Runway expands from AI video tools into “world models”

In short: Runway, a company known for AI video tools, is trying to build “world models” and compete with bigger labs like Google.

What's going on

Runway started in 2018 and became popular by helping filmmakers and ad agencies generate and edit video using AI. Its tools have been used by major media companies, and TechCrunch reports it has deals with Lionsgate and AMC Networks.

Now Runway is pushing beyond video creation. The company says it is working on “world models,” which are AI systems that can simulate an environment well enough to predict what will happen next (like a flight simulator, but for many kinds of scenes). Runway released its first world model in December 2025 and plans to launch another one in 2026.

Runway’s leaders argue that many AI companies focus too much on text, like chatbots trained on internet writing. They believe training AI on video and other real-world observations could help it learn how the world works, not just how people describe it. Runway’s co-CEO Anastasis Germanidis said this could lead to tools useful outside entertainment, including robotics testing and scientific research.

The challenge is scale. Training these systems can require huge amounts of computing power, meaning expensive specialized computers running for long periods. TechCrunch notes Runway has partnerships with CoreWeave and Nvidia, but the company did not confirm whether it has guaranteed access to the largest computing setups that top AI labs often rely on.

What to watch

Runway is valued at $5.3 billion and says it added $40 million in annual recurring revenue in the second quarter of 2026, but it faces deep-pocketed rivals. Google is a direct competitor in AI video and is also building its own world model projects. The key question is whether Runway’s strength in video will translate into world models that work reliably in the real world.

Source: TechCrunch AI