Meta Buys Deeper Into the Humanoid Intelligence Stack
Assured Robot Intelligence brings Meta a compact team working across robot learning, imitation, manipulation and whole-body control. That matters more than the company’s size.
Meta buys deeper into humanoid intelligence with Assured Robot Intelligence acquisition
Meta has acquired Assured Robot Intelligence, a young robotics AI company focused on humanoid intelligence and robot control. Financial terms were not disclosed. The ARI team, including co-founders Lerrel Pinto and Xiaolong Wang, is joining Meta Superintelligence Labs and will work closely with Meta Robotics Studio.
The acquisition is not really a hardware story.
ARI was not known as a public humanoid OEM with a commercial robot fleet. Its relevance sits closer to the intelligence layer behind humanoids. Reporting describes the company as developing robotic intelligence that can help robots understand, predict and adapt to human behaviour in complex, dynamic environments.
That makes the team an important part of this story.
Pinto’s background is especially notable. Before ARI, he co-founded Fauna Robotics, then left in 2025. Amazon acquired Fauna in March 2026, adding another layer of big-tech competition around humanoid robotics talent.
His public research sits deep in robot learning. Across recent work, Pinto has been tied to imitation learning, teleoperation, mobile manipulation, tactile dexterity, robot memory, behaviour generation and deployment adaptation. That is the kind of work needed when humanoids move beyond choreographed demos and into changing, contact-heavy environments.
Wang brings another part of the same stack. He is a UC San Diego professor and co-founder of ARI. His own site describes ARI’s mission as building “super-intelligent humanoid robots at scale,” while Nvidia’s research profile lists his work across computer vision, robotics and robot learning, including legged robot vision-language-action navigation and object-centric manipulation.
His recent humanoid work also points directly at whole-body control. AMO, a 2025 Robotics Science and Systems project with Wang as an author, focuses on adaptive motion optimisation for hyper-dexterous humanoid whole-body control.
That fits the broader direction of Meta’s robotics effort.
Meta has been building around embodied AI for years, through simulation, egocentric perception, tactile sensing and robot-learning research. Its newer humanoid push appears to be pulling that research base toward physical systems. The ARI acquisition adds a compact team working much closer to the problems that still limit humanoids in the real world.