Autonomous excavators have been three years away for about a decade. The money that landed Monday is betting the problem was never the machine — it was that everyone kept trying to sell a new one.
Gravis Robotics announced a $200 million Series A with SoftBank Group as sole investor. The company calls it the largest Series A in construction robotics history, and nothing in the public record contradicts that. Gravis spun out of ETH Zurich in 2022 and runs offices in Zurich, Austin and Oxford.
The Retrofit Argument
The product is the Gravis Rack: a roof-mounted computing appliance carrying cameras and lidar, plus two rear masts with additional cameras, lidar and GNSS. It pairs with a tablet called Slate for remote operation. Supported machines span Caterpillar, Case, Develon, John Deere, JCB, Hitachi, Sumitomo, Yanmar and Volvo.
That list is the whole pitch. A fleet owner running mixed iron doesn’t have to standardize on one OEM or trade in serviceable machines to get autonomy. Gravis argues roughly two-thirds of global excavator demand sits outside the top three manufacturers, which is a company claim with no third-party citation attached, but the strategic logic holds either way.
Operating modes run a spectrum. Gravis Copilot keeps the operator in the cab with 3D guidance and hazard detection. At the other end, one operator supervises a fleet. Every equipped machine also functions as a site sensor, which automates survey and hazard mapping as a byproduct.
Why Dirt Is Harder Than Warehouses
CTO and co-founder Dominic Jud framed the technical problem in a way worth quoting: “Skilled operators read the earth through subtle physical feedback — listening to the engine strain, sensing the machine vibration and reacting to hydraulic resistance. Our AI takes that same physical input and grounds it in machine telemetry, responding to varying subterranean forces and soil mechanics at microsecond speeds.”
That’s the distinction between navigation autonomy and interaction autonomy. A self-driving car succeeds by not disturbing its environment. An excavator’s entire function is to destroy and reshape it, against soil conditions that change every bucket. It’s why earthmoving has trailed warehouse robotics by roughly a decade.
The Questions to Ask
The 30 percent productivity claim is measured against peak manual operation, and it comes from the company. Treat it as a hypothesis until somebody publishes third-party field data. On the heavy-civil jobs where this iron actually works, like the Susquehanna River bridge program, the constraint is skilled operators rather than machines. Gravis says it’s deployed across four continents and is leading an $8 million UK government-backed CAM Pathfinder project with Flannery Plant Hire.
The harder question is liability. When one operator supervises four machines and one of them strikes an unmarked utility, the contract language on responsibility hasn’t been written yet in most of the industry. Fleet owners evaluating a bolt-on kit should get that answer in writing before the first machine gets racked. The ROI math on a retrofit is genuinely different from buying autonomous iron, and so is the risk allocation.