Construction is getting its biggest autonomy bet yet: SoftBank led a $200M Series A for Gravis Robotics, a Zurich startup that upgrades excavators and wheel loaders contractors already own, in what it calls the largest round of its kind in the sector.
Gravis, spun out of ETH Zurich in late 2022, deliberately skips the new-machine path. Its RACK kit adds cameras, 3D LiDAR, GNSS RTK positioning, and edge compute to existing fleets, and it already supports Caterpillar, John Deere, JCB, Hitachi, Volvo, Yanmar, Case, Develon, and Sumitomo models.
The technical challenge is unusual: every bucket swing reshapes the ground the machine must interpret next. Gravis leans on simulation-heavy training and sensor fusion, blending hydraulics data with lidar and camera feeds to react to soil and slope changes instead of following a fixed script. It claims throughput gains of up to 30% on terrain-aware excavation.
A tablet interface gives operators three levels of control, from in-cab guidance to full remote orchestration of several machines, which the company sees as the realistic adoption path.
Gravis is past the demo stage: it has worked with Taylor Woodrow at Manchester Airport, Holcim, and HD Hyundai, and a UK program will trial six fully autonomous excavators.
The round underlines how physical AI capital is rotating toward industries where the equipment and the tasks already exist.
