Newsletter Subscribe
Enter your email address below and subscribe to our newsletter
Enter your email address below and subscribe to our newsletter

General Intuition, the AI lab that emerged from gaming clip platform Medal after rejecting a reported $500 million acquisition offer from OpenAI, is raising approximately $300 million at a valuation of just over $2 billion, according to TechCrunch reporting cited by The Next Web.kucoin
The New York-based startup, founded by Pim de Witte, spun out of Medal roughly eight months ago with a $134 million seed round — one of the largest on record — led by Khosla Ventures and General Catalyst. The new round reportedly includes backing from Jeff Bezos, whose physical AI venture Prometheus recently raised $12 billion, and former Google CEO Eric Schmidt, alongside existing investors.letsdatascience
General Intuition trains world models and embodied AI agents using Medal’s dataset of approximately 2 billion gaming video clips per year from more than 10 million monthly active users across tens of thousands of games. Unlike platforms such as Twitch or YouTube, Medal’s data captures controller inputs alongside visual streams, allowing agents to learn spatial-temporal reasoning by observing how players navigate virtual environments.thenextweb
The company’s approach treats world models as training grounds for AI agents that can reason about space and time. Its models can reportedly understand environments they were not trained on and predict actions within them using only visual input. Initial applications target gaming, search-and-rescue drones, and autonomous navigation of unfamiliar physical environments.techcrunch
The fresh capital will be used to scale compute capacity, with the company planning to release a new product by late summer or early fall.kucoin
The raise comes amid intensifying competition in the physical AI space. OpenAI recently relaunched its robotics program, and startups like XDOF have emerged to build data pipelines for training robots. General Intuition’s pitch is that gaming data — rich in physics, spatial relationships, and agent interactions — offers a uniquely scalable path to training embodied intelligence without the costly real-world data collection that robotics typically requires.techcrunch