Runway has launched a $10 million venture fund to support early-stage AI startups while transitioning toward building general world-model AI [3].

This strategic pivot signals a move away from niche video generation toward a broader ambition to map and simulate physical reality. By developing world models, Runway aims to compete with larger laboratories such as Google and OpenAI, moving beyond simple content creation to create a foundation for a wider ecosystem of AI applications.

Based in New York, the company has raised approximately $860 million [1] and currently holds a valuation of $5.3 billion [2]. The new initiative, known as the Builders program, was announced on March 31, 2026 [3].

The company's shift into world models became a point of public discussion in April 2026 [4]. Unlike traditional video AI, which predicts the next pixel in a sequence, world models attempt to understand the underlying laws of physics and spatial relationships. This evolution allows the AI to simulate environments that are more consistent and interactive, a critical step for applications ranging from robotics to advanced simulation.

Runway is utilizing the $10 million fund to foster a network of developers who can build on top of its evolving models [3]. By backing early-stage startups, the company intends to accelerate the adoption of its technology across various industries. The New York-based firm is positioning itself as a platform provider rather than just a tool for creators [1, 4].

The transition comes as the AI industry moves toward "embodied AI," where models must understand the physical world to function in real-time environments [4]. Runway's focus on world-modeling is intended to bridge the gap between generative media and functional physical intelligence.

“Runway is moving from AI‑generated video to building general world‑model AI”

Runway's transition from a creative tool to a world-model developer reflects a broader industry trend toward General Artificial Intelligence. By investing in a startup ecosystem and focusing on physical simulations, the company is attempting to move up the value chain from a software application to a foundational infrastructure provider, directly challenging the dominance of the largest AI labs in the U.S.