The expensive part of putting a robot to work is often not the robot. It is teaching the machine where it is, and a startup that skips that step is drawing major backing.
FieldAI builds models that let robots navigate without a map. Business Insider reports the company is seeking about $700M, in a round that could value it near $10B, some five times its worth last August. A term sheet is said to be signed.
Existing backers include Bezos Expeditions, Nvidia’s NVentures and Intel Capital. Names of new investors were not reported.
The technical promise targets a familiar headache. A hand-built map takes time to prepare and goes stale as a site changes, which throws off accuracy. FieldAI says its models need no map, GPS, internet link or preset route, and can react on their own when conditions shift, slowing a robot in a factory section where lights are failing.
The software spans platforms, from autonomous vehicles to humanoids. A March tie-up added support for Boston Dynamics’ Spot, aimed at inspecting equipment.
Sensor feeds also become digital twins that stay current, which customers can use to rehearse site changes or train models. FieldAI says revenue and contracts passed $135M recently, up from about $100M in June, across more than 30 customers.
