During the next Years, humanoid robots will change the nature of the work. Or at least, this is what humanoid robotics companies have always promising, allowing them to collect hundreds of millions of dollars to assessments that come up against billions.
Make these promises will require a lot of robots. Agility Robotics plans to send “hundreds” of its figures in 2025 and has an Oregon factory capable of building more than 10,000 robots per year. Tesla plans to produce 5,000 of its optimus robots in 2025, and at least 50,000 in 2026. The figure estimates that “there is a path to 100,000 robots” by 2029. And these are only three of the largest companies in an increasingly congested space.
The amplification of this message is many financial analysts: Bank of America Global Research, for example, predicts that global shipments of humanoid robots will reach 18,000 units in 2025. And Morgan Stanley Research estimates that in 2050, there could be more than 1 billion humanoid robots, part of an American market of $ 5 billion.
But for the moment, the humanoid robots market is almost entirely hypothetical. Even the most prosperous companies in this space have deployed only a small handful of robots in carefully controlled pilot projects. And future projections seem to be based on an extraordinarily wide interpretation of the jobs that a capable, effective and safe humanoid robot – which does not currently exist – perhaps able to do. Can current reality connect with the promised scale?
What will it be necessary to put humanoid robots on the scale?
Building tens of thousands, even hundreds of thousands, physically, of humanoid robots is certainly possible in the short term. In 2023, in the order of 500,000 industrial robots was installed worldwide. Under the basic hypothesis that a humanoid robot is approximately equivalent to four industrial arms in terms of components, existing supply chains should be able to support even the most optimistic short -term projections for humanoid manufacturing.
But the simple fact of building robots is undoubtedly the simplest part of the Humanoid scale, explains Melonee Wise, who was product manager at Agility Robotics until this month. “The most important problem is demand – I don’t think anyone has found an application for humanoids that would require several thousand robots by installation.” Large deployments, explains Wise, are the most realistic way for a robotics company to develop its business, because the integration of any new customer can take weeks or months. Another approach to deploying several thousand robots to do a single job is to deploy several hundred robots that can each do 10 jobs, which seems to be what most of the humanoid industry is betting in the long term.
Although there is a belief in a large part of the humanoid robotics industry that rapid progress in AI must somehow result in rapid progress towards versatile robots, it is not clear how, when or if it happens. “I think many people hope is that they go to the AI of their way,” explains Wise. “But the reality of the situation is that attempts at the AI is not robust enough to meet the requirements of the market.”
Bring humanoid robots to the market
Market requirements for humanoid robots include a multitude of extremely dull and extremely critical things such as battery life, reliability and safety. Among these, the battery life is the simplest – for a robot to usefully work, it cannot spend most of its time loading. The next version of the agility figure robot, which can manage useful loads up to 16 kilograms, includes a bulky “backpack” containing a battery with a load ratio of 10 to 1: the robot can operate for 90 minutes and completely recharge in 9 minutes. The thinner humanoid robots of other societies must necessarily compromise to maintain their svelte factors.
In operation, Digit will probably spend a few minutes to load after having run for 30 minutes. Indeed, 60 minutes of execution of Digit is essentially a reserve in case something happens in its workspace which requires it to stop temporarily, an occurrence not inframed in logistics and manufacturing environments as target agility. Without reserve of 60 minutes, the robot would be much more likely to lack power in the middle of the task and must be recharged manually. Consider what it might look like a modest deployment of several hundred robots weighing more than a hundred kilograms each. “No one wants to face this,” comments Sage.
Potential customers for humanoid robots are very concerned about downtime. During one month, a factory operating with 99% reliability will see around 5 hours of stopping. Wise says that any time of stopping that stops something like a production chain can cost tens of thousands of dollars per minute, which is why many industrial customers are waiting for a few other 9s of reliability: 99.99%. Wise says that agility has demonstrated this level of reliability in certain specific applications, but not in the context of versatile or general use.
A humanoid robot in an industrial environment must respond to the general security requirements For industrial machines. In the past, robotic systems such as vehicles and autonomous drones have benefited from immature regulatory environments to evolve quickly. But Wise says that this approach cannot work for humanoids, because the industry is already strongly regulated – the robot is simply considered as another piece of machinery.
There are also more specific safety standards currently in development for humanoid robots, explains Matt Powers, associate director of R&D of autonomy at Boston Dynamics. He notes that his business helps develop a safety standard for the International Organization for Standardization (ISO) to dynamically balance the leg robots. “We are very happy that the best players in the field, such as agility and figure, join us to develop a way to explain why we think that the systems we deploy are safe,” explains Powers.
These standards are needed because the traditional cutting up safety approach may not be a good option for a dynamically balancing system. This will drop a humanoid robot, which worsens the situation. There is no simple solution to this problem, and the initial approach that Boston Dynamics expects to adopt with his atlas robot is to keep the robot outside the situations where the simple fact of feeding it may not be the best option. “We are going to start with relatively low risk deployments, then we develop as we strengthen confidence in our security systems,” explains Powers. “I think a methodical approach will really be the winner here.”
In practice, a low risk means to keep people’s humanoid robots away. But the humanoids limited by the jobs they can do safely and where they can move safely have more difficulty finding tasks that bring value.
Are humanoids the answer?
The problems of demand, battery life, reliability and security must all be resolved before humanoid robots can evolve. But a more fundamental question to ask is whether a biped robot is really worth it.
Dynamic balancing with the legs would theoretically allow these robots to navigate in complex environments like a human. However, demonstration videos show these humanoid robots as being mainly stationary or moving short distances on flat floors. The promise is that what we see now is only the first step towards human mobility. But in the short and medium term, there are much more reliable, effective and profitable platforms that can take over in these situations: robots with arms, but with wheels instead of the legs.
Safe and reliable humanoid robots have the potential to revolutionize the labor market at some point in the future. But the potential is only that, and despite the humanoid enthusiasm, we must be realistic on what it will take to transform the potential into reality.
This article appears in the October 2025 printing problem as “why humanoid robots are not on the scale”.
From your site items
Related items on the web
