
The recent partnership between Boston Dynamics and Google DeepMind, announced at CES 2026, marks a significant milestone in the evolution of Physical Intelligence. This collaboration represents a high-stakes move in the ongoing competition for the “brain” of the next generation of robots. By integrating DeepMind’s Gemini Robotics with the Atlas humanoid platform, the two companies aim to enable humanoid robots capable of performing a wide variety of industrial tasks.
Google DeepMind’s contribution aligns with Part 3 of ARC’s Physical Intelligence taxonomy: the Software & Intelligence Stack. The Gemini Robotics foundation models function as the robot’s “embodied mind,” providing high-level reasoning, multimodal perception, and the ability to interact with humans and use tools. This approach is intended to move robots beyond rigid, preprogrammed automation toward more responsive collaboration, allowing them to adapt to dynamic industrial environments in real time.
Despite the intense hype surrounding humanoid robots, the partnership’s focus on the automotive industry reflects a pragmatic starting point. The automotive sector has historically led the adoption of robotics, and humanoids could help address labor shortages in tasks such as kitting and machine tending, particularly in brownfield facilities that are difficult to automate using traditional fixed robots.
Overall, this alliance represents a clear attempt to bridge the gap between digital intelligence and physical action. By combining a mature physical platform with an advanced AI “brain,” the partners are positioning Physical Intelligence to move beyond pilot projects toward real-world deployment. The critical test will be whether these joint research efforts can scale safely and efficiently while delivering measurable impact across industrial operations.