SoftBank leads $200M into Gravis Robotics to put autonomous AI on excavators

SoftBank is leading a $200 million Series A into Gravis Robotics, a Zurich-based construction-robotics startup that retrofits existing excavators with an autonomy kit rather than selling new machines. Gravis describes the round as the largest in construction-robotics history.

Gravis was spun out of ETH Zurich in 2022 and sells the Gravis Rack, a control-and-software package that bolts onto installed iron. The retrofit approach lets contractors add autonomy to fleets they already own rather than replace machines.

Earthmoving inverts the self-driving problem. Most physical-AI work targets static worlds, with self-driving cars on fixed pavement as the example. The machine is reshaping the world it has to navigate, so a model trained on yesterday's jobsite is working a different site tomorrow. CEO Ryan Luke Johns frames earthmoving as the foundational bottleneck for housing, energy grids, and the AI data-center buildout.

CTO Dominic Jud says the system grounds operator-style physical feedback (engine strain, vibration, hydraulic resistance) in machine telemetry at microsecond speeds, and trains across billions of cubic yards of synthetic material, from soft clay to rock-filled soil, to bridge the sim-to-real gap.

Gravis did not disclose valuation. The round's size and SoftBank's name put the company at the front of a category that has been slow to automate; whether the retrofit model scales across fragmented contractor fleets is the open question.