The interesting signal is the convergence of robotics, physical-world data, and foundation models. Atoms is not pitching a single robot; it is trying to build a platform that spans food, mining, transportation, and other physical operations.
That makes the data layer especially valuable. Real-world robotics depends on large volumes of messy sensory and action data, and companies that can collect, label, replay, and evaluate that data are becoming core infrastructure.
Atoms has hired Vikas Chandra, a former Meta AI executive known for lightweight models used in smart glasses, as vice president of AI.
The company has raised about $1.7 billion and is trying to build foundation models for robots that operate in real environments rather than just chat interfaces.