Many managers are hoping to replace factory workers with humanoids. Increasingly, they believe their wishes are about to be heard.
At the 2026 Hannover Fair, the buzz was all about humanoids. I noted meaningful progress in manufacturing software, but that’s not where the crowds gathered. The humanoid booths were impossible to miss. They were packed with spectators standing shoulder to shoulder. Me included.
The humanoids were waving and walking. Some were talking and dancing. Others were working on mock-up assembly lines or conducting quality inspections. Some booths without fully functional systems still displayed human-like robotic mannequins; just standing there was enough to draw attention. In the most spectacular setup, two humanoids were assembling a third humanoid—at least that was the narrative being presented.
I have a running bet with a senior manager at a global vehicle manufacturer. His stance: at least 3 humanoids will be doing value-added work in his main factory by late 2027. I think not. But first, let’s look at the market and case for humanoids.
The case for humanoids in manufacturing
As I witnessed at the 2026 Hannover Fair, the humanoid market is growing rapidly. If these ambitions materialize, an army of humanoids could be checking into assembly lines sooner than many expect. Market research firms agree; estimates for the global humanoid markets are in the range of several trillion USD by 2030.
The central claim is common sense: Because factories are already designed for humans, a robot with a human-like form can operate in existing production environments without major redesign. Humanoids could be deployed across a wide range of tasks currently performed by humans. Instead of investing in expensive redesign to accommodate clunky industrial robots or buying new machines, simply deploy flexible, general-purpose humanoids.
Automobile supplier Schaeffler used the fair to announce the planned deployment of 1,000 AEON humanoids from Hexagon Robotics across its factory network by 2032. It is not the only company with such plans. At Tesla, for example, humanoids are part of its long-term manufacturing vision, and its Optimus has already been working in the factory—at least to train the algorithm. Another known humanoid company, Figure AI, piloted humanoids in collaboration with BMW. Reports suggest that the Figure 02 humanoid logged 1,250 hours of value-adding work under controlled ten-hour shift conditions, helping produce 30,000 BMW X3s. Evidently, humanoids are already generating money for some; last year, Figure AI raised more than USD 1 billion in its Series C funding round.
The pitch is appealing: 24-hour shifts, no sick leaves, no bonus discussions, no complaints. Especially in times of worker shortages and skill gaps, the idea of replacing shop floor worker Mathias with a Maschinenmensch (machine human) is attractive to many business owners.
There is also a more subtle factor. Humans are drawn to human-like systems [1]. Humanoids capture attention in ways conventional automation does not. That was evident at the fair and is evident in early deployments in service environments such as airports and restaurants. Especially promising markets for humanoids are adult entertainment and military operations. However, this post is not about the broader market for humanoids. It is about their role in manufacturing operations.
Why humanoids won’t survive real factory conditions
When assessed from an operations perspective, the case for humanoids becomes less convincing.
First, factories are not designed for generality [2]. They are designed for performance under specific product and process conditions. Factories compete on product quality, throughput, reliability, speed, cost efficiency, and stability over time. System performance depends on synchronization across processes, while maintaining safety and consistency. These constraints are not visible in technology demonstrations. What is visible today is that humanoids are still slow and clumsy. Proponents argue that this is temporary, but they forget that safety concerns are real. No one wants a K.O. from an iron-fist humanoid coworker.
Second, while humanoids may not get sick, they drift and defect. Any experienced engineer knows how difficult it is to design robust, error-proof automation. An industrial robot typically operates with around 6 degrees of freedom, enabling precise positioning and orientation in space, whereas a humanoid robot usually has 30 or more degrees of freedom to replicate the complex, multi-joint movements of the human body, including limbs, torso, and hands. This increases flexibility, but it also multiplies the number of potential failure modes. Many of them will surface in real operation. Factories adopting humanoid workers should plan for extensive maintenance infrastructure—a kind of robotic “lazaret”—and specialized engineering support to keep them operational.
Third, just because a use case exists doesn’t mean there’s a business case. Training, supervising, and maintaining a humanoid workforce will be costly. Once downtime, failures, and engineering support are factored in, the total cost is likely to exceed the labor costs being replaced. Not forgetting the costs of jumping in to keep operations going when the robots err. The most optimistic humanoid evangelists seem to think that even those tasks can be delegated to humanoids. It cannot. Current humanoids struggle with anything outside their training data. Problem-solving will remain a human responsibility.
Humanoids are show-ready, not shop-floor ready
Humanoid robots represent an interesting development in robotics and may, over time, find applications in specific niche industrial contexts.
However, in the near term, their role in factories will be limited. Where they do appear, it is likely in pilot settings or in low-impact tasks that are visible to visitors but not critical. I bet there will always be better, more effective, and economical ways to automate (or eliminate!) the work tasks assigned to humanoids.
Timely shared on the International Labor Day—this is my current perspective on humanoids in manufacturing. What is yours?
[1] Even a pair of sticky eyes on a steel structure can make humans relate to it more. I recall a conference discussion of a study conducted in a supermarket in Japan that showed sales increased when promotion AGVs were equipped with sticky eyes.
[2] Often, factories have also not been designed for human work but for machines and material transport. Unfortunately, human work, the subject of industrial engineering, has usually been treated as an afterthought.
What's on your mind?