Humanoid Robot Production Is Straining Suppliers Amid US-China Tensions

Humanoid robots are slowly ceasing to be demo units, and the first to feel it are not assembly plants but suppliers. According to AI Business, rising production volumes are straining component supply, and tensions between the US and China may leave Western manufacturers without a fallback: finding replacements for Chinese suppliers is getting harder. The original report gives no specific figures. The industry context, however, is well known, and it explains why the problem is real.

Where exactly the squeeze is

A humanoid is built from dozens of actuators. Every joint contains a motor, a gearbox, a torque sensor, bearings and control electronics. Linear joints need planetary roller screws, and the hands need micro-actuators. For a long time, these parts were made in small batches for industrial automation. The jump from hundreds of machines to thousands breaks that model. Machine tools, tooling and skilled production lines do not scale in a single quarter, so a supplier's queue becomes a real constraint.

Magnets are a separate story. China produces an overwhelming share of the world's sintered neodymium magnets, usually put at around 90%. In April 2025, Beijing introduced export licensing for a number of rare-earth elements, including samarium, gadolinium, terbium, dysprosium, lutetium, scandium and yttrium. Elon Musk said at the time that the licensing requirement also affects Tesla Optimus. If a single robot uses dozens of actuators with magnets, any hiccup in supply immediately turns into a stopped line.

Why there is no alternative

Chinese humanoid makers are growing together with their supply chain. Suppliers are nearby, a prototype becomes a production run faster, and prices fall: the Unitree G1 launched in 2024 at a price of around $16,000. Western companies lack such a short loop. They do have Japanese and European makers of gearboxes and bearings, such as Harmonic Drive Systems and Nabtesco. But those capacities were built up over years for industrial robots, not for mass production of humanoids. Additional lines cost money and take time, and trade tensions make multi-year planning even less reliable.

According to industry estimates, the most sensitive components are these:

• rare-earth magnets for joint motors;
• harmonic and planetary gearboxes;
• planetary roller screws for linear actuators;
• torque and force sensors, power electronics;
• traction batteries and their cells.

This list does not come from the source; it is a typical risk map for robotics. But these are exactly the places where shortages usually appear when demand grows faster than factories expand.

How to live with it

For manufacturers, the takeaway is pragmatic. Multisourcing is needed at the design stage, not after the first missed delivery. Designs with a lower content of heavy rare-earth elements, for example magnets that drop dysprosium and terbium, provide a safety margin. Long-term supplier contracts and buffer stocks of critical parts are expensive, but cheaper than a shutdown.

Investors and corporate customers can use a simple filter. When a company announces plans to ship thousands of robots, it should explain where the actuators, magnets and gearboxes will come from, and what happens if even one supplier is late. Without a clear answer, volume figures remain a promise. The bottleneck in this industry has already shifted from algorithms and models to hardware, and the winners will be those with the more reliable supply chain.