Runway AI has announced a new $315 million funding round backed by Nvidia and AMD Ventures.
General Atlantic, the primary investor in Runway's previous financing, led this round. TechCrunch reports that Runway's valuation has surpassed $3 billion, with a source indicating it now stands at $5.3 billion.
The company's world model algorithm creates 3D virtual environments from user prompts. Its models are used by clients like Shutterstock and Robinhood Markets. As of last April, Runway reportedly aimed to achieve $300 million in annual recurring revenue by the end of 2025.
Its latest model, GWM-1, launched in mid-December, enables engineers to build virtual environments for robot testing and onboard neural network training. For instance, logistics operators can simulate how robotic arms interact with various package types.
GWM-1 offers two additional versions optimized for tasks like 3D avatar generation. All three versions are built on Gen-4.5, a video generation model released in early December that quickly topped AI video generator rankings.
Gen-4.5 shows improved rendering of physical phenomena like momentum and fluid dynamics compared to its predecessor. However, it has limitations: objects may disappear randomly in one frame and reappear in another, and effects sometimes appear before the actions that cause them.
Runway states the new capital will fund model development, likely including fixes for Gen-4.5's rendering issues. The company also reportedly plans to hire more developers and marketing professionals.
This investment follows reports that rival World Labs, led by AI pioneer Fei-Fei Li, is seeking up to $500 million at a $5 billion valuation. The company offers a world model called Marble for integration into developer applications.
Like Runway, World Labs counts Nvidia and AMD Ventures among its backers. Adobe Ventures has invested in both startups, with the software maker joining several large financial institutions in Runway's latest round.
Both companies face competition from Google, which last month launched Project Genie. The tool generates 3D virtual environments from natural language prompts by combining Google's Nano Banana Pro image generator with the Genie 3 world model.
