Europe makes world class robots on semiconductors it does not make.

European robotics is genuinely strong. Collaborative arms from Odense, industrial systems from Bavaria and Baden-Württemberg, motion control from Switzerland and Sweden. The engineering is not the vulnerability.

The vulnerability is two layers down. Motor controllers, field programmable gate arrays, vision processors and the power electronics that sit between them are sourced from a supply chain with critical concentration outside Europe.

What the chip policy actually delivered

European semiconductor policy has added leading edge logic capacity and, more usefully for robotics, power and analogue capacity. Robotics does not primarily need the most advanced process node. It needs reliable supply of mature node components with long product lifetimes.

That is the quieter success of the past three years, and it matters more to a robot builder than any announcement about advanced logic.

The remaining exposure

Advanced packaging and test remain concentrated in Asia. A European-fabricated die that is packaged elsewhere carries most of the same geographic risk as a fully imported component, a point that supply chain announcements consistently understate.

Vision processing is the second exposure. The specialist silicon that makes machine vision economical at the edge comes from a small number of suppliers, and robot builders have limited ability to design around them.

What builders are doing

Three responses recur. Dual sourcing at the component level, accepting a bill of materials penalty for supply certainty. Designing for component substitution, so the controller board can accept alternates without recertification. And carrying more inventory than lean orthodoxy permits, a permanent change since the shortages of the early decade.

The honest position

Europe has world leading capability in a few upstream links, notably lithography equipment, and in the downstream application layer. The middle of the chain is where dependency sits, and no realistic policy closes that in a decade.

Recognising this produces better industrial strategy than pretending otherwise. The productive question is which dependencies are tolerable and which are not.