General Intuition Closes $320M to Build Action AI

The big picture: A New York AI lab that trains its models on billions of gameplay clips from real human players has raised $320M at a $2.3B valuation, and is already in conversations for another round. General Intuition closed its Series A in January 2026, just three months after a $133.7M seed round, with Khosla Ventures leading both financings.

Why it matters:

• Next AI Frontier: Most leading AI labs have focused on models that generate text, images, or code; General Intuition is tackling AI that can reliably act, perceive environments, and make real-time decisions in the physical world.
• Market Shift: Global robotics startups raised over $23 billion in 2026 alone, nearly matching all of 2025, signaling a significant investor bet on AI models that can act rather than just reason.
• Unique Data Advantage: The company leverages billions of action-labelled gameplay clips from Medal’s 17 million monthly active users, providing a dataset with precise human input metadata that most competitors lack.

How it works:

• Large Action Models: General Intuition trains what it calls large action models on gameplay data, learning from moment-to-moment player decisions like mouse movements and button presses, which are theorized to generalize to robotics.
• Dual Model Approach: The startup builds world models to predict environment evolution based on actions, and action models to generate optimal actions given observations, working in parallel.
• Real-World Application: An AI agent trained on a Fortnite-like game for 100 hours simultaneously powered a quadrupedal robot navigating a room using a single camera and minimal real-world data.

The catch: General Intuition operates in a highly competitive and well-funded field, with rivals like Physical Intelligence, Skild AI, Google DeepMind, Odyssey, and Rhoda also developing foundational models for physical AI. While its gaming data offers a unique advantage in capturing human intention, the challenge remains in generalizing these learned behaviors from simulated game environments to the full complexity and unpredictability of the real world. The rapid pace of fundraising and significant compute deal with CoreWeave indicate a substantial capital burn rate, betting heavily on research trajectory before a fully commercialized product.