Google DeepMind Partners with Agile Robots to Put Gemini Robotics Inside 20,000+ Industrial Machines
Google DeepMind and Agile Robots have announced a research partnership to integrate Gemini Robotics foundation models directly into Agile Robots' fleet of over 20,000 deployed industrial systems. The collaboration is designed to create a continuous feedback loop: real-world data from Agile Robots' physical hardware will flow back into DeepMind's models, refining them against the messy realities of factory floors and industrial environments rather than controlled lab conditions.
This matters because most AI robotics partnerships are exactly that — partnerships announced at conferences with demos that rarely leave the research lab. Agile Robots is different. With more than 20,000 units already operating in production environments, they bring one of the largest physical deployment pipelines for embodied AI that Google DeepMind could tap outside of its own hardware projects. Every real-world interaction becomes training data, and that kind of scale is genuinely hard to replicate in simulation.
For Google, the deal is part of a wider push to establish Gemini Robotics as the foundation model of choice for industrial automation — a race that also includes competitors like Physical Intelligence and Boston Dynamics' expanding software ambitions. Getting into Agile Robots' installed base early gives DeepMind a durable foothold in the industrial segment before the market consolidates around a few dominant AI-for-robots platforms.