Robotics data startup Foxglove raises $40M led by Bessemer to build the data stack powering the Physical AI revolution

Foxglove, a San Francisco-based startup building a data and observability platform for robotics companies, has raised $40 million in Series B funding led by Bessemer Venture Partners, with participation from existing investors Eclipse and Amplify Partners. The round brings the company’s total funding to more than $58 million since its founding in 2021.

For a startup barely four years old, Foxglove has quickly become a go-to name in robotics infrastructure. Its founders, Adrian Macneil and Roman Shtylman, previously worked at Cruise, where they built similar internal tools for autonomous vehicle development. They realized that most robotics teams lacked reliable off-the-shelf options for collecting, visualizing, and analyzing the massive amounts of sensor data generated by their robots. Foxglove was born to fill that gap — and now counts major players like NVIDIA, Amazon, Anduril, and Wayve among its customers.

Macneil describes the company’s mission as building the “AWS or Datadog for robotics.” In the early 2000s, he said, tech giants like Google and Amazon had to build their own infrastructure before startups emerged to make those tools accessible to everyone. Foxglove aims to do the same for the growing field of Physical AI — the integration of artificial intelligence into machines that interact with the physical world.

Macneil said the new funding will help the 50-person team expand and accelerate product development. “Every Physical AI company faces the same challenge: building a flywheel that lets robots capture and learn from vast quantities of data in complex, real-world environments,” he said. “Our mission is to build that infrastructure so our customers can focus on solving unique, domain-specific problems. This funding allows us to expand our platform to support the complete data lifecycle across development, testing, and operations.”

The company’s platform helps developers collect and analyze multimodal data — everything from 3D visuals and video to audio, GPS, and time-series data — in one unified workspace. This allows teams to better debug, train, and deploy robots in complex environments. Foxglove’s open-source logging format, MCAP, launched in 2022, has already become a standard in the robotics community and is included in frameworks such as ROS 2 and NVIDIA Isaac.

Foxglove’s technology has become deeply embedded across the Physical AI ecosystem. One of its customers, Shield AI, initially used Foxglove’s tools internally but later integrated them into its HiveMind autonomy stack, making Foxglove part of its SDK for other developers. Dexterity, another robotics company focused on logistics, said Foxglove has saved it more than 20% in development time and about $150,000 annually.

At Bessemer Venture Partners, partner Jeremy Levine sees Physical AI as the next major platform shift, on par with mobile and cloud computing.

“At Bessemer, we believe Physical AI represents the next generational platform shift — as impactful as mobile computing or cloud infrastructure,” said Jeremy Levine, partner at Bessemer Venture Partners. “Foxglove is the clear category leader, building the developer tools and infrastructure stack that every robotics company will rely on. We’re proud to partner with Adrian and his team as they accelerate this industry-defining opportunity.”

With fresh capital and growing adoption, Foxglove is positioning itself as the backbone of the Physical AI movement — a company building the unseen data infrastructure that may soon power the robots shaping the future of work, industry, and everyday life.