The Hardest Problem in Physical AI: The Magnet, Not the Model
A manufacturing executive argues that scaling robots is a factory problem rather than an AI one, which is true and stops short of the binding constraint because the motor in every joint needs a rare earth magnet that China refines.
August 31, 2026 – 9:00 pm
AI Learning and Artificial Intelligence
Actuators account for 40% to 60% of a humanoid robot’s bill of materials, and China refines roughly 90% of the rare earth magnets inside them. (Nate Evans, Unite.ai)
The Shortage: Actuators and Rare Earth Magnets
The hardest problem in physical AI is not the AI itself. Actuators are in short supply from a small pool of qualified manufacturers, with new capacity taking years, not quarters.
These actuators, which make up 40% to 60% of a humanoid robot’s materials, rely on neodymium magnets—a key component sourced heavily from China.
The European Perspective
Europe has recognized this problem through the Critical Raw Materials Act, setting a 2030 cap on single-country supply for certain rare earths and strategic raw materials, including light and heavy rare earths.
Articles 28 and 29 of the act specifically name industrial robots, with magnet requirements coming into effect by 2028.
A Limited Supply Chain
Currently, Europe has one primary company addressing this challenge: Schaeffler, which invests in and supplies key components for both humanoid robot manufacturers (like Humanoid) and assembles its own robots.
Despite European efforts like the CRMA, Schaeffler acknowledges that an actuator assembled in Germany still contains magnets refined in China due to limited refining capacity within Europe.
The Policy Gap
The gap lies in policy: While Europe has set targets for supply chain independence, it lacks domestic refining capacity to meet these goals.