Who's building it
A handful of labs openly say they are aiming at AGI or superintelligence. A separate group is building the bodies.
Frontier AI labs
| Organization | Based | Stated direction |
|---|---|---|
| OpenAI | San Francisco | Mission framed around ensuring AGI benefits all of humanity; makers of ChatGPT and the GPT models. |
| Google DeepMind | London and Mountain View | Google's combined AI lab; Gemini models, plus a dedicated robotics program. |
| Anthropic | San Francisco | Safety-focused lab; makers of the Claude models. |
| Meta Superintelligence Labs | Menlo Park | Meta's AI group, reorganized in 2025 with superintelligence as its stated goal. |
| xAI | Bay Area and Memphis | Elon Musk's AI company; makers of the Grok models. |
| Safe Superintelligence Inc. | Palo Alto and Tel Aviv | Founded in 2024 by Ilya Sutskever with a single goal: a safe superintelligence. |
Robot brains
Companies building the AI models that control robots, often for many different machines.
| Organization | Focus |
|---|---|
| Google DeepMind | Gemini Robotics models for general-purpose robot control. |
| Nvidia | Isaac simulation tools and GR00T foundation models for humanoids. |
| Physical Intelligence | General-purpose robot foundation models (the π series). |
| Skild AI | A shared "robot brain" meant to work across many robot bodies. |
Robot bodies
Tesla, Figure AI, Boston Dynamics, Agility Robotics, Apptronik, 1X and Unitree lead the humanoid race. Their status changes monthly, so we track them separately at sirobot.us.
Last reviewed October 2026.