General Intuition raises $320 million in Series A at $2.3 billion valuation

General Intuition has closed a $320 million Series A funding round, bringing the company's post-money valuation to $2.3 billion. The round was completed in January 2026 and publicly announced in June 2026. Combined with an earlier seed round of $133.7 million in October 2025, total capital raised stands at $454 million.

General Intuition trains AI models on billions of gameplay clips sourced from Medal, its sister company that allows gamers to save and share their gameplay footage. The approach focuses on spatial reasoning and world modelling through visual data from games, a direction that is attracting growing attention in the broader AI field.

Who participated in the round

Khosla Ventures led the Series A, as it did in the earlier 2025 seed round. Additional participants include General Catalyst, Jeff Bezos and Eric Schmidt through his family office Hillspire. Other investors include Nico Rosberg, the funds FuturePresent, Hexagon, Coalition and NPHard, co-founders of Remote and Cradle, and researchers affiliated with Google DeepMind and MIT.

The presence of technology veterans such as Bezos and Schmidt alongside specialised AI and deeptech funds illustrates the broad interest in General Intuition's approach. The company was founded in 2025 and has completed two substantial funding rounds in less than a year.

How the capital will be deployed

The capital will be directed towards four areas: research, model development, expanding compute capacity and hiring AI researchers and engineers. General Intuition aims to scale its training infrastructure in order to run increasingly large models on Medal's video data.

Concretely, an expansion of the commercial API is planned. The API is currently available to a limited group, but the company intends to open access more broadly to developers and businesses before the end of summer 2026.

MIRA as first public model

Alongside the funding announcement, General Intuition introduced MIRA, described as a multiplayer world model. MIRA generates interactive, multiplayer-compatible environments based on gameplay data. It is the first model from the company to be made externally visible.

World models are an active area of AI research: they are systems that build an internal representation of an environment and can reason within it about future states. General Intuition uses games as a training domain because the data available from Medal is large and structured, and because games provide a measurable and reproducible environment for evaluation.

Background and Dutch roots

General Intuition is described as a startup with Dutch roots, although the company operates internationally. The link with Medal provides a direct connection to an existing data source containing billions of hours of footage, which sets the approach apart from labs that rely on costly, manually collected datasets.

The combination of a large sister company in Medal and the speed at which General Intuition is attracting funding signals that investors are taking the use of gameplay data as a legitimate training domain seriously. For the broader European AI scene, the round is an indication that substantial capital is available for early-stage AI labs with a clear data-advantage strategy, even outside established American hubs. Dutch and European founders working on niche models with proprietary training data may see this as a precedent.