Atoms raises $1.7B for physical AI
Travis Kalanick’s new robotics company is combining automation, logistics infrastructure and physical operations in one unusually capital-intensive platform bet.
Physical AI is attracting software-scale ambition with infrastructure-scale capital. Atoms has raised $1.7 billion to build a robotics platform that reaches beyond a single machine or model.
What happened
Atoms is led by former Uber chief executive Travis Kalanick and brings together assets from CloudKitchens and industrial-automation company Pronto. Andreessen Horowitz led the financing, with Bain Capital, Fifth Wall and Uber also participating. Ben Horowitz is joining the board.
The company has not yet disclosed a complete product roadmap, but its stated direction is broader than selling an individual robot. Atoms is working toward a common operating foundation for physical machines: software, autonomous systems, deployment infrastructure and the real-world environments in which robots perform work.
That makes the financing unusually ambitious. Instead of developing a robot and asking customers to integrate it, Atoms appears to be assembling more of the operational stack itself, drawing on CloudKitchens’ physical sites and Pronto’s industrial-autonomy expertise.
Why it matters
Robotics startups frequently stall between a successful demonstration and a repeatable commercial deployment. Hardware must work reliably, software must adapt to messy environments, and customers need installation, maintenance and operational support.
Atoms’ vertically integrated approach could shorten that gap by testing systems inside operating businesses rather than isolated laboratories. It could also generate proprietary operational data—one of the most valuable inputs for improving physical-AI systems.
The bigger picture
The round shows how the AI investment cycle is expanding from models and data centres into the physical economy. Investors are increasingly backing companies that control not only intelligence, but also the machines, facilities and workflows through which that intelligence produces economic output.
The risk is equally large: owning more of the stack means higher capital requirements and greater execution complexity. Atoms now has the funding to build at scale, but it still has to prove that one platform can operate across sufficiently valuable real-world use cases.