General Intuition, an AI startup training agents on millions of hours of video game footage, has raised $320 million in a funding round led by Khosla Ventures. The round values the company at $2.3 billion, according to TechCrunch. Total funding for General Intuition now stands at $454 million.
The bet is that gameplay action data can teach AI systems something akin to human intuition — specifically spatial reasoning and decision-making. Instead of relying on text or video alone, the company feeds raw gameplay data into its models, aiming to produce agents that can navigate physical spaces, operate machinery, or control robots.
The company's R&D floor in New York houses its 31-year-old co-founder and team. The approach is unusual even by AI startup standards: most competitors train on static datasets or real-world sensor feeds. General Intuition argues that games offer millions of hours of low-cost, high-variety interaction data that can be repurposed for real-world tasks.
If successful, General Intuition's technology could reshape industries from logistics to autonomous vehicles. But skeptics question whether virtual-world reflexes transfer reliably to physical environments. The company has not disclosed specific deployment partners or commercial timelines.
A rival AI researcher noted that gaming data often embeds unrealistic physics and simplified outcomes, which may limit real-world generalization.