← All news·2026-06-18·5 min read

General Intuition Raises $300M To Build AIs With World Models

World Models: AI that understands space and time.

aiworld-modelsembodied-aifundingrobotics

General Intuition, the spin-out from Medal, a social network for gamers, which launched in October 2025 with a $134M seed round led by Jeff Bezos, Eric Schmidt, Khosla Ventures and General Catalyst, is close to raising a $300M round at a $2B valuation to train world models. These are foundation models that enable AI to understand space and time, as opposed to just language. A world model is like a language model, except instead of learning to generate the next word in a sentence, it learns to generate the next frame in a scene, e.g. how objects will move, what the layout of a room is, what the view from a first person perspective will be like one second into the future.

General Intuition trains these world models on Medal’s data. Medal is a social network for gamers, with 10 million monthly active users who upload ~2 billion videos per year, mostly interactive first person gameplay in a 3D world. All of these videos have some kind of causal relationship between frames — the player is navigating through the world and causing things to happen — giving the model a rich sense of agency and cause-effect relationships. This makes it perfect for training AI agents that can navigate the physical world, whether that be a home assistant robot, self driving vehicle, drone, or non-player character in a game that feels like it understands its surroundings. In fact, OpenAI allegedly offered to buy Medal’s data before General Intuition was spun off into a separate company. Competitors in the space include Decart, Runway, and World Labs (founded by Fei-Fei Li).

Why gameplay video is the prize

Why does the world need World Models?

Why this matters for you

World models are the final piece needed to create intelligent agents that can interact with the physical world. We already have chatbots that can hold intelligent conversations but don’t really understand much about the physical world — they will gladly say that the sofa will fit through the door even though it won’t. But if you want an AI agent to do anything in the real world — whether that be driving a car, flying a drone, or controlling a home assistant robot — you need it to understand space and time, and that’s what world models provide. The fact that General Intuition is raising another $300M round on top of their $134M seed round from just a few months ago shows that there is a mad dash to create the next generation of AI after chatbots: AI that can do stuff. They plan to use the funds to grow their infrastructure and launch their first product later this summer/early fall 2026. But I wouldn’t get too excited just yet — world models look great on paper but tend to fall apart when you throw them into the real world, and a $2B valuation for videos of people playing games is a pretty big leap of faith.

ℹWhat I'd actually do

Don't trade on the hype, but do watch the category. If you build products, the practical takeaway is that 'AI that understands physical space' is about to get cheaper and more available — useful for anything involving robotics, AR, navigation or simulation. When General Intuition's first product lands, judge it on one question: does it predict what happens next in a real scene better than today's video models, or just look impressive in a curated clip?

Source: techcrunch.com

Free course

Stop reading about AI — start building with it

The free Claude Code course: your first site, tool or game — no coding. No upsells, no cross-sells — nothing to buy here.

Start free →
EAEvgenii Arsentev

Author

Evgenii Arsentev

PhD · Founder & CEO, Dad4Kids