Humanoid Robots: The Scaling Challenge

Over the next few years, humanoid robots the nature of work will change. Or at least that's what a humanoid robotics companies have always been promising, which allowed them to attract hundreds from millions from dollars estimated to be in the billions.

It will take a lot of robots to deliver on these promises. Dexterity Robotics awaiting dispatch”hundredsof its Digit robots in 2025 and has a factory in Oregon able to build more than 10,000 robots per year. Tesla plans produce 5,000 of its Optimus robots in 2025 and at least 50,000 in 2026.there is a path to 100,000 robotsby 2029. And these are just three of the biggest companies in an increasingly crowded space.

Many financial analysts reinforce this message: Bank of America Global Researchfor example, predicts that global shipments of humanoid robots will reach 18,000 units in 2025. Morgan Stanley Research estimates that by 2050 the number of humanoid robots could exceed 1 billion, becoming part of a US$5 trillion market.

But for now, the market for humanoid robots is almost entirely hypothetical. Even the most successful companies in this space have deployed only a small handful of robots under carefully controlled conditions. pilot projects. And predictions for the future seem to be based on an extremely broad interpretation of the jobs that a capable, efficient, and safe humanoid robot that does not currently exist is supposed to be able to do. Can the current reality live up to the promised scale?

What will it take to scale humanoid robots?

The physical creation of tens of thousands or even hundreds of thousands of humanoid robots is certainly possible in the near future. In 2023 Around 500,000 industrial robots are installed worldwide. Assuming that a humanoid robot is roughly equivalent to four industrial robots in terms of components, existing supply chains should be able to support even the most optimistic short-term humanoid production forecasts.

But simply creating robots is perhaps the easiest part of scaling humanoids, says Melonie Wisewho until this month served as chief product officer at Agility Robotics. “The bigger problem is demand: I don't think anyone has found a use for humanoids that would require several thousand robots per facility.” Large deployments, Wise explains, are the most realistic way for a robotics company to scale its business, since onboarding any new customer could take weeks or months. An alternative approach to deploying several thousand robots to do a single job is to deploy several hundred robots, each capable of 10 tasks, and this appears to be what much of the humanoid industry is betting on in the medium to long term.

Although among most humanoids there is a belief robotics industry that rapid progress in artificial intelligence should somehow lead to rapid progress in creating multi-purpose robots. No transparent how, when and whether this will happen. “I think a lot of people are hoping that they can find a way out of this with AI,” Wise says. “But the reality of the situation is that AI is currently not robust enough to meet market demands.”

Bringing humanoid robots to market

Market requirements for humanoid robots include a lot of extremely boring and extremely important things, such as battery life, reliability and safety. Of these, battery life is the simplest: in order for a robot to do its job usefully, it cannot spend most of its time charging. Next version of Agility's Digital robotCapable of supporting a payload of up to 16 kilograms, it includes a bulky “backpack” containing a battery with a charge ratio of 10 to 1: the robot can operate for 90 minutes and is fully recharged in 9 minutes. Thinner humanoid robots from other companies are bound to make compromises to maintain their slender form factor.

While in use, the Digit will likely spend a few minutes charging after 30 minutes of use. This is because Digit's 60 minutes of runtime is essentially a reserve in case something happens in its workspace that requires a temporary suspension, which is often the case in the logistics and manufacturing environments that Agility is targeting. Without the 60-minute reserve, the robot is much more likely to run out of energy mid-task and have to be recharged manually. Consider what this might look like with even a modest deployment of a few hundred robots weighing over a hundred kilograms each. “Nobody wants to deal with it,” Wise comments.

Potential customers of humanoid robots are very concerned about downtime. Over the course of a month, a plant operating with 99 percent reliability will be idle for approximately 5 hours. Wise says any downtime that shuts down something like a production line can cost tens of thousands of dollars per minute, so many industrial customers expect a couple more nines of reliability: 99.99 percent. Wise says Agility has demonstrated this level of reliability in some specific applications, but not in the context of multi-purpose or general-purpose functionality.

A humanoid robot in an industrial environment must comply with the general safety requirements for industrial machines. In the past, robotic systems such as autonomous vehicles And drones benefited from an immature regulatory environment to scale quickly. But Wise says this approach can't work for humanoids because the industry is already heavily regulated—the robot is simply considered another piece of machinery.

There are also more specific safety standards “Humanoid robots are currently being developed,” explains Matt Powers, associate director of autonomy R&D at the company. Boston Dynamics. He notes that his company helps develop International Organization for Standardization (ISO) safety standard for robots with dynamically balancing legs.. “We're thrilled that leading players in the space like Agility and Fig are joining us in developing a way to explain why we believe the systems we deploy are secure,” says Powers.

These standards are necessary because the traditional safety approach of reducing power may not be suitable for a dynamic balancing system. This will cause the humanoid robot to fall, which can make the situation even worse. There is no simple solution to this problem, and the initial approach Boston Dynamics plans to take is Atlas robot is to avoid situations where simply turning off the robot may not be the best option. “We're going to start with relatively low-risk deployments and then expand as we build confidence in our security systems,” Powers says. “I think the methodical approach will really win here.”

In practice, low risk means keeping humanoid robots away from humans. But humanoids who are limited in what work they can safely do and where they can safely move will have a harder time finding tasks of value.

Are humanoids the answer?

Before humanoid robots can scale, issues of demand, battery life, reliability and safety must be addressed. But the more fundamental question to ask is whether it's even worth spending the money on a bipedal robot.

Dynamic balancing using legs would theoretically allow these robots to navigate complex environments like a human. However, demonstration videos show these humanoid robots to be either mostly stationary or periodically moving short distances across a flat floor. The promise is that what we see now is just the first step towards human-like mobility. But in the short to medium term, there are much more reliable, efficient and cost-effective platforms that can take over in such situations: robots with arms, but with wheels instead of legs.

Safe and reliable humanoid robots could revolutionize the job market at some point in the future. But potential is just potential, and despite humanoid enthusiasm, we must be realistic about what it will take to turn potential into reality.

This article appears in the October 2025 print issue entitled “Why Humanoid Robots Don't Scale.”