robots, the source material offers several points of guidance, along with some notable gaps in information.
First, the technology has demonstrably improved. The source material identifies three specific advances — battery density, actuator precision and cost, and AI learning algorithms — that have made humanoids more practical than they were a decade ago. Buyers should evaluate these three components when assessing any humanoid robot system. A robot with excellent AI but poor battery life will not be useful for extended shifts. A robot with good hardware but limited learning capabilities will require extensive programming for each new task.
Second, the deployment model is shifting toward data-driven learning. Both UBTECH and Boston Dynamics are using remote-operated data collection to train their robots. This means that the value of a humanoid robot is not solely in the hardware but also in the data infrastructure that supports it. Buyers should ask suppliers about their data-collection capabilities and how much training data has been accumulated for the specific tasks they need the robot to perform.
Third, the automotive industry is the current proving ground. The source material identifies automotive as "an ideal setting" for humanoids, and UBTECH has already deployed over 1,000 units to factories. Buyers in other industries should recognise that humanoid robots are likely to be most mature in automotive applications. Deployments in other sectors may be less proven, and buyers should seek evidence of successful implementations in their specific industry.
Fourth, the source material does not disclose several critical commercial details. There is no information about the purchase price or leasing costs of the Walker S2 or any other humanoid robot. There is no information about maintenance requirements, expected lifespan, or reliability metrics. There is no information about the availability of spare parts or the speed of service response. Buyers should not assume that any of these factors are favourable based on the information provided here.
Fifth, the remote-operation model has implications for ongoing operational costs. If robots require human operators to train them and to handle edge cases, then the total cost of ownership includes not just the robot hardware but also the personnel and infrastructure needed for training and supervision. Buyers should ask suppliers about the ratio of robots to human operators required for effective operation.
Sixth, the partnership between Google DeepMind and Boston Dynamics suggests that AI capability is becoming a key differentiator in the humanoid market. Buyers should evaluate the AI software that controls a robot as carefully as they evaluate the mechanical hardware. The ability of a robot to learn new tasks and adapt to changing conditions will determine its long-term usefulness.
Seventh, the source material does not address safety certifications, compliance standards, or insurance considerations. Humanoid robots are a relatively new category of industrial equipment, and regulatory frameworks may still be evolving. Buyers should investigate the regulatory status of humanoid robots in their jurisdiction before making purchasing decisions.
Eighth, the source material does not provide information about the total addressable market for humanoid robots or the production capacity of the companies involved. While UBTECH's deployment of over 1,000 units is significant, it is not clear how many units the company can produce annually or how quickly production can be scaled. Buyers should ask suppliers about production capacity and lead times.
Ninth, the source material does not discuss the total cost of ownership over the lifespan of a humanoid robot. While the initial purchase price is an important consideration, the ongoing costs of energy, maintenance, software updates, and training data collection may be substantial. Buyers should request detailed cost projections from suppliers.
Tenth, the source material does not address the question of interoperability. Can humanoid robots from different manufacturers work together in the same facility? Can they be integrated with existing industrial automation systems? These are important questions for buyers planning large-scale deployments, but the source material provides no information on these topics.
In summary, the source material indicates that humanoid robots have made significant technological progress and are being deployed in industrial settings at scale. However, many commercial details remain undisclosed. Buyers should approach the market with a clear understanding of what is known and what is not known, and they should seek additional information from suppliers on the specific factors that will determine the economic viability of humanoid deployment in their operations.
The next twelve to twenty-four months will likely be decisive for the humanoid robot industry. If the deployments described in the source material prove successful, we can expect to see rapid expansion. If they encounter unexpected problems, the industry may face a period of consolidation. Either way, European buyers and service providers should monitor these developments closely and prepare for a future in which humanoid robots are a routine part of industrial operations.
Sources
https://www.nature.com/articles/d41586-026-00164-0
Published by Vigla Media OÜ (Estonia).