Inside the Humanoid Robot BOM: Actuators, Reducers, and the Wiring That Connects Them

A China-built humanoid robot carried a bill of materials (BOM) of roughly **$35,000 per unit at the end of 2025**, and analysts at McKinsey and Bank of America expect that figure to fall to **$13,000–$17,000 by 2030–2035** as volume scales. Most of that cost sits in one subsystem: the actuators that move the joints. Far less attention goes to the wiring that links those actuators to power and control — yet that wiring is where a surprising share of field failures begin.

This article breaks down where the money actually goes in a humanoid robot, why harmonic reducers and planetary roller screws dominate the actuator bill, and where high-flex cable assembly fits into the same BOM. We build robot cable assemblies; we do not make reducers or screws. That boundary is exactly why the wiring layer is worth a clear-eyed look.

What's actually in a humanoid robot's bill of materials?

Actuators are the dominant cost line, accounting for **40–60% of the total BOM**. In a high-configuration humanoid with dexterous hands, joint actuators alone exceed 30% of BOM; in simpler models without hands or rich sensing, they can pass 50%. A common reference design uses **16 rotary actuators and 14 linear actuators** — roughly 30 powered joints in a single machine.

Inside each actuator, the gearbox (the reducer or screw) represents **30–50% of the actuator's cost**. So the mechanical transmission elements are the single biggest swing factor in the whole robot's cost structure. Sensors, compute, and the structural frame split most of the remainder. The wiring harness is a smaller line by dollar value, but it is the connective tissue: every actuator, encoder, and tactile sensor needs power and signal routed across moving joints.

Why harmonic reducers dominate the rotary joints

Harmonic (strain-wave) reducers are the default for rotary joints because they deliver high gear ratios in a compact, lightweight package — exactly what a robot shoulder, elbow, or wrist needs. The trade-off is sensitivity to back-driving forces, such as the shock load when a humanoid falls.

China's largest harmonic-reducer maker, **LeaderDrive (Leader Harmonious Drive Systems, 688017)**, shows how fast this segment is moving. J.P. Morgan estimates LeaderDrive holds **30–40% of China's harmonic-reducer market** — the domestic leader, though notably below the inflated “60%+” figure that circulates on social media. In 2025 the company reported revenue of **570.7 million RMB (up 47%)** and net profit that more than doubled to **124.4 million RMB**. Its customers include humanoid developers such as **Agibot and UBTech**.

Planetary roller screws and the linear actuators

Linear joints — hips, knees, and ankles that must carry large torques — increasingly use **planetary roller screws** rather than ball screws. According to Interact Analysis, joint actuators are the fundamental cost-and-performance component of humanoids, and Morgan Stanley expects roller screws to become the dominant linear-actuator transmission because they handle higher loads and wear better under constant cycling.

China's supply chain moved early here too. Industry research tied to Tesla's Optimus program suggests Chinese suppliers — led by names like Tuopu, Sanhua, LeaderDrive, and Wuzhou Xinchun — secured roughly **70% of the Gen-3 component share** across reducers, screws, and motors, leaning on automotive-supply-chain know-how and cost. The pattern repeats: the expensive headline components are mechanical, and they keep pushing more powered joints into the same body — which multiplies the cables that must survive in motion.

The part of the BOM nobody benchmarks: the wiring

A humanoid robot has **20+ articulated joints**, and every one forces power and signal lines to bend and twist — millions of times over the machine's service life. The harness must do this inside tight, weight-constrained limbs while keeping noisy power lines away from low-level sensor signals.

Tactile and force-feedback sensing make this harder. As control moves into the hands and fingertips, a single noisy or intermittently open conductor stops being a cosmetic defect and becomes a control fault. High-flex cable assemblies for these platforms need controlled strip lengths, consistent crimps, clean conductor preparation, and reinforcement (aramid yarn or braided shields) that add flex life without adding stiffness.

A humanoid actuator can be rebuilt at a bench. A conductor that fatigues inside a sealed wrist after 800,000 cycles fails in the field, in front of a customer. The wiring is the cheapest line in the BOM and the most expensive place to cut corners.

— Hommer Zhao, General Manager and Wire Harness Engineer

None of this requires us to make a reducer or a screw. It requires the harness that ties 30 actuators, their encoders, and their sensors into one reliable, serviceable system — the robot actuator cable assembly and power-distribution harness work that sits right next to the mechanical bill of materials.

Where each component sits in the BOM

The table makes the strategic point: the mechanical components set the cost, but the harness sets much of the reliability. Cutting harness cost to save a few dollars on a $35,000 machine is the classic false economy — it shifts spend from the factory to the field-service budget.

What this means for OEMs sourcing the build

If you are sourcing a humanoid or advanced service-robot program, separate the two buying problems. The reducer and screw decision is a mechanical-engineering and cost negotiation with specialist suppliers. The wiring decision is a reliability-and-integration problem best solved with a cable assembly partner early in the design — before joint geometry, bend radius, and connector placement are locked.

We are firmly in the second camp: a cable assembly and integration partner, not a component reseller. The earlier the harness routing, sensor-signal cabling, and strain relief are reviewed against the joint design, the fewer late-stage surprises hit the humanoid robot program.