General Intuition raises $320M at $2.3B valuation to build agents trained on gameplay data
Tags AI · Infrastructure · Enterprise
General Intuition secured $320 million led by Khosla Ventures, with participation from General Catalyst, Jeff Bezos, Eric Schmidt, Nico Rosberg, Google DeepMind researchers, and MIT researchers. The startup, spun out of gaming clip platform Medal, uses hundreds of millions of hours of gameplay with embedded action labels to train a single model that generalizes from game environments to real-world robotics. The company's quadruped robot required only eight minutes of real-world data for fine-tuning. Total disclosed funding reaches $454 million.
Technical significance
The $2.3B valuation for a company with a novel data moat (proprietary action-labeled gameplay) signals investor willingness to bet on alternative training paradigms beyond raw compute scaling. If gameplay-to-robotics transfer works at scale, it could dramatically reduce the cost of training embodied AI, potentially disrupting the economics of companies relying on expensive real-world data collection.