A manufacturing industry figure has made the case that robot scaling depends primarily on factory capabilities rather than algorithmic advances. This observation holds merit, yet it glosses over a more fundamental limitation: every motorized joint in a humanoid robot requires a rare earth magnet sourced from Chinese refineries.
Actuators represent between 40% and 60% of the total cost structure in humanoid robots, according to McKinsey research. China's dominance in refining rare earth magnets—controlling approximately 90% of global processing capacity—transforms these components into the actual ceiling for expansion. Meanwhile, Europe's Critical Raw Materials Act establishes a 2030 ceiling limiting any single non-EU nation to supplying no more than 65% of the continent's consumption. Currently, only Schaeffler operates as a significant actuator manufacturer at scale within Europe.
The assertion that hardware rather than intelligence represents the true bottleneck carries weight. Actuator availability remains constrained by a narrow supplier base, and establishing new manufacturing capacity demands years of development rather than months. "The hardest problem in physical AI is not the AI," wrote Nate Evans in Unite.ai. Evans, who cofounded Fictiv—a manufacturing sourcing platform acquired by Japanese components supplier MISUMI for $350 million—now leads MISUMI AI, lending particular credibility to his perspective.
The analysis becomes more precise when examining the actuator's internal structure. Gearboxes consume between 30% and 50% of an actuator's value, creating a critical chokepoint. The gearbox supply chain depends on a limited roster of manufacturers: Harmonic Drive, Nabtesco, and a handful of others. Qualification timelines for these components stretch across years.
The neodymium magnet embedded within each high-torque motor joint represents an even tighter constraint. China processes roughly 90% of rare earth magnets globally. While China extracts approximately 69% of rare earth ore—a less concentrated figure—the refining stage is where the actual bottleneck emerges.
European policymakers have begun addressing this vulnerability. The Critical Raw Materials Act explicitly targets this issue, placing both light and heavy rare earths on the strategic materials list. The legislation sets a 2030 deadline for ensuring no single third country supplies more than 65% of European consumption. Additionally, Articles 28 and 29 of the act directly reference industrial robots, mandating that magnets weighing more than 0.2 kilograms carry labeling, digital documentation, and recycled content information beginning in 2028.
Europe's industrial response remains limited to a single major player. Schaeffler has invested in Humanoid, the London-based startup that secured $152 million in funding during July. Bosch manufactures robots, while Schaeffler supplies the joints. Under an agreement extending through 2031, Schaeffler serves as the preferred supplier for more than half of Humanoid's joint actuators, covering a seven-digit unit volume.
Schaeffler is simultaneously expanding its own robot deployment. The company plans to install a four-digit quantity of robots across its facilities by 2032, with initial deployment starting in December at its Herzogenaurach plant. Some observers contend that Europe could achieve competitive advantage in robot deployment despite China controlling 63% of the hardware supply chain.
Yet this argument reaches its limits when confronted with physical reality. An actuator assembled in Herzogenaurach still incorporates a magnet refined in China, and no amount of supplier diversification alters this dependency. The challenge transcends purchasing strategy and enters the policy domain. Europe possesses a 65% supply cap dated 2030, a labeling requirement dated 2028, and virtually no domestic rare earth refining infrastructure to satisfy either mandate.
Source: The Next Web
