Why U.S. Humanoid Robots Are Stuck on Actuators and Gears
The U.S. plan to build humanoid robots in large numbers hinges on actuators, motion parts that are hard to get from domestic suppliers.
That is the central point of an Oct. 2 report titled "To win the-humanoid robots race, the U.S. needs this critical machine component that's hard to find" MarketWatch. Apollo and similar humanoid platforms require actuators. Supply in the U.S. is tight.
An actuator is a compact motor and gear package that moves a joint. It sets joint torque, or turning strength, plus speed and precision. Strain wave gearing sits next to it in that drive chain. Think of it like a bicycle in low gear: many fast, weak turns go in, one slow, strong turn comes out. Schaeffler presented formed strain wave gearboxes for the mass market in early September. Strain wave gearboxes can transmit motion precisely with high torque capacity by using a high gear reduction ratio.
How they are made matters for cost. Forming parts into shape, rather than cutting them, points to faster cycle time and lower cost per unit. That fits volume production. For savers and investors, that is the difference between lab-grade precision gearing and a purchased component that can clear automotive or industrial PPAP-style qualification, the formal quality approval carmakers use, and scale to tens of thousands of joints.
Scale is still small. Global humanoid robot shipments nearly quadrupled to 19,100 units in the first half of the year, with Chinese manufacturers dominating shipments Reuters. That 19,100-unit base is where any U.S. ramp must start. It is also concentrated.
Chinese robot makers showed humanoids sorting parcels, packing mobile phones and helping with household chores at a Beijing conference in August Reuters. The task list spans logistics, light assembly and domestic use.
Two supply facts frame the U.S. position. China dominates refining of critical minerals needed for robot production Reuters. The Trump administration banned new Chinese robots and inverters to protect the U.S. AI buildout Reuters. China's comprehensive supply chain gives it an edge in lowering humanoid robot production costs significantly.
The broader context here is cost curve versus control. A 19,100-unit half year, dominated by one geography, leaves limited independent data on learning rates for actuators and precision gearing outside that supply chain. Learning rate means how fast costs fall as output rises. The Beijing demos show breadth of trial uses, not deployment at scale. If U.S. developers cannot source high-reduction, high-torque-density gearing and complete actuator assemblies at comparable yield, or share of good parts, unit cost stays high and delivery schedules slip. That hits capex planning, or spending on equipment, working capital tied up in pilot fleets, and the timing of any service revenue from deployed units.
In my view, investors should separate what is known from what is priced in. What is known is physical concentration: refining, component volume, and first-half shipments all lean toward China, while U.S. sourcing for the key motion components remains constrained. What is not known is how quickly formed gearing and other design-for-manufacture approaches close the cost and yield gap, and whether import restrictions accelerate domestic capacity or simply raise input costs during the build phase. Policy can redirect demand. It cannot create precision capacity overnight.
Looking at what this means for diligence, the questions are narrow. Who supplies the actuator stack, at what yield and lead time, or wait for delivery. Whether strain wave capacity is captive or merchant, meaning made in-house or sold to others, and under what qualification regime. How mineral refining exposure flows through magnet and motor pricing. Those answers will decide whether humanoids move from conference demonstrations of parcel sorting and phone packing to repeatable factory and logistics deployment with auditable unit economics, or costs and revenues per robot that can be checked.
