By Tony Seba, Adam Dorr and Bradd Libby
The robots are coming, and things really are different this time. This technology is genuinely something new under the sun. But not everything about the imminent disruption of labor by humanoid robotics is novel. We’ve seen versions of this movie before. We can use the Seba Technology Disruption Framework (STDF) to predict how the story will play out.
The STDF, which draws from data around hundreds of technological transformations throughout history, explains that disruptions occur when a new technology or set of technologies emerge that let us produce goods and services which offer far greater value than ones produced with older technologies. New disruptive products enter the market being competitive either on cost (a disruption from below), on capability (a disruption from above) or both (a “Big Bang” disruption). But the cost capability of these new products doesn’t stand still—it continues to improve as the underlying technologies are refined and manufacturing scales up. Over time, the value proposition of a disruptive product becomes overwhelmingly competitive, and as a result virtually all users end up adopting it within 15 years of launch in their area. This is why—from the perspective of an incumbent industry in the crosshairs of new technology—disruptions appear to occur so swiftly.
Many disruptions are triggered by the convergence of multiple technologies. Smartphones, for example, were enabled by a convergence of advancements in batteries, sensors, cameras, touchscreens, computer processor chips and memory, wireless communications and software, among others.
The same is now true of humanoid robots. A convergence of similar advancements in batteries, sensors, cameras, computer processor chips and memory, motors, actuators and software (most especially artificial intelligence) has just begun to enable the deployment of general purpose humanoid robots capable of doing useful work.
Taken together, the component technologies add up to an overall cost capability that we can now begin to measure and track over time with the metric of dollars per hour for their machine labor. Charting the cost capability metric of a disruptive technology as a time series yields a cost curve.
In the case of humanoid robotics, the cost capability metric is cost per hour of labor ($/hour).
To estimate the cost capability metric for humanoid robots, we must divide the total lifetime costs (comprising capital expenditures or capex, operating expenditures or opex, financing and any other soft costs such as permitting) by the number of useful hours of labor the robot performs during its lifetime.
Capex is comprised of all component hardware and software technologies that have converged to enable useful humanoid robots.
Notably, the opex of humanoid robots is likely to be very low from the start and to trend quickly toward near-zero as the disruption proceeds. This means the marginal cost of labor will also approach zero—a profound implication we will focus on in a future blog post.
Today, just prior to any significant deployment of humanoid robots, we estimate lifetime costs are unlikely to exceed $200,000 per unit, and could well enter the market from some manufacturers below $40,000.
Humanoid robots will be capable of working at least 20 hours per day, and perhaps almost nonstop if they have swappable battery packs or tethered power. And apart from minimal downtime for maintenance, they will work without vacation or illness or complaint.
A reasonable estimate of productive uptime is perhaps 7,000 hours per year. Just as importantly, robots can be actively productive for the entirety of those 7,000 hours. Unlike human beings, robots on the clock will not get distracted, procrastinate, check their phones, take bathroom or cigarette breaks, or otherwise shift the focus of their attention away from their work.
We expect initial units to have a financial working life of just three years. They could certainly function for longer, just as the first digital cameras and smartphones technically could, but given how rapidly they are likely to become obsolete in the early years of the disruption (just as cameras and phones did) it is more reasonable to assume they will be decommissioned and either upcycled or recycled. This gives roughly 20,000 lifetime hours as the denominator for the first generation of humanoid robots.
Taken together, we estimate early data points for the humanoid robotics cost curve to start sometime in 2025 at between $2 and $10 per hour. At this early stage, our rough estimate is that the cost capability of humanoid robots will improve by at least one order of magnitude every eight years—although possibly a great deal more.
At first, humanoid robots will only be able to perform relatively simple tasks, but their capabilities will grow rapidly.
Labor is one of the primary factors of production, and ultimately the limiting factor. Regardless of the industry or economy or nation, and regardless of the level of skill required, tasks that can only be performed by human beings are limited by the number of humans available to do them.
By the end of the 2030s, robots are likely to be performing as much total labor as human beings. During the 2040s, economic productivity will explode worldwide as the cost of labor falls toward zero while the supply expands by at least one and perhaps two or even three orders of magnitude.
There are two crucial and related facts to acknowledge here:
The disruption of labor has staggering implications, and it is therefore imperative that we embrace the immense potential that humanoid robots offer but also proactively address the challenges they present.
We’ve seen stories like this play out before. Cost curves are one of the tools in the STDF that can be used to understand and prepare for the radical transformations that lie ahead. The advent of electricity and nuclear weapons offer instructive historical examples of the level of national investment, policymaking and planning that is now called for in response to humanoid robotics.
Any country which fails to go all-in on humanoid robots will look like it is standing still next to peers whose annual GDP-equivalent growth could exceed 10% by the early 2030s, and 100% (yes, an annual doubling of the national economy) by the late 2030s. Seldom in human history has a disruptive new technology offered such a radical expansion of productivity, or such a comprehensive threat to the socioeconomic status quo.
The future that awaits us is one where the ingenuity of humans and the efficiency of machines will intertwine, creating a world of possibilities limited only by our imagination and our capacity to adapt to this new paradigm of labor and production.
This is the second blog in our series on humanoid robotsics.
