Meta has acquired Assured Robot Intelligence (ARI), a startup building AI systems that enable humanoid machines to predict and respond to human behavior, the latest move in the accelerating corporate race to own the physical AI layer of the next computing platform. Financial terms were not disclosed.
ARI was founded by Xiaolong Wang, Zuxin Cheng, and Lerrell Pinto, three researchers with deep roots in academic and industrial robotics. Wang previously worked on chip design at Nvidia before joining the University of California, San Diego as a faculty member.
Pinto holds a professorship at New York University and co-founded Fauna Robotics, a startup making child-sized humanoid robots, which Amazon acquired last month before Pinto departed to start ARI.
The startup had raised seed funding from AIX Ventures, though the amount was not made public. Its entire team is now moving to Meta’s Superintelligence Labs, the division that oversees the company’s most advanced AI research.
In a post on
A spokesperson for Meta told Bloomberg that ARI brings expertise in designing models for “whole-body humanoid control” and self-supervised learning capabilities that Meta has identified as central to its robotics ambitions.
Internal documents from 2025 revealed that the company was already developing a consumer robot combining custom hardware with AI software, and CTO Andrew Bosworth has publicly described the software layer as a “bottleneck” across the board.
Bosworth’s stated approach mirrors Meta’s strategy in mobile: Build licensable software based on what Google did with Android, starting with efficient hand controls and expanding outward.
Amazon absorbed Pinto’s former company Fauna Robotics within the last month to boost its own robotic workforce program.
Meanwhile, Tesla halted production of its Model S and Model
