General Intuition seeks $300M raise at roughly $2B valuation

General Intuition is in talks to raise around $300 million at a valuation just over $2 billion, sources told TechCrunch. The New York startup plans to use the funds to scale up compute so it can release a new product by the end of summer or early fall.

What is General Intuition building?

General Intuition builds a foundation model designed to train AI agents how to move through space and time, training embodied AI and world models rather than selling models as the product. It trains these systems using Medal’s dataset, which the article says contains 2 billion videos per year from 10 million monthly active users, a dataset the company positions as uniquely useful because it captures interactive, first-person gameplay.

The company’s pitch centers on spatial-temporal reasoning: teaching machines to perceive, anticipate, and interact in real time in simulation using large volumes of gameplay footage. The founders named in the report are Pim de Witte, Eloi Alonso, Adam Jelley, and Vincent Micheli, with de Witte having co-founded Medal and now leading General Intuition.

Who is backing the round and what’s the company’s history?

General Intuition is pursuing a roughly $300 million raise and has secured funds from backers including Jeff Bezos and Eric Schmidt, along with existing investors Khosla Ventures and General Catalyst, according to sources. The raise comes eight months after General Intuition spun out of Medal with a $134 million seed round.

Medal is the platform for uploading and sharing video game clips that supplied the dataset General Intuition uses. The article notes that OpenAI previously attempted to acquire Medal and that other large AI labs have expressed interest in the dataset and the company.

How does General Intuition fit into the wider world-models race?

General Intuition is operating in a crowded, active field: the article lists startups like Runway, Decart, and World Labs as recently releasing world models, while Google’s Genie 3 has begun integrating Google Maps data for more real-world simulation capabilities. Where many companies build world models to sell the models themselves, General Intuition’s stated approach is to use world models to train agents, and to position the agents as the product.

That choice rests on the company’s unique data asset, Medal’s interactive gameplay footage. The report frames gaming and robotics training as near-term commercial use cases for world models, and it presents General Intuition’s dataset and agent-first product strategy as its route to commercial viability.

Why it matters

A dataset of first-person, interactive gameplay at the scale cited in the report — 2 billion videos per year from 10 million monthly active users — gives General Intuition an unusual training input for embodied AI. If the company converts that data into agents that perform reliably in simulation and real-world tasks, it could carve out a different commercial path from labs that sell world models themselves. The reported involvement of high-profile backers such as Jeff Bezos and Eric Schmidt signals investor confidence in that approach and adds fuel to competition over unique training datasets.

What to watch

Whether the round closes at around $300 million and the company follows through on its plan to release a product by the end of summer or early fall are immediate milestones. Also watch for any further interest or moves from big AI labs, given the article’s note that OpenAI previously tried to acquire Medal and that other labs have approached the company.