
Where does the money actually go when building a humanoid robot? Morgan Stanley’s analysis of Tesla Optimus Gen 2 reveals a striking pattern.
Hands and legs together consume 56% of the entire bill of materials.
This single insight explains why humanoid robotics is fundamentally different from other hardware businesses—and why it’s so hard to scale.
The Distribution
Legs (thigh + calf + feet): 38.6% — $21,300 of the ~$55,000 total. Eight linear actuators, planetary roller screws, and 6D force sensors work together to handle the dynamic loads of bipedal locomotion.
Arms + shoulders: 29.5% — $16,300. Ten rotary actuators with harmonic reducers enable the range of motion needed for manipulation tasks.
Hands: 17.2% — $9,500. The single most expensive component relative to size. Twelve coreless motors, planetary reducers, and precision encoders packed into a human-sized form factor.
Head + torso: 14.7% — $8,100. This includes the FSD chips, cameras, battery, and thermal management. The “brain” of the robot is the cheapest part.
Why This Matters
The cost concentration tells us where engineering innovation will have outsized impact.
Smartphones scale through chip fabrication improvements. Electric vehicles scale through battery chemistry breakthroughs. But humanoids must scale through precision mechanical systems—and there’s no Moore’s Law for gearboxes.
Actuators account for 47-50% of total humanoid costs. Each robot needs 28 of them, performing precise movements under variable loads. The dominant suppliers—Maxon Motor, Harmonic Drive, Kollmorgen—maintain premium positioning through superior precision: less than 4 arcminutes of backlash, 97% efficiency.
Chinese competitors offer alternatives at 5x lower cost, but established suppliers still lead in reliability and precision at the tolerances humanoids require.
The Strategic Implication
The 56% concentration in hands and legs creates a clear strategic map.
This is where the race will be won or lost. Companies that crack low-cost, high-reliability actuators for extremities will dominate. Those that don’t will remain dependent on constrained supply chains.
Tesla’s approach: design custom actuators from scratch, vertically integrate, and leverage automotive manufacturing expertise. The company acknowledges that no existing components meet volume production requirements.
The cost distribution chart is a roadmap. Hands + legs = 56% of BOM. This is where engineering innovation wins the physical AI race.









