Bedrock Robotics
Autonomous construction equipment company retrofitting heavy machinery with an AI autonomy kit for self-driving earthmoving and excavation. The reversible Bedrock Operator kit mounts to a customer's existing fleet to run mass excavation and grading.
Best for: General contractors and site-prep firms wanting to run mass excavation and grading autonomously on their existing heavy equipment fleet.
Last reviewed 6/2/26
- Property types
- Construction sites, Industrial/manufacturing development, Site preparation, Aggregates/quarries
- Geography served
- United States (Arizona, Texas, Arkansas deployments)
How Bedrock Robotics uses AI[1][2][3][4]
Retrofits existing excavators and dozers into supervised-autonomous machines using the Bedrock Operator kit, which combines lidar, GPS, cameras, and onboard AI to run mass excavation and grading to project specs.
- • Perception and localization from fused lidar, GPS, IMUs, and eight HD cameras processed by an in-cab computer
- • Site mapping, motion planning, and safety control onboard each machine
- • Bolt-on kit mounts to existing equipment same-day with reversible, non-permanent installs
- • Runs supervised-autonomy mass excavation across excavators ranging from 20 to 80 tons
- • Founders previously led Waymo's autonomous driving programs
- • Raised $270M Series B in February 2026, co-led by CapitalG and Valor Atreides AI Fund, at a $1.75B valuation
AI type: Computer Vision, Perception, Autonomous Control, Mapping/Localization (LiDAR + cameras)
API: unknown MCP: unknown
Key numbers[2][4][5][6]
- • $270M Series B at a $1.75B valuation (Feb 2026), bringing total funding to over $350M
- • ran the industry's largest-known supervised-autonomy mass excavation on a 130-acre manufacturing site in Phoenix with Sundt Construction (Nov 2025), moving more than 65,000 cubic yards on a project targeting ~700,000 cubic yards
- • deploying across live sites in Arizona, Texas, and Arkansas
- • targeting first fully operator-less deployments in 2026
Credibility[2][4][6]
- Founders
- Founded 2024 by Boris Sofman (CEO, ex-Waymo/Anki), Kevin Peterson (CTO, ex-Waymo), Ajay Gummalla and Tom Eliaz (VP Engineering); a majority of the founding team worked together at Waymo on road autonomy. Senior hires include Vincent Gonguet (Head of Evaluation, ex-Meta Llama AI safety) and John Chu (Head of People, ex-Waymo).
- Customers
- Named partners and customers include Sundt Construction, Zachry Construction, Champion Site Prep, and Capitol Aggregates.
- Investors
- Series B co-led by CapitalG and Valor Atreides AI Fund, with Xora, 8VC, Eclipse, Emergence Capital, Perry Creek Capital, NVentures (NVIDIA), Tishman Speyer, MIT, Georgian, Incharge Capital, and C4 Ventures. Earlier $80M Seed/Series A led by Eclipse (Seed) and 8VC (Series A).
Founded
2024
Headquarters
San Francisco, CA
Stage
Series B
Sources
- Bedrock Robotics product site
- Series B funding (PR Newswire)
- Series B coverage (The Robot Report)
- Supervised autonomy / 130-acre Phoenix deployment, equipment specs, customers (Equipment World)
- 65,000 cubic yards moved on Southwest project (Engineering News-Record)
- Emerged from stealth with $80M Seed/Series A, AZ/TX/AR deployments, founders (PR Newswire)
Related companies
Toggle
Toggle uses computer vision and 3D sensing to localize rebar, verify fabricated parts, and guide robots that assemble reinforcement cages for construction.
Toggle uses computer vision and 3D sensing to localize rebar, verify fabricated parts, and guide robots that assemble reinforcement cages for construction.
AIM Intelligent Machines
AIM Intelligent Machines converts heavy earthmoving equipment into autonomous machines for construction and mining. Its plug-and-play retrofit kit and AI autonomy software let dozers, excavators, and loaders grade sites, dig, and move earth without an operator, supervised remotely.
AIM retrofits existing bulldozers, excavators, loaders, and skid steers with plug-and-play sensor and edge-compute hardware, then runs them with an embodied AI autonomy stack. The system uses 360-degree LiDAR perception to build real-time 3D maps of a site and executes dig, haul, plow, fill, and grade tasks without an operator in the cab.
T2D2
AI building inspection platform using computer vision to detect 80+ damage types from drone, photo, and thermal imagery, replacing manual visual surveys.
Computer vision and deep learning, trained on hundreds of thousands of forensic inspection images from Thornton Tomasetti, detect and classify visible damage on facade and structural materials.