Meta Acquires Assured Robot Intelligence for Humanoid AI

Meta Platforms acquired Assured Robot Intelligence (ARI), a startup developing AI models for humanoid robots, Bloomberg reports, citing a company spokesperson. Financial terms of the deal were not disclosed, Bloomberg and the Wall Street Journal note. A Meta spokesperson told TechCrunch in an emailed statement, "We acquired Assured Robot Intelligence, a company at the frontier of robotic intelligence designed to enable robots to understand, predict and adapt to human behaviors in complex and dynamic environments." TechCrunch and Engadget report that ARI's team, including co-founders Xiaolong Wang and Lerrel Pinto, will join Meta's Superintelligence Labs.

Per reporting by TechCrunch, ARI had been building foundation models for humanoid robots aimed at whole-body control to perform a variety of physical tasks, including household chores and other forms of high-value manual labor. TechCrunch documents the co-founders backgrounds: Xiaolong Wang formerly at Nvidia and UC San Diego, and Lerrel Pinto previously at NYU and a co-founder of Fauna Robotics. Engadget quotes ARI co-founder Xiaolong Wang saying the company pursued "training a truly general-purpose physical agent," and that the founders believe the agent will be humanoid. Engadget also quotes Meta CTO Andrew Bosworth saying, "Software is the bottleneck," and describing early work starting with software for a dexterous hand.

Companies and researchers working on robotic generalization increasingly treat foundation models and large-scale self-supervised learning as necessary components for flexible robot control. Industry-pattern observations show that integrating perception, prediction, and control into a single model family raises engineering demands around real-world data collection, sim-to-real transfer, safety validation, and compute cost. For practitioners, those demands typically translate into heavier investment in on-robot data pipelines, simulation fidelity, and hybrid learning loops that combine simulation, human demonstrations, and online fine-tuning.

Industry context: Public coverage frames this acquisition as another major tech company moving from purely virtual AI toward embodied systems. Reporting places Meta alongside other firms pursuing humanoid robotics, including Amazon and Tesla, and it highlights a broader narrative --- present in TechCrunch and eWeek reporting --- that training agents in the physical world is viewed by many researchers as a pathway to more general capabilities. For practitioners, the move underscores increased demand for cross-disciplinary expertise that spans perception, control, systems engineering, and safety engineering.

Observers should track several indicators over the coming months: 1) whether Meta publishes technical papers or open-sourced components that reveal ARI's modeling approach; 2) hiring patterns or job postings in areas such as on-robot data engineering, sim-to-real research, and dexterous manipulation; and 3) any partnerships or licensing announcements around robot-control software, a direction Engadget reports Meta is considering. Reporting to date has not disclosed the transaction value.