General Intuition, the New York startup building a foundation model that trains AI agents to move through space and time, is in talks to raise money at a $6 billion pre-money valuation. According to TechCrunch AI, the round brings in new backers including Valor Equity Partners, Point72 Ventures, and Seven Seven Six, with existing investors Khosla Ventures and General Catalyst also joining. What stands out here is the speed: TechCrunch AI reports the raise comes just weeks after the company closed $320 million at a $2.3 billion valuation.
That is nearly a 3x jump in valuation in a matter of weeks. It tells you how hard investors are competing to get into physical AI right now.
What the company actually does
CEO Pim de Witte spun General Intuition out of Medal, his video game clip-sharing platform, last October. The clever part is the data. Medal sits on hundreds of millions of hours of gameplay, and crucially, it captured “action labels” alongside the footage. Those are records of which buttons a player pressed and exactly when.
Why does that matter? Most video data shows you what happened on screen but not the intent behind it. Action labels connect the visual world to the decisions that drove it. Investor Vinod Khosla told TechCrunch that he sees those labels as central to the “emergence of intuition,” meaning a model’s ability to generalize to tasks it was never explicitly trained on.
That is the whole bet. Train an agent on enough action-plus-outcome data across virtual worlds, and it learns to navigate real ones.
Where the money goes
General Intuition plans to point the fresh capital at robotics. The company wants to push its general model toward physical embodiments, which means two big cost centers:
- Compute. The startup has a partnership with neocloud provider CoreWeave to feed its training runs.
- Talent. Expect aggressive hiring as it moves from game worlds into hardware.
A source close to the deal told TechCrunch the round is oversubscribed, and the company is still fielding investor interest. That is not a company scrambling for cash. That is a company picking who gets in.
Why this matters
This is significant because it signals where the smart money thinks the next frontier sits. Large language models trained the industry on text. The race now is toward models that understand action, movement, and the physical world. General Intuition is one of the clearest bets on using gameplay as the training ground for that shift.
The investor lineup carries its own message. TechCrunch AI notes that Valor Equity Partners is best known for backing SpaceX, and that this would be the first AI lab the fund has invested in since. When a fund with that track record moves into a new category, other capital tends to follow.
Here is the context worth holding onto. A year ago, “physical AI” and “world models” were mostly research talking points. Now they are commanding multi-billion-dollar valuations before the products ship. General Intuition going from spinout to a reported $6 billion valuation in under a year is a marker of how fast this category is heating up.
What to watch next
A few things will tell you whether the hype matches the substance:
- Does the round close at $6 billion? TechCrunch AI notes the deal is still being finalized, and Valor’s investment has not been formally confirmed.
- Robotics results. The real test is whether gameplay-trained agents actually transfer to physical hardware. Talk is cheap; a robot that generalizes is not.
- The data moat. If action labels prove as valuable as Khosla claims, expect rivals to start hunting for their own labeled action datasets.
For practitioners, the takeaway is practical. If you work in robotics, simulation, or agent training, the value of paired action-and-outcome data just got a very loud endorsement. Datasets that capture not just what happened but why are becoming the currency of this next wave.
The story is still developing, and the numbers could shift before the round is signed. You can find the full details in the original report at TechCrunch AI.