Meta has acquired Assured Robot Intelligence (ARI), a robotics AI startup building foundation models for humanoid robots, the company confirmed on May 1, 2026. The deal hands Mark Zuckerberg's Superintelligence Labs a fresh team of physical-AI researchers as the social-media giant tries to catch up in a race that already includes Tesla, Figure, 1X, and a wave of well-funded Chinese rivals.
The price was not disclosed. ARI's full team, including its two co-founders, will join Meta's AI unit.
Who ARI is
ARI was co-founded by Xiaolong Wang, a former Nvidia researcher and associate professor at UC San Diego, and Lerrel Pinto, who previously taught at NYU and co-founded the kid-size humanoid startup Fauna Robotics — a company Amazon acquired in March 2026, according to TechCrunch's reporting.
The startup was working on foundation models intended to let humanoid robots understand, predict, and adapt to human behavior in complex environments, with household chores and other physical-labor tasks as early target use cases. ARI had raised an undisclosed seed round from AIX Ventures before the Meta deal.
What Meta said
In a statement provided to reporters, Meta framed the acquisition as a step toward "how we can design our models and frontier capabilities for robot control and self-learning to whole-body humanoid control." The team will sit inside Superintelligence Labs, the research group Meta reorganized last year under Alexandr Wang.
The move fits a years-long pattern. A 2025 internal memo previously revealed Meta's intent to build consumer-oriented humanoid robots that combine its own AI models with custom hardware, even though the company has not yet shipped a robot of its own.
Why it matters
The acquisition lands at a moment when humanoid robotics is shifting from research demos to industrial pilots. BMW, Mercedes, Toyota, and Amazon warehouses have all begun deploying humanoid units in production environments over the past year, and investors continue to pour money into the category despite uncertain unit economics.
Market forecasts are wildly divergent: Goldman Sachs projects the humanoid robot market will reach roughly $38 billion by 2035, while Morgan Stanley has floated a $5 trillion ceiling by 2050. Either way, the prize is large enough that no major AI lab can afford to ignore the physical-AI stack.
For Meta, ARI also brings something less tangible — a credibility signal that its Superintelligence Labs reorganisation is producing real moves rather than headlines. With OpenAI, Google DeepMind, and Nvidia all expanding their robotics work, Zuckerberg is betting that controlling the foundation models for humanoid bodies will matter as much as controlling the chatbot layer.
What to watch
The open questions now are how quickly Meta integrates ARI's research into a shippable product, whether Pinto and Wang stay long enough to lead it, and whether Meta licenses its humanoid stack to hardware partners or builds its own robots end-to-end. The company has not committed to a public timeline.
— Michael Ouroumis
