General Intuition seeks $300M at $2B valuation

General Intuition, a New York-based startup building foundation models for embodied AI, is reportedly in discussions to raise approximately $300 million, according to sources familiar with the matter. This funding round would value the company at just over $2 billion, marking a significant jump from its $134 million seed round eight months ago. The startup spun out from Medal, a platform for sharing video game clips, leveraging Medal's massive dataset of 2 billion videos per year from 10 million monthly active users. Sources indicate that prominent backers including Jeff Bezos and Eric Schmidt have committed funds, alongside existing investors Khosla Ventures and General Catalyst. The fresh capital is expected to accelerate development of General Intuition's world models, which train AI agents to navigate physical and simulated environments.

Unique dataset from Medal powers AI training

General Intuition's core advantage lies in its access to Medal's proprietary dataset, which captures 2 billion first-person gameplay videos annually from 10 million active users. This dataset is distinct because it provides interactive, real-time perspectives from gamers, allowing the AI to learn spatial-temporal reasoning—how objects move, interact, and change over time in 3D environments. The startup's co-founders, including Pim de Witte, Eloi Alonso, Adam Jelley, and Vincent Micheli, bring expertise in world modeling and simulation, aiming to train agents that can perceive, anticipate, and act in real-world scenarios. Unlike competitors that sell world models as products, General Intuition uses them to train AI agents for robotics and gaming applications. The dataset has reportedly attracted interest from major AI labs, including OpenAI, which previously attempted to acquire Medal, according to sources.

Competitive landscape in world models heats up

The world model space is becoming increasingly crowded, with startups like Runway, Decart, and World Labs releasing their own models for simulating environments. Google's Genie 3 recently integrated Google Maps data for more realistic real-world simulations, signaling big tech's entry into the field. However, General Intuition differentiates itself by focusing on agent training rather than selling world models as standalone products. The startup's unique dataset from Medal offers a path to viability that competitors lack, according to sources. Near-term commercial use cases include gaming and robotics training, where AI agents can learn to navigate complex environments through simulated interactions. The fundraising round reflects growing investor confidence in embodied AI, with General Intuition positioning itself as a leader in this niche.

Funding to scale compute and launch new product

General Intuition plans to use the $300 million raise to expand its compute capacity, a critical requirement for training large-scale world models, according to a source familiar with the matter. The startup aims to release a new product by the end of summer or early fall 2026, though details remain undisclosed. This timeline aligns with the company's rapid growth since spinning out from Medal in late 2025. The compute scaling is essential for processing the massive dataset of 2 billion videos annually, which requires significant GPU resources. The product launch could target robotics companies or gaming studios, leveraging the trained agents for real-world applications. Sources say the funding round is still in negotiations, with terms subject to change.

Implications for embodied AI and robotics

General Intuition's focus on training AI agents through gaming data could have broad implications for embodied AI, where machines learn to interact with physical environments. The startup's world models enable agents to perceive and react in real time, bridging the gap between simulation and reality. This approach could accelerate robotics development by providing a cost-effective training method without physical hardware. However, ethical considerations arise around data privacy, as Medal's user-generated content is used for AI training. The startup's valuation surge—from $134 million seed to $2 billion in eight months—reflects market enthusiasm, but questions remain about long-term scalability and competition from big tech. As the world model space evolves, General Intuition's unique dataset may give it a moat, but rivals are also racing to secure similar data sources.