Actuators and advanced sensors now the biggest cost for humanoids, 1X says
- 1X’s new Hayward factory puts the cost story in plain view: humanoid prices are now bottlenecked less by AI chips than by joints and perception. - McKinsey pegs actuators at 40% to 60% of a humanoid’s bill of materials, while 1X says it sold out 10,000 NEO units in 5 days. - That matters because the next race is manufacturing scale — not demos — and whoever cheapens motion hardware fastest gets to mass market first.
Humanoid robots are starting to look less like an AI story and more like a factory story. That’s the big takeaway from 1X kicking off full-scale NEO production in Hayward on April 30. The hard part is no longer just getting a robot to walk, see, and pick things up in a demo. The hard part is building the expensive physical guts — especially actuators and advanced sensing — cheaply enough that a humanoid can become a real product. (1x.tech) ### What changed this week? 1X said its 58,000-square-foot California plant has started full-scale production of NEO, its home humanoid, with 200-plus employees already on site. The company says it booked out a full year of production — 10,000 robots — in five days after opening preorders, and still plans first shipments in 2026. That is a useful reality check: demand might be there, but now the bottleneck is manufacturing. (1x.tech) ### Why are actuators such a big deal? Actuators are basically the robot’s muscles and joints — motors, gearboxes, transmissions, torque control, and the sensing wrapped into those assemblies. In a humanoid, you need a lot of them, and they all have to be strong, precise, quiet, efficient, and safe around people. McKinsey’s latest supply-chain breakdown puts actuators at 40% to 60% o(1x.tech)re bucket. (mckinsey.com) ### Aren’t sensors expensive too? Yes — and that’s the second half of the problem. Humanoids need cameras, depth sensing, force feedback, joint encoders, tactile sensing, and all the wiring and compute that make those signals usable in real time. McKinsey puts sensing and perception at another(mckinsey.com)d before the model can act on it. (mckinsey.com) ### Why doesn’t scale fix this quickly? Because these are not commodity laptop parts. A humanoid actuator has to survive constant motion, absorb shocks, hit tight tolerances, and still be manufacturable by the tens of thousands. That is closer to building a drivetrain than assembling a gadget. (mckinsey.com)ement electronics, and even soft outer materials in-house to control cost and iteration speed. (forbes.com) ### What does 1X think the answer is? Make more of the stack yourself, then automate the factory around it. 1X says the Hayward plant has already produced 17,000 motors and is built around its in-house Revo2 motor and tendon-driven actuation system. Basically, the company is treating motion hardware the way EV makers treated battery packs a few years ago — as the cost center you have to own if you want margins and scale. (forbes.com) ### Why do preorders matter so much? Because preorders are not just marketing. They tell a hardware company how many units to tool for, where customers are, and how aggressively to lock in suppliers. 1X says the preorder launch was crucial to production planning, not just buzz. If the expensive parts are actuato(forbes.com)cale. (1x.tech) ### Is this just a 1X problem? Not really. Apptronik just raised its total Series A to more than $935 million to scale Apollo production, which tells you investors also see manufacturing as the real choke point now. Across the sector, the question is shifting from “can the robot work?” to “can the supply chain support it at a sane price?” (apptronik.com)) ### So what’s the bottom line? Humanoids are entering the same phase EVs and drones went through: the wow factor is moving from software demos to costed, repeatable hardware. If actuators stay dominant and sensing stays pricey, the winners will be the companies that industrialize those parts first — not just the ones with the flashiest robot videos. (mckinsey.com)
