Meta acquires robotics startup Assured Robot Intelligence to advance humanoid robot control models
Meta has acquired humanoid robotics startup ARI, bringing together experts in robot learning and control. The acquisition signals Meta's continued push into physical AI and embodied intelligence research.
1 source · cross-referenced
- Meta acquired Assured Robot Intelligence, a startup focused on foundation models for humanoid robots, for an undisclosed amount.
- Co-founders Lerrel Pinto (formerly NYU, Fauna Robotics) and Xiaolong Wang (formerly Nvidia, UC San Diego) will join Meta's Superintelligence Labs division.
- The team will focus on robot control, self-learning, and whole-body humanoid capabilities as part of Meta's broader AI research strategy.
- Industry forecasts for humanoid robotics value range from $38 billion (Goldman Sachs, 2035) to $5 trillion (Morgan Stanley, 2050), reflecting uncertainty in the emerging field.
Meta has acquired Assured Robot Intelligence, a robotics startup founded by researchers Lerrel Pinto and Xiaolong Wang, according to a statement from Meta to TechCrunch. The acquisition price was not disclosed. The startup specialized in foundation models designed to enable humanoid robots to understand and adapt to human behavior in dynamic environments, with applications to household and labor tasks.
Pinto comes from an academic background at NYU and previously co-founded Fauna Robotics, which Amazon acquired last month. Wang is a former Nvidia researcher and associate professor at UC San Diego who has received multiple prestigious research awards. Both founders will now lead robotic intelligence efforts within Meta's Superintelligence Labs research division.
Meta has been pursuing humanoid robotics research for at least a year, according to internal documents that surfaced in 2025. The company's strategy encompasses both AI models and hardware intended eventually for consumer release. Experts in the field increasingly argue that advancing AI—particularly toward artificial general intelligence—may require training models through direct robot interaction with the physical world rather than relying solely on pre-existing datasets.
The robotics sector remains highly speculative, with wildly divergent market forecasts. Goldman Sachs projects the humanoid robotics market will reach $38 billion by 2035, while Morgan Stanley estimates $5 trillion by 2050. This divergence reflects both the sector's acknowledged potential and persistent uncertainty about technical feasibility and adoption timelines.
- Apr 30, 2026 · TechCrunch — AI
Amazon's AWS cloud business posts fastest growth in 15 quarters, fueled by AI infrastructure demand
Trust71 - Apr 29, 2026 · Hugging Face
DeepInfra Now Available as Hugging Face Inference Provider with SDK Support
Trust61 - Apr 29, 2026 · Hugging Face
Granite 4.1 LLMs use five-stage pre-training and multi-stage reinforcement learning to achieve dense model efficiency
Trust79