Vangrid raises $9M for Physical AI maps
Vangrid raised $9M to build a decentralised spatial-intelligence network for Physical AI.
Physical AI does not only need smarter robots. It also needs better maps of the real world — constantly updated, verifiable and detailed enough for machines to understand.
What happened
Dutch startup Vangrid raised a $9M seed round to build a decentralised spatial-intelligence network using smartphone cameras and sensors.
The company is trying to create high-fidelity spatial data for Physical AI systems, with verification, provenance and privacy-preserving capture built into the network. Instead of building the robot itself, Vangrid is focusing on the data layer those robots and autonomous systems may eventually need.
Why it matters
Physical AI depends on reliable real-world context. A warehouse robot, delivery vehicle, industrial inspection system or future home robot all need to understand where objects are, how spaces change and whether the data they rely on can be trusted.
That makes spatial intelligence a useful infrastructure layer. If Vangrid can turn ordinary smartphone sensors into a distributed mapping network, it could sit upstream of many robotics and autonomous-system applications.
The bigger picture
The Physical AI stack is starting to separate into layers: hardware, models, sensors, simulation, deployment software and real-world data. Vangrid is a reminder that the less glamorous data-provenance layer may become just as important as the robot people can actually see.