Meta says it has acquired Assured Robot Intelligence (ARI), an early-stage startup focused on foundation models for humanoid robots, for an undisclosed amount. In a statement reported by TechCrunch, Meta said ARI's team, including co-founders Xiaolong Wang and Lerrel Pinto, will join the company's Superintelligence Labs research organization.

What ARI was building

ARI's public site describes the company simply as "Frontier AI for Robots." On their personal pages, Wang and Pinto say ARI's mission is to build frontier or super-intelligent robots that can make physical labor more abundant. Those pages also help verify the team's background: Wang says he previously built foundation models at NVIDIA and is an associate professor at UC San Diego, while Pinto describes his work around large-scale robot learning, behavior modeling, and affordable open-source robots.

Why it matters

The deal is notable less for size - Meta did not disclose terms - than for talent and direction. ARI appears to have been a small, research-heavy company, so the acquisition looks more like an acqui-hire aimed at strengthening Meta's in-house embodied AI effort than a mature product purchase.

That fits a broader pattern across big AI labs: frontier model work is increasingly spilling into robotics, where companies hope reasoning systems trained on text, images, and video can eventually control machines in the physical world. Meta is now signaling that it wants more of that work under its own roof.