Fundraise

Italy's Mosaico raises €7.5M to fix robotics' data bottleneck

What's the deal? Italian startup MosaicoDealroom has a profile for this one. Try Dealroom → has raised €7.5m in seed funding to help companies manage the vast amounts of sensor data needed to develop AI-powered robots, Sifted reported. Based in Parma and founded in 2023, the company builds an open-source, on-premise data platform for robotics and physical AI.

What's the endgame? Mosaico ingests large volumes of multi-sensor data — ROS bags, MCAP files, camera, LiDAR, radar, IMU, and GPS — and turns it into structured, training-ready datasets. Features for data ontology, multimodal search, and data lineage let robotics teams find and reuse specific sensor events, rather than building fragile in-house pipelines.

Who's behind it? The company was founded by chief executive officer Gabriele Lini, chief science officer Francesco Di Corato, and chief technology officer Federico Cabassi. The three are engineers with backgrounds at Ambarella and Magneti MarelliDealroom has a profile for this one. Try Dealroom →, plus autonomous-driving research at SISSADealroom has a profile for this one. Try Dealroom →, the University of PisaDealroom has a profile for this one. Try Dealroom →, and the University of ParmaDealroom has a profile for this one. Try Dealroom →.

Why now? The raise follows a $1m pre-seed round in January 2026, led by Connect VenturesDealroom has a profile for this one. Try Dealroom → alongside VentoDealroom has a profile for this one. Try Dealroom → and angel investors. It also lands amid a growing push to solve the data bottleneck in physical AI, where robotics firms increasingly see data collection and handling — not just models or hardware — as the main constraint on moving AI from lab demos to reliable industrial use.

The signal: Investors are betting that the infrastructure layer of robotics is where value accrues. Mosaico's platform approach contrasts with rivals building physical data factories, but both reflect the same wager: better training data, not bigger robots, is what gets machines working in the real world.

Read more: Sifted

Image credit: "Google Robocar Racetrack Ride" by jurvetson, licensed under CC BY 2.0

More top stories