The hardest machine to automate on a jobsite isn’t the dozer. It’s the excavator.
Bedrock Robotics is betting it can crack that this year. The San Francisco startup, founded in 2024 by former Waymo engineers, says it will put fully operator-less excavators to work on customer sites in 2026. A $270 million Series B that closed in February, co-led by CapitalG and the Valor Atreides AI Fund, pushed total funding past $350 million and set the company’s valuation at $1.75 billion. That’s a lot of capital riding on articulated steel.
Why operator-less excavators are the real test
Plenty of vendors have automated haul trucks and rollers. Those machines follow predictable paths. An excavator does not. It swings, reaches, judges soil, and adjusts mid-cut, which is exactly why operator pay scales with experience. Getting a machine to read a trench and dig it unsupervised is a genuinely harder control problem than keeping a truck in its lane.
Bedrock’s approach is a retrofit. The Bedrock Operator mounts onto existing equipment using lidar, GPS and motion sensors, installs in a few hours, and needs no permanent modification. That matters for contractors who already own seven-figure fleets and aren’t going to scrap them for purpose-built robots. It’s the same logic driving rivals such as August Robotics and the bricklaying push from Buildroid: bolt intelligence onto tools crews already trust.
Supervised today, unmanned tomorrow
Here’s the part the headlines skip. Bedrock’s largest run so far, a 130-acre mass-excavation job in November 2025, was supervised autonomy. A human stayed in the loop. Mass grading on an open pad is close to the friendliest scenario an autonomous excavator will ever see: few obstacles, repetitive cuts, generous tolerances. Trenching next to live utilities is a different animal, and that’s where most of the dirt work actually happens.
So the 2026 milestone is real, but narrow. “Operator-less” on a controlled site is a believable near-term claim. Fully unmanned across the messy middle of a typical project is a longer road, and Bedrock’s founders, who watched self-driving cars take a decade to reach limited commercial service, know it.
The labor math behind the bet
Autonomy keeps drawing capital for one stubborn reason. Crews are short. Contractors keep telling surveyors that worker shortages are now the top cause of project delays, and equipment operators are among the hardest seats to fill. A retrofit kit that lets one experienced operator oversee three machines doesn’t replace the workforce. It stretches it.
The technology will get judged on a simple question: can a Bedrock-equipped excavator hit grade, on spec, without a person in the cab, on a real project with a real schedule? If the answer in late 2026 is yes even on the easy jobs, the case for autonomous earthmoving stops being a pitch deck and starts being a line item.
Sources: The Robot Report, Construction Dive.