Runway raises $315M Series E to scale video generation compute and hit millions of media and entertainment customers

Summary

Runway has raised a $315M Series E, with CEO and co-founder Cristóbal Valenzuela describing the round as primarily a bet on compute. The company is scaling training infrastructure for its next generation of frontier video models and supporting inference demand from what Valenzuela says is now a base of millions of customers. The round closed late last year; Runway is only disclosing it now.

The immediate pressure point is capacity. Runway's Gen-4.5 video model has been drawing significantly more usage than the company anticipated, and Valenzuela says managing inference volume has become the central operational challenge. Alongside compute, the company is hiring across research, engineering, and sales.

Customer traction

Valenzuela names PayPal, Allstate, AMC, and Lionsgate as signed customers, alongside a broad set of agencies and marketing teams. His framing is that AI-first production is already the default decision for sophisticated buyers — studios are hiring AI chiefs and reorganizing workflows around it. The pitch he makes to enterprise customers is straightforward: tasks that took six months and an external agency can now be done internally, in days.

Model strategy

Valenzuela is explicit that Runway has to do both frontier model research and workflow product development simultaneously. Companies that focus only on workflows get leapfrogged by more capable models, he argues — a pattern he says has repeated consistently across the seven to eight years Runway has been operating. The companies that win are ambidextrous: they push the frontier and build the tooling on top of it.

On what video models actually are, Valenzuela's framing is more ambitious than a media tool pitch. He describes them as world simulation engines — systems that develop an innate understanding of the world by watching it, which means they can generalize to new tasks (rotoscoping, object removal, novel camera angles) from relatively few examples. That same simulation capability, he argues, has direct value outside entertainment, including robotics and autonomous vehicles, where models trained on video can teach robots rather than humans.

Runway has been first to market with video generation, world models, and real-time generation. Valenzuela is betting the $315M keeps that lead intact long enough for the next model cycle to compound it.