Meta Platforms has acquired Assured Robot Intelligence, the frontier robotics-AI startup building foundation models for whole-body humanoid control, the company confirmed on 1 May 2026. Financial terms were not disclosed. The ARI team, including co-founders Xiaolong Wang and Lerrel Pinto, joins Meta Superintelligence Labs and will work directly with Meta Robotics Studio, the humanoid-hardware group Meta stood up in 2025.
What Meta Just Bought
ARI was founded in 2021 in San Diego. Xiaolong Wang is a former Nvidia researcher and UC San Diego professor; Lerrel Pinto is an NYU faculty member and previously co-founded kid-size humanoid startup Fauna Robotics — which Amazon acquired just weeks before the ARI deal. ARI's technical focus was foundation models that let humanoids understand, predict and adapt to human behaviour in dynamic settings, plus its e-Flesh tactile-sensing platform. The startup had raised an undisclosed seed round backed by AIX Ventures.
The Superintelligence Labs Fit
Meta framed the deal as an accelerant, not a pivot. "This team, led by Lerrel Pinto and Xiaolong Wang, will bring deep expertise in how we can design our models and frontier capabilities for robot control and self-learning to whole-body humanoid control," a Meta spokesperson told TechCrunch. Meta's humanoid group inside Reality Labs has been staffing up on hardware for months — sensors, actuators, batteries and software stacks — and reportedly wants to sell the stack to third-party OEMs, an Android-for-humanoids play. ARI's model layer plugs directly into that.
A Sprint Across Big Tech Robotics
Meta is the latest hyperscaler to buy its way into physical AI. Amazon snapped up Fauna Robotics in March; Google DeepMind is deep in a partnership with Boston Dynamics; Apple is reportedly staffing an in-house humanoid team; and Chinese peers such as Alibaba and Xiaomi are moving in parallel. In parallel with M&A, capital keeps flowing to the pure-play side: Neura Robotics closed $1.4 billion Series C, and unicorns like LimX Dynamics continue to raise pre-IPO rounds.
Physical Data Is The New Moat
Meta's chief AI officer has previously argued that the road to AGI runs through embodied learning — models that train against reality, not just text. ARI's foundation-model work, combined with Meta Robotics Studio's hardware, sensor and simulation stacks, gives Meta a shot at collecting proprietary interaction data at scale. Whether Meta ships a consumer robot or licenses the stack, the acquisition tightens Meta's grip on the physical layer of AI at exactly the moment the humanoid market is professionalising.
Reporting based on coverage from TechCrunch, Bloomberg and Meta.