Why Runway Is Turning Toward Robotics as the Next Frontier for Growth

For the past seven years, Runway has focused on building tools that generate visuals for creative industries. But now, the company sees a far bigger horizon for its technology: robotics.

The New York–based startup is widely recognized for its AI-driven video and photo generation platforms, known as “world models” — systems that construct digital replicas of real-world environments. In March, Runway launched its video-generation model Gen-4, followed in July by Runway Elif, a video-editing model.

According to co-founder and CTO Anastasios Germanidis, once Runway’s models began producing simulations that looked convincingly real, interest started pouring in from robotics and autonomous vehicle companies eager to adapt the technology.

“The ability to digitally mimic the physical world has enormous utility beyond entertainment — even though entertainment remains a fast-growing core market for us,” Germanidis explained. “For robotics and self-driving systems, our models make training policies far more scalable and cost-efficient than relying exclusively on physical-world testing.”

When Runway launched in 2018, it never set out to serve robotics or autonomous driving. But once other companies in those sectors began approaching them, it became clear that the models had far broader applications than initially imagined.

Robotics companies are now using Runway’s simulations for training scenarios. Germanidis noted that running experiments in the real world is expensive, time-consuming, and often impractical to scale. Simulations, by contrast, allow companies to tweak highly specific variables without altering the broader environment.

“You can step back, test alternate decisions, and see what outcomes unfold,” he said. “If a car took this turn instead of that one — what would happen? Running those controlled variations is extremely hard in the physical world, but effortless in a simulation.”

Runway is not alone in this pursuit. Earlier this month, NVIDIA unveiled the latest version of its Cosmos world models, aimed at supporting robotics training infrastructure.

Unlike NVIDIA, however, Runway doesn’t plan to spin off dedicated models just for robotics or self-driving cars. Instead, the company intends to refine its existing platforms to serve those industries, while also assembling a specialized robotics team.

Interestingly, robotics was never part of the company’s original pitch to investors. Yet, Germanidis said investors have embraced the shift. Runway has already raised over $500 million from backers such as NVIDIA, Google, and General Atlantic.

“Runway was never built around chasing one specific market,” Germanidis emphasized. “It’s built on a principle — the principle of simulation: creating increasingly powerful representations of the world. Once you have models that strong, you can apply them across an extraordinary range of industries. Some of those industries are already being reshaped, and others will inevitably follow as these generative models grow more capable.”