On April 17, 2026, McKinsey & Company published a detailed analysis of humanoid robotics economics titled “Turning humanoid supply-chain constraints into billion-dollar wins.” Its central thesis: the primary constraint on industry scaling today is not algorithms or foundation models but the supply chain for physical components — chiefly actuators, which account for nearly half of total cost. The current cost of a single humanoid ranges from $30,000 to $150,000 per unit. The target for mass-market viability, per McKinsey, sits below $20,000. Between these two points lies the need for a roughly fivefold cost reduction, and the report focuses on how to deliver it.
Context: 2026 marks the year humanoid robotics exits the laboratory phase and enters serial production. Chinese manufacturer AGIBOT shipped its 10,000th unit in March, taking the global lead in deliveries according to Omdia and IDC data for 2025. Unitree sells its G1 model starting at $13,500. Goldman Sachs reports humanoid manufacturing costs declined 40% year-over-year in 2025, against earlier projections of just 15–20%. The April McKinsey report is a sequel to October’s Crossing the chasm from concept to commercial reality, in which the same consultants mapped the operational scenarios where humanoids already work commercially.
Cost structure
McKinsey breaks down the bill of materials (BOM) of a humanoid robot into five core domains:
— Actuators — 40–60% of cost (motors, gearboxes, joint assemblies, integrated sensors and drivers);
— Sensing and perception — 10–20% (cameras, LiDAR, radar, tactile sensors);
— Compute and control — 10–15% (processors, accelerators, controllers);
— Mechanical structure — 5–10% (frames, shells, mounting hardware);
— Battery — 5–10%.
Together these five domains account for 85–90% of total unit cost. The remainder goes to cooling, wiring harnesses, and miscellaneous components. Specific component pricing from the analysis: a single harmonic drive actuator costs $2,000 to $5,000, with custom high-torque variants reaching $15,000. A robot with forty or more joints carries tens of thousands of dollars in actuation hardware alone.
McKinsey also singles out “hands and manipulation” as a strategically important sub-category, emphasizing that without a stable solution to this problem, commercial applications of humanoids remain limited regardless of how other component costs trend.
The scaling dilemma
The most expensive subsystem in a humanoid — the actuator — simultaneously has the least developed supplier ecosystem. McKinsey documents a tenfold gap between distributor list prices and should-cost manufacturing estimates derived from teardown analysis. That gap reflects neither material economics nor production complexity but a structural shortage of specialized suppliers and low production volumes.
The underlying contradiction is structural: suppliers will not commit capital to dedicated high-volume production lines for humanoid actuators because orders are measured in dozens, not thousands of units. But orders stay small because the high cost of low-volume components keeps the final product unviable for the mass market.
Engineering metrics compound the issue: current designs require more than 200 internal cables, and assembly of a single unit takes approximately four days. Neither is compatible with any serious volume production plan, which is why reducing assembly time is among the primary engineering targets for every leading player. Tesla addresses this through vertical integration — developing actuators and inference chips in-house, bypassing external suppliers altogether. McKinsey, by contrast, recommends industry-wide movement toward standardized actuator interfaces and form factors, allowing suppliers to produce unified modules serving multiple OEMs simultaneously.
China as structural advantage
McKinsey identifies China’s industrial base as the one case where the scaling dilemma is being overcome. The reason: direct overlap between the EV supply chain and humanoid robotics requirements. High-torque motors, power electronics, precision bearings, permanent magnets — all are already produced in China at EV-industry volumes orders of magnitude beyond current humanoid demand.
The specific figures:
— China produces 90% of permanent magnets and 40% of precision bearings and encoders used in humanoid robots;
— Per South China Morning Post analysis cited in the report, building Tesla’s Optimus Gen 2 without Chinese suppliers would cost roughly three times as much — the BOM would rise from approximately $46,000 to $131,000;
— China filed approximately 7,700 humanoid-related patents over the past five years, against ~1,560 in the United States and ~1,100 in Japan;
— Industrial clusters in Hangzhou, Ningbo, Shenzhen, and Suzhou allow Chinese OEMs to source motors, CNC housings, printed circuit boards, gearing, sensors, and batteries within an hour’s radius of any production site.
McKinsey emphasizes that China’s advantage is not driven by cheap labor. It is a structural advantage rooted in supplier clustering, EV-industry economies of scale, and component-base maturity. Reproducing this configuration outside China is possible, but it requires state-level coordination and massive capital investment.
What to watch
— Will an industry standard for actuator interfaces emerge? — Without one, scaling economics will not engage. Likely centers for standardization: IEEE, Japan’s AIRoA, Chinese industry associations.
— How will Western industry respond to Chinese component dependence? — Three scenarios are possible: vertical integration along the Tesla model, formation of alternative supply clusters (Japan, Korea, Germany), or de facto acceptance of Chinese dependence for certain component classes.
— Where do Japanese players land in this picture? — Honda with its newly announced multi-fingered hand at Humanoids Summit Tokyo, Toyota with its Humanoid Research Unit — Japan has the engineering base for an alternative supply chain, but no commercial timelines have been announced yet.
— What will McKinsey add after Tokyo Summit? — McKinsey partner Ani Kelkar speaks at the summit on May 28–29. A follow-up will cover the substance.