On June 25, 2026, New York startup General Intuition announced a $320 million round at a $2.3 billion valuation, led by Khosla Ventures with participation from General Catalyst, Jeff Bezos, Eric Schmidt, F1 world champion Nico Rosberg, and researchers at Google DeepMind and MIT. Added to the $134 million seed the company raised in October 2025, total disclosed funding now stands at $454 million.
The company is betting on one thing: the best material for teaching AI about the world is not text scraped from the internet, but games. General Intuition was spun out of Medal, the game-clip sharing platform founded by CEO Pim de Witte. Its core asset is hundreds of millions of hours of Medal gameplay, and every clip carries embedded action labels recording exactly which buttons the player pressed, and when. De Witte puts it plainly: “It works on anything that you can control using a game controller or a keyboard mouse.”
Deal Highlights
The round had been rumored since mid-June, when TechCrunch reported talks to raise roughly $300 million at around a $2 billion valuation; the final numbers landed slightly above the leaks. Vinod Khosla framed the thesis in one line: “In world models, I think the quantum leap is the emergence of intuition in the AI.” Intriguingly, Medal turned down an acquisition offer from a major AI lab before the spinout, and more offers have arrived since. Khosla’s response to the acquisition interest was confident: “At this point, it would be a data acquisition, which is sort of uninteresting.”
Why Gameplay Data Is The Secret Weapon
Most world models watch video and guess what happened beyond the frame. General Intuition’s data comes with the answers attached: the model does not just see a car take a corner, it knows what inputs the driver made at that instant. That lets one model do three jobs at once. It plays games — an agent demonstrated by the company played for 100 hours straight, per CPO Kent Rollins. It runs a generated world model used internally as a training environment nicknamed “the gym.” And it controls physical robots. Alongside de Witte, the founding team includes Eloi Alonso, Adam Jelley, and Vincent Micheli, and the bet rests on a data moat nobody else holds: action labels.
From The Virtual Gym To Real Robots
The flagship result is a quadruped robot fine-tuned with just eight minutes of real-world data. The logic: robots learn to walk, run, and carry objects inside simulation, then fine-tune in reality — dramatically cheaper than collecting real-world data from scratch. De Witte is explicit that the company will not become a vehicle maker: “We’re not gonna build a self-driving car company.” Instead, “We’re gonna make it 10 times easier for the next person to build” one.
Use Of Funds And The Nerve Marketplace
Most of the round goes to compute, through a CoreWeave deal, to pre-train the next version of the model. The company also plans to make its API broadly available by the end of the summer, and its current customers already span gaming, simulation, and robotics. A second product worth watching is Nerve, a jobs marketplace where gamers earn money by labeling data and teleoperating robots — effectively crowdsourcing the data supply chain to the player community, and giving ordinary people a slice of the robotics economy. The marketplace also solves a problem the company created for itself: as its robot fleet grows, the demand for human demonstration data and teleoperation hours grows with it, and Nerve turns that cost center into a paid gig network.
The Competitive Field And Its Red Lines
The world-model race is heating up. Fei-Fei Li’s World Labs has shipped its first commercial product, Decart’s new model can simulate hours of photorealistic driving, and Odyssey closed a round at a $1.45 billion valuation in mid-June. Most rivals sell simulation or generation; General Intuition’s differentiation is the action-labeled dataset and its robotics path, where the same weights that predict the next frame also decide the next action. The company has also drawn red lines: de Witte said “We don’t want to be an escalatory part of the system” — no lethal autonomy, though search-and-rescue applications are acceptable. For a startup sitting on defense-adjacent technology and raising money from billionaires, saying that out loud is itself a positioning choice.
Sources
- General Intuition’s $2.3B bet — TechCrunch
- General Intuition in talks to raise $300M — TechCrunch
- General Intuition lands $134M seed — TechCrunch
AI-assisted summary compiled from the sources above, reviewed by a human before publishing.