The humanoid robot narrative has shifted decisively over the past twelve months. What was once a showcase of laboratory prototypes and carefully choreographed demonstration videos has become a story about manufacturing scale, supply chains, and the unglamorous business of putting robots into real-world workflows. For anyone tracking the robotics industry in Europe, the question is no longer whether humanoid robots will arrive, but who will build them at volume, who will control the cost curve, and what that means for operators who are expected to deploy these machines in factories, warehouses, and eventually homes.
The source material for this analysis comes from a report by analyst firm Omdia, titled ‘General-purpose Embodied Intelligent Robots’, which was covered by Telecoms.com. The report’s central claim is straightforward: China holds a significant early advantage in the manufacture of humanoid robots, and that advantage is already visible in the shift from prototypes to scaled pilots. The report’s author, Lian Jye Su, is quoted as saying that by late 2025, embodied intelligence had moved from prototypes to scaled pilots, driven by Chinese volume production on one side and US AI innovation on the other. That sentence captures the current geopolitical and industrial split neatly: China is winning on hardware volume, while the United States continues to lead on the software and model side of the equation.
This is not a trivial distinction. The humanoid robot market is often discussed as if it were a single race, but the Omdia analysis suggests two parallel tracks that are now converging. Chinese manufacturers have the factory floor expertise, the component supply chains, and the willingness to produce at scale. US firms, meanwhile, have the AI models, the cloud infrastructure, and the research ecosystem that makes these robots intelligent in the first place. The report’s framing implies that neither side can fully succeed without the other, at least for now. For European operators, this creates a complex procurement landscape where the hardware may come from one region, the software from another, and the integration work will likely fall to local system integrators.
The market forecast in the source material is aggressive. Omdia projects that the global market for general-purpose embodied intelligent robots will roughly double every year for a period, reaching 2.6 million units by 2035. That figure is worth pausing on. It is not a forecast for all robots, nor for industrial arms, nor for collaborative robots. It is specifically for general-purpose embodied intelligent robots — machines that can navigate unstructured environments, manipulate objects with dexterous hands, and learn new tasks through imitation and reinforcement learning. A market of 2.6 million units by 2035 implies a compound annual growth rate that is steep by any standard, and it suggests that the humanoid form factor is not a niche experiment but a mainstream industrial tool within a decade.
The source material also highlights a specific technical milestone that underpins this growth: dexterous hands with 16 degrees of freedom (DoF). This is not a cosmetic upgrade. Hands with 16 DoF can perform a much wider range of manipulation tasks than the simple grippers or low-DoF claws that dominated earlier robot designs. Combined with end-to-end imitation learning and self-reinforcement learning, these hands allow robots to be trained on tasks by demonstration rather than by explicit programming. That shift from hand-coded behavior to learned behavior is what makes general-purpose robots economically viable in settings where task variety is high and volumes are low. In a warehouse, for example, a robot that can be shown how to pick, pack, and sort a dozen different product types without reprogramming is far more valuable than a robot that needs a software update for every new SKU.
The Omdia report’s title — ‘General-purpose Embodied Intelligent Robots’ — is itself a signal. The industry has moved away from the term ‘humanoid’ in some analyst circles, preferring language that emphasizes the embodied intelligence aspect rather than the anthropomorphic form. This is a subtle but important shift. It suggests that the value proposition is not about looking like a human, but about having a body that can interact with a world designed for humans. Stairs, doors, tools, and workspaces are all built for human dimensions and human dexterity. A robot with a human-like form factor can operate in those environments without requiring extensive retrofitting. That is the economic argument for the form factor, and it is why the market forecast is so aggressive.
Key findings
The first key finding from the source material is that China has a massive head-start in the manufacture of humanoid robots. The Omdia report is explicit on this point, and the timing is notable. The report states that by late 2025, embodied intelligence had shifted from prototypes to scaled pilots. That is a very short window between demonstration and deployment. Most emerging technologies take years to move from lab to pilot, but the combination of Chinese volume production and US AI innovation appears to have compressed that timeline significantly. The source does not provide specific production numbers, factory locations, or manufacturer names, so we cannot verify the scale of Chinese output from this report alone. What we can say is that the analyst firm’s assessment is unambiguous: China is the early leader in this race.
The second key finding is the convergence of three technical pillars: AI models, dexterous hands with 16 DoF, and end-to-end imitation and self-reinforcement learning. The report treats these as mutually reinforcing. AI models provide the reasoning and perception capabilities. The 16-DoF hands provide the physical manipulation capability. And the learning methods allow the robot to acquire new skills without explicit programming. Together, these three pillars make the robots viable for industrial roles first, then service roles, and eventually household roles. The source material does not provide a timeline for when each of these roles becomes mainstream, but the ordering is clear: industrial first, service second, household last. This is consistent with the economics of each market. Industrial customers have the highest willingness to pay for labor replacement, the most controlled environments, and the clearest return on investment. Households, by contrast, have the lowest willingness to pay, the most unstructured environments, and the highest safety requirements.
The third key finding is the market forecast itself. Omdia projects that the global market for general-purpose embodied intelligent robots will reach 2.6 million units by 2035, with the market roughly doubling every year for a period. The source does not specify the exact start year for this doubling, nor does it provide a breakdown by region, industry, or form factor. We also do not know whether this forecast includes only humanoid robots or also other embodied intelligent systems. The report title suggests a broad category, but the source material focuses on humanoid robots specifically. For the purposes of this analysis, we will treat the 2.6 million figure as a total for general-purpose embodied intelligent robots, which includes but may not be limited to humanoid form factors.
The fourth key finding, implicit in the report’s framing, is the division of labor between China and the United States. The source material credits Chinese volume production and US AI innovation as the two drivers of the shift to scaled pilots. This is a complementary relationship, but it is also a fragile one. If trade tensions escalate, if export controls on AI chips tighten further, or if intellectual property disputes arise, the convergence could stall. The source material does not address these risks, so we flag them as open questions rather than findings. What the report does tell us is that the current trajectory is one of convergence, not divergence.
What it means for European operators
For European operators — whether they run factories, logistics networks, or service businesses — the Omdia forecast has several practical implications. The first is that the window for early adoption is now. If the market is indeed doubling every year and reaching 2.6 million units by 2035, then the first wave of scaled pilots is already happening. European operators who wait for the technology to mature further may find themselves competing against early adopters who have already worked out the integration challenges, trained their staff, and refined their use cases. The source material does not provide specific deployment examples or pilot results, so we cannot quantify the benefits that early adopters have seen. But the direction of travel is clear.
The second implication is that European operators will need to navigate a two-region supply chain. The hardware is likely to come from China, given the manufacturing head-start documented in the report. The AI software and models are likely to come from the United States, given the innovation advantage noted by Omdia. European operators will therefore need to manage procurement relationships across two very different regulatory and geopolitical environments. This is not a new challenge — European manufacturers have long sourced components from Asia and software from the US — but the stakes are higher with humanoid robots because the hardware and software are tightly integrated. A robot’s dexterous hands are useless without the AI models that control them, and the AI models are useless without the physical platform. The source material does not specify whether the 16-DoF hands are a Chinese or US innovation, nor does it identify specific vendors. We flag this as a gap in the available information.
The third implication is around standards and safety. The source material notes that the robots are becoming viable for industrial, service, and eventual household roles. Each of these roles has different safety requirements, and Europe has some of the most stringent regulations in the world for machinery, data protection, and worker safety. The source material does not address regulatory compliance, so we cannot say how the Omdia forecast interacts with European standards. What we can say is that European operators will need to work with suppliers who can demonstrate compliance with local regulations. This may slow adoption relative to markets with lighter regulatory burdens, but it may also create a competitive advantage for European integrators who can navigate the regulatory landscape.
The fourth implication is about workforce and skills. The report’s emphasis on end-to-end imitation learning and self-reinforcement learning suggests that the robots will be trained rather than programmed. This has implications for the skills required on the factory floor. Instead of hiring robot programmers, operators may need to hire trainers who can demonstrate tasks to the robots. The source material does not provide details on the training process, the time required to train a new task, or the level of expertise needed. We flag these as unknowns. What is clear is that the skill mix will shift, and European operators should start planning for that shift now.
The fifth implication is financial. A market that reaches 2.6 million units by 2035 will require significant capital investment across the value chain — from component manufacturers to robot assemblers to software developers to system integrators. European operators will need to budget for these investments, but they will also have opportunities to participate in the value chain. The source material does not provide pricing information, total cost of ownership estimates, or payback period calculations. We cannot invent these figures. What we can say is that the scale of the forecast implies that unit costs will need to fall dramatically from current levels, and that this cost reduction will likely be driven by Chinese volume production, as noted in the report.
The sixth implication is strategic. The source material describes a market that is doubling every year, led by China in manufacturing and the US in AI. Europe is not mentioned as a leader in either domain. This does not mean Europe is irrelevant — it means European operators will likely be buyers and integrators rather than producers of core technology. That is a viable position, but it requires a clear strategy. European operators should identify their use cases, build relationships with suppliers from both China and the US, and invest in the integration and training capabilities that will be needed to deploy these robots effectively. The source material does not provide a roadmap for this, so we offer it as analysis rather than as a finding from the report.
There are also open questions that the source material does not answer. We do not know the current installed base of general-purpose embodied intelligent robots, so we cannot calculate the implied growth rate from today to 2.6 million units in 2035. We do not know the average selling price, so we cannot estimate the market value in euros or dollars. We do not know the failure rates, maintenance requirements, or operational lifetimes of these robots, so we cannot advise on total cost of ownership. We do not know the specific industrial or service applications that are driving the scaled pilots, so we cannot recommend which sectors should prioritize adoption. These are significant gaps, and we flag them explicitly rather than filling them with speculation.
What the source material does tell us, with reasonable confidence, is that the era of general-purpose embodied intelligent robots is beginning, that China has a manufacturing head-start, that the US leads in AI innovation, and that the market is forecast to grow at a dramatic pace through 2035. For European operators, the message is to prepare now — not by making speculative bets on specific vendors, but by building the organizational capabilities to evaluate, deploy, and integrate these systems as they become available. The report’s timeline suggests that scaled pilots are already underway as of late 2025, which means the window for early learning is open now. Operators who wait for the technology to stabilize may find that the competitive landscape has already shifted.
In summary, the Omdia report, as covered by Telecoms.com, provides a clear picture of a market in transition. China leads in manufacturing, the US leads in AI, and the combination is driving humanoid robots from prototypes to scaled pilots. The forecast of 2.6 million units by 2035 implies a doubling market for several years. European operators should treat this as a strategic signal, not a forecast to be taken at face value. The report does not provide enough detail to build a business case, but it provides enough direction to start planning. That is the practical takeaway for a European audience.
Sources
https://www.telecoms.com/ai/humanoid-robot-production-goes-exponential-led-by-china
Published by Robot Service Map.