General Intuition is raising $300 million to train AI agents on the video game data OpenAI tried to buy

What happened

General Intuition, a New York-based AI lab focused on training agents to understand space and time through video game data, is in talks to raise about $300 million at a valuation slightly above $2 billion. This funding round comes just eight months after the company raised $134 million at launch. General Intuition builds AI models called world models that learn from billions of video game clips to teach agents how to reason and plan in complex environments. Notably, the video game data source they leverage was reportedly sought by OpenAI, highlighting its value for next-level AI training.

Why it matters

This sizable raise signals growing investor appetite for AI labs training agents with alternative data forms, especially those that offer reasoning capabilities beyond text or static images. Video game data offers rich environments where AI agents can learn causal relationships, strategy, and planning in a way real-world data often cannot. For startups and investors, General Intuition’s approach stresses that world models trained on interactive, spatial-temporal data remain a promising avenue to push AI beyond passive pattern recognition. The company’s valuation above $2 billion also pressures competitors and AI platforms chasing new training data sources and agent capabilities.

What to watch next

Watch if General Intuition’s world models accelerate advances in AI agents for practical automation scenarios, like robotics, simulation, or gaming AI. Also track how they leverage the large funding to scale their dataset, model size, and real-world integration. Their progress will test whether video game-derived data can compete with or complement mainstream AI training heavily reliant on web text and images. Finally, keep an eye on the startup and investors competing for scarce large-scale interactive datasets that can unlock new agent reasoning skills in AI.

AI Quick Briefs Editorial Desk