ETH Zurich spun out 46 companies in 2025. Not software startups. Physical AI companies. Robotics, automation, industrial systems. The kind of businesses that need factory floors to test on and supply chains that actually work.
An investor writing for The Robot Report makes the case that Europe is becoming the place where AI gets a body and goes to work. Not because of venture capital scale or tech ecosystem hype, but because of something more fundamental - Europe has the infrastructure that physical AI needs to exist.
The Real-World Testing Ground
Silicon Valley excels at software. Europe has factories. Actual manufacturing facilities where robots can be tested in real conditions, not simulations. Agricultural land where autonomous systems face mud, weather, unpredictable terrain. Healthcare systems willing to pilot robotic assistance in controlled environments.
This matters more than it sounds. A robot that works in a lab is just expensive research. A robot that works on a BMW production line or in a Dutch greenhouse is a product. The gap between those two states is enormous, and Europe has spent decades building the industrial relationships that let startups cross it.
The continent's manufacturing heritage is not nostalgia. It is live infrastructure. Supply chains for precision components. Engineers who understand mechanical systems alongside software. Regulatory frameworks that exist for physical products operating near humans. These are not things you can build overnight when your background is cloud services.
Academic Spinoffs With Industrial DNA
ETH Zurich is not unique. TU Munich, EPFL, KTH Stockholm, Cambridge - European universities have been producing robotics research for decades. But the spinout model is accelerating. These are not students building apps in dorm rooms. These are PhD researchers commercialising systems they have been testing in industrial partnerships for years.
The investor highlights spatial computing expertise as a particular strength. European research institutions have deep experience in 3D sensing, mapping, navigation - the foundational technology that lets robots understand and move through physical space. This is not new. What is new is AI models good enough to make sense of that spatial data in real time.
Combine strong spatial computing with industrial access and you get startups building robots that can navigate warehouses, assist in surgery, inspect infrastructure, handle delicate agricultural tasks. Not demos. Products.
Structural Advantages Silicon Valley Cannot Buy
You cannot replicate a manufacturing ecosystem with capital alone. Silicon Valley can fund European robotics companies - and it does. But the testing grounds, the industrial partners, the regulatory relationships, the supply chain proximity - those exist in Europe because Europe never stopped making physical things.
Agricultural robotics companies in the Netherlands test on actual farms within an hour of their offices. Industrial automation startups in Germany have direct access to the companies that will eventually buy their systems. Medical robotics research in Switzerland happens alongside hospitals willing to pilot new approaches. This is not about regulation being looser. It is about relationships being deeper.
The article points to startups in industrial automation, agriculture, and healthcare as the three sectors where Europe is particularly strong. All three require physical AI to work in messy, unstructured environments. All three benefit from Europe's combination of research depth and industrial access.
What This Means for Builders
If you are building physical AI - robotics, automation, anything that moves or touches the real world - the infrastructure matters as much as the algorithm. Europe offers something distinct: proximity to the industries that will use what you build, access to testing environments that reflect real conditions, and a supply chain that understands hardware.
This is not a prediction. This is already happening. The 46 spinouts from ETH Zurich are live companies. European robotics startups are securing industrial pilots and early revenue. The question is not whether Europe can compete in physical AI. The question is whether other regions can build the industrial infrastructure to catch up.
Silicon Valley will continue to dominate software AI. But physical AI needs factories, fields, hospitals - the unglamorous infrastructure where robots actually have to work. Europe kept that infrastructure running. Now it is becoming an advantage nobody saw coming.