General Intuition nears $6B valuation on gaming-data bet
The New York startup trains AI agents on hundreds of millions of hours of gameplay — and has nearly tripled its valuation within months.

Illustration · AI-generated (AI IN LIFE)
At a glance
- Reported: 24 August 2026 (TechCrunch, Stockpil)
- Valuation in talks: $6B pre-money — up from $2.3B in June 2026
- Last round: $320M; new investors: Valor, Point72, Seven Seven Six
- Training data: hundreds of millions of hours of gameplay plus action labels from Medal
- Compute partner: CoreWeave; focus: robotics applications
General Intuition is in talks to raise at a $6 billion pre-money valuation — nearly three times the $2.3 billion it commanded in its $320 million round in June 2026. TechCrunch reports the round is oversubscribed. New backers include Valor Equity Partners, Point72 Ventures and Seven Seven Six, alongside existing investors Khosla Ventures and General Catalyst.
The company's edge is its data. Spun out of the gameplay clip platform Medal in October 2025 by CEO Pim de Witte, it trains on hundreds of millions of hours of gameplay footage — including so-called action labels, the recorded player inputs behind every moment of video.
From that combination it builds a foundation model teaching agents to navigate space and time — first in games, increasingly in robotics. Investor Vinod Khosla puts it this way: action labels will be "a key part of the emergence of intuition" in machines.
The fresh capital is earmarked for compute — CoreWeave is the infrastructure partner — and for hiring, with a clear focus on robotic embodiments. That puts General Intuition squarely in the trend XPeng and Bosch are fuelling as well: the industry is betting embodied AI is the next big market.
The source of the advantage is worth noting: not web text, but interaction data from games. Companies holding proprietary, hard-to-copy data sources can reach multi-billion valuations in the AI race without running a frontier lab.
FAQ
What does General Intuition do?
It trains a foundation model for agents that navigate space and time — using gameplay video and recorded player inputs as training data.
Why is gaming data valuable?
It pairs video with the actions behind it (action labels). Models learn how actions change the world — a foundation for robotics.
Is the round closed?
No. Reports describe advanced talks over an oversubscribed round at a $6 billion valuation.


