From Fortnite to robots: General Intuition raises $2.3B on bet that video games can train AI agents for the real world

June 25, 2026 - 18:13

General Intuition has secured $2.3 billion in new funding to pursue an unusual bet: that the fast-paced action of video games can teach artificial intelligence how to operate in the real world. The company plans to use millions of hours of gameplay footage to train AI agents, aiming to develop what it calls "machine intuition" rather than simple pattern recognition.

The core idea is that games like Fortnite require split-second decisions, spatial awareness, and adaptive strategy. By feeding this data into neural networks, General Intuition hopes its AI will learn to handle uncertainty and react to changing environments the way a human player does. The company argues that traditional AI training methods, which rely on static datasets or controlled simulations, fail to capture the messy, unpredictable nature of real-world tasks.

The new funding round values General Intuition at over $10 billion. Investors are betting that this approach could eventually power robots, autonomous vehicles, and industrial systems. The company has already begun testing its models on simple robotic arms, using the gameplay-trained AI to guide physical movements.

Critics point out that translating virtual skills into physical actions remains a major challenge. A game character's ability to dodge a bullet does not automatically translate to a robot avoiding a falling object. But General Intuition believes its massive dataset and unique training method will bridge that gap. The company plans to double its engineering team and build a dedicated simulation facility to test real-world applications over the next two years.