AI Startup General Intuition Raises $320 Million in Series A Funding

AI Startup General Intuition Raises $320 Million in Series A Funding General Intuition is turning video game play into a training ground for real‑world robots—and investors are wagering billions that the bet will pay off.

Founded in 2025 as a spin‑out from Medal, a platform for sharing gaming clips, General Intuition began by repurposing “hundreds of millions of hours” of uploaded gameplay as training data for AI agents, teaching them spatial‑temporal reasoning, or how to move through space and time. Early internal demos showed the same AI “brain” controlling both a Fortnite‑style game agent and a quadrupedal robot, with just eight minutes of real‑world robotics data used to fine‑tune the model for physical navigation.

By October of its launch year, the lab had already raised $134 million, positioning itself as a “frontier” AI shop focused on building agentic models that can generalize from gameplay to simulations and physical embodiment. Its core thesis: that action‑rich gaming data—both video and player inputs—can serve as superior labels for training large action models.

On June 25, 2026, TechCrunch reported that General Intuition had secured a further $320 million, bringing its total disclosed funding to $454 million and valuing the company at $2.3 billion. The article framed this as a “$2.3B bet that video games can train AI agents for the real world,” emphasizing the firm’s focus on agents that can understand their “place in the world” across digital and physical settings.

Axios followed on June 26, confirming the $320 million Series A round at a $2.3 billion post‑money valuation and noting that General Intuition is already in talks for a Series B. Khosla Ventures led the round, joined by General Catalyst, Hedosophia, Bezos Expeditions, Innovation Endeavors, and Nico Rosberg. Citing a Goldman Sachs report, Axios framed the deal within a broader shift of AI investment into the “physical economy,” with General Intuition’s gaming‑based approach pitched as a faster, cheaper path to building both world models and large action models.

In public comments, CEO Pim de Witte cast the lab’s existence itself as a challenge to industry orthodoxy: “It shouldn’t have been possible to start a frontier lab in 2025. The doors were shut, they said. But thanks to Vinod [Khosla] … it was.”

Continue reading https://foxvector.com/stories/019f1132-1e2f-1e43-723c-16c36b791eb0