General Intuition is securing $300 million for AI developed using gaming data.

      **Summary:** General Intuition, an AI lab that emerged from the gaming clip platform Medal after declining a reported $500 million acquisition offer from OpenAI, is seeking to raise around $300 million at a valuation exceeding $2 billion. Backers are said to include Jeff Bezos and Eric Schmidt.

      Located in New York, General Intuition focuses on training agents to understand space and time through billions of video game clips and is in discussions to secure about $300 million, as reported by TechCrunch. This funding round follows an impressive $134 million seed round launched eight months ago.

      Notable investors reportedly include Jeff Bezos, whose physical AI venture Prometheus recently obtained $12 billion, and Eric Schmidt, alongside existing supporters such as Khosla Ventures and General Catalyst.

      **The Medal Backstory:** General Intuition was created after a deal fell through. In late 2024, OpenAI allegedly proposed $500 million to acquire Medal, a platform where gamers share short gameplay clips, to gain data for model training, according to The Information. However, Medal's founder, Pim de Witte, declined the offer and instead launched General Intuition in October 2025 with co-founders Eloi Alonso, Adam Jelley, and Vincent Micheli—all of whom have made significant contributions to world modeling and diffusion-based simulation.

      The primary asset of the company is Medal's dataset, which comprises approximately two billion video clips each year from over 10 million active users across various games. Unlike footage from YouTube or Twitch, which showcases gameplay from a viewer's perspective, Medal’s clips offer a first-person, interactive view, reflecting the spatial reasoning, timing, and decision-making that gaming entails.

      **Agents Over Models:** General Intuition distinguishes itself from competitors by focusing on training agents rather than merely building world models as standalone products. Companies like Decart, which recently secured $300 million, and Google’s Project Genie, which has integrated its world model with Street View’s 280 billion images, sell simulation environments to developers and enterprises. In contrast, General Intuition aims to make the agents the primary product, using the world models solely for training, thereby directly linking revenue to the agents’ capabilities rather than the quality of virtual scene rendering.

      **A Competitive and Funded Landscape:** The race for world models garnered substantial investment in 2026. Runway, valued at $5.3 billion, introduced its first world model in December and has been generating $40 million in annual recurring revenue each quarter. World Labs, founded by Stanford's Fei-Fei Li, raised $1 billion at a $5.4 billion valuation in February. Odyssey AI secured $310 million with support from AWS and AMD, while Yann LeCun’s AMI Labs raised over $1 billion at a valuation of $3.5 billion.

      The prevailing belief is that AI must comprehend physics alongside language, and video data serves as the vital link. The founders of General Intuition believe their dataset offers a competitive advantage over many rivals. Alonso and Micheli have developed DIAMOND, a diffusion-based world model that predicts future frames directly instead of compressing them into tokens. The Medal dataset provides dense, interactive footage that is unmatched by static videos.

      **Future Plans:** The company reportedly intends to utilize the new funds for expanding computing capabilities and aims to launch a product by late summer or early autumn. A broader concern is whether the agent market that OpenAI and others are pursuing will develop swiftly enough to validate the high valuations assigned to companies creating the foundational reasoning. OpenAI's strong interest in Medal's data, evidenced by its $500 million offer, and General Intuition's current valuation being four times that amount, highlights how the market values control of training data in a time when every major AI lab is in search of it.