Stockholm’s IPercept nets $16.5M Series A to bring physical AI to factory floors

IPercept has pulled in a $16.5 million Series A to do something industrial software has largely ignored: watch CNC machines closely, without plugging into the controller. The Stockholm startup says that matters because these machines sit at the heart of manufacturing, yet many have been running with little or no monitoring since the 1940s.

Isogon Ventures and 2150 led the round, with backing from Luminar Ventures, RunwayFBU, J12 Ventures, and AI.Fund. The company says it already works with more than 30 customers, including Airbus, Bosch, Hitachi Energy, Scania and Volvo, and that its team has grown to more than 20 people. The new money will help it hire in the US for the first time.

IPercept’s pitch is built around a stubborn industry problem rather than a flashy AI breakthrough. Founder and CEO Karoly Szipka says major Swedish manufacturers first tried to solve CNC monitoring by hiring data scientists, only to find the real issue was domain knowledge. He has a PhD in CNC machine diagnostics from KTH Royal Institute of Technology, and he co-founded the company there in 2019 with Andreas Archenti, a professor at KTH. The product hit the market in 2022.

The device mounts on the moving part of a CNC machine and measures motion at the micrometre level. It does not need access to the machine controller or the customer’s IT network, and IPercept says it works across roughly 300 CNC manufacturers. Installation takes only a few hours. Szipka compares it to a full-body MRI for industrial equipment, which is a more memorable image than most factory tech gets.

The timing is decent. Predictive maintenance is projected to reach $97.37 billion by 2034, and CNC machines make up a meaningful slice of that market. IPercept says the installed base is worth about $1 trillion worldwide, and it wants a piece of that by making older machines easier to understand, not by replacing them.