World model startup General Intuition closes $220M investment
SiliconANGLE Maria Deutscher
General Intuition raised $220M at a $6.2B valuation. Its world model tech turns game footage into training data for robots and other AI.
Based on reporting by SiliconANGLE, Maria Deutscher — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
General Intuition has pulled in $220 million in fresh funding, and the price tag is doing a lot of the talking: a $6.2 billion valuation for a company that only spun out of Medal a year ago. Valor Equity Partners led the round, with Atreides, Seven Seven Six, Point72, Khosla Ventures and General Catalyst also putting in money.
The basic idea is simple, if the execution is not. Medal gives people a free way to capture and record video game footage. General Intuition takes that sort of footage and uses it to train AI models, then goes a step further by building world models that can generate synthetic footage for training projects. That matters because training data for robots is still awkward, slow and often painfully manual to create.
Its newest model, MIRA, arrived in June. General Intuition says it can keep running “infinitely without diverging,” which is a big claim in a field where many video generators top out at short clips. The company says MIRA can also handle scenes with multiple fast-moving objects, which is the sort of thing that matters when you’re trying to teach a robot not to crash into things.
The system is built for efficiency as much as capability. General Intuition says MIRA can generate 20 frames per second at 720 by 576 resolution on a single B200 graphics card. It uses latent diffusion, so it works on a compressed latent space instead of full video frames, and the model itself is relatively small at 5.6 billion parameters.
There’s a catch, though. MIRA is still a research demo, not a commercial training-data engine, and right now it can only generate synthetic footage for a single video game. Even so, General Intuition is already testing a commercial version with a limited set of customers in robotics, simulation and entertainment, and it has opened a waitlist. The new money will go toward hiring more AI researchers.
My take — AI-written commentary, not fact-checked reporting
This is another reminder that the AI gold rush is moving from chatbots to the messier business of simulation. The interesting part isn’t the valuation circus; it’s that people are paying up for tools that can make training data less miserable to produce. That’s a lot more grounded than the usual model-hype parade, which is nice for a change.
Read more about this at: SiliconANGLE