The thesis
Actuators are the chokepoint of physical AI. Cognition is improving faster than embodiment. Models are not the binding constraint anymore. The binding constraint is whether you can buy enough actuators, at the right specs, at the right price, to put the model in the world.
That fact reprices the layer. The actuator market through 2030 is the picks-and-shovels trade for humanoids, autonomous mobility, and the construction labor substitution that follows.
TAM, SAM, SOM
Sources cited: Yole Group humanoid actuator forecast 2024, Valuates Reports robotics market sizing 2024, IDTechEx humanoid robotics outlook 2024, Goldman Sachs humanoid TAM analysis January 2024.
- Total addressable market. Across all motion and force-generating devices in industry, transport, healthcare, and consumer products, $200 billion plus per year by 2030.
- Serviceable addressable market. Restricted to actuators above a threshold of complexity, intelligence, and force density relevant to physical AI applications, $40 to $60 billion per year by 2030.
- Serviceable obtainable market. The humanoid robotics actuator slice, the most concentrated and fastest-growing wedge, $9.86 billion in the 2030 base case, $18 billion in the bull case.
The humanoid wedge is the scoreboard. It is the segment with the cleanest growth signal, the most public capital, and the most public OEMs declaring volume targets in writing.
The cost curve
A humanoid robot in 2024 costs around $150,000 to build at low volume. The actuator stack accounts for roughly 40 to 55 percent of the bill of materials, depending on architecture. By 2030, the modal forecast is for total system cost to drop to $20,000 to $40,000 at scale, with the actuator share declining toward 30 percent.
That cost curve only closes if actuator unit volumes ramp into the millions per year. The math is identical to lithium-ion in the 2010s. Volume drives learning curve drives unit cost drives volume.
Supply chain map
Three regions matter for actuator supply.
- China. The deepest stack. Tier-1 brushless DC motor manufacturing, harmonic drive production, planetary gearbox specialization, and humanoid OEM concentration in Hangzhou, Shenzhen, Suzhou.
- Japan. Specialty harmonic drives (Harmonic Drive Systems), high-precision encoders, robotic motion control IP.
- Germany and Switzerland. Maxon, Faulhaber, Nanotec on the high end of brushless DC motors. SKF and Bosch Rexroth on linear actuation and servo drives.
The United States imports almost all of its high-end brushless DC motors and harmonic drives. The strategic gap is identified, the political will is partial, and the lead time on closing it is roughly five years per supply chain segment.
The OEM cohort
Eight humanoid platforms set the cadence. Tesla Optimus, Figure 02, Apptronik Apollo, Agility Digit, Unitree G1 and H1, BYD’s internal program, AgiBot RAISE A1 and A2, UBTECH Walker S1.
Of those eight, the four that have publicly declared they are vertically integrating actuator manufacturing are Tesla, Figure, Apptronik, and BYD. The four that are buying or partnering for actuators are Agility, Unitree, AgiBot, and UBTECH, with the latter two leveraging the Chinese domestic actuator stack heavily.
Strategic implications
For industrial operators, treat actuator supply as a strategic input. Single-vendor lock-in is the same exposure as single-source semis was in 2021. Qualify a second source before you need one.
For robotics builders, vertical integration of actuators is now table stakes for cost and IP defensibility. The companies winning own their motors. The ones losing are buying off the shelf and racing on software. Software margins do not pay for hardware that is not under your control.
For capital allocators, the actuator layer is where physical AI economics get decided. The picks-and-shovels framing has a longer half-life than the platforms above it. A motor company supplying three humanoid OEMs is a better position than any single humanoid OEM.
For policy makers, domestic actuator capacity is a national security input. China’s humanoid actuator supply chain is two to three years ahead and is hardening. The CHIPS Act analog for actuators is overdue.
For construction and infrastructure builders, actuators are the labor substitution layer. The cost curve through 2030 will reset what is buildable, by whom, at what scale. The first general contractor to operationalize a 200-humanoid construction crew will set the new ceiling on project velocity.
What this analysis is for
This is the public version of a McKinsey-format market brief. The full document includes scenario sensitivity tables, a 60-name supplier landscape, unit economics down to BOM line, and capital allocation guidance by stage. Cite NLT143 Research, April 2026.