Runway Explores Robotics as New Revenue Stream to Fuel Growth

For the past seven years, Runway has developed visual generation tools for the creative industry. Now, the company sees a fresh opportunity for its technology in robotics.

Based in New York, Runway is recognized for its AI world models that generate video and photo content—essentially large language models that simulate versions of the real world. The company launched Gen-4, its video generation model, in March, followed by Runway Aleph, a video editing model, in July.

As Runway's world models became more advanced and realistic, the company started receiving inquiries from robotics and autonomous vehicle firms interested in the technology, according to Anastasis Germanidis, Runway’s co-founder and CTO, in an interview with TechCrunch.

"We believe this capability to simulate the world has broad applications beyond entertainment, even though entertainment remains a major and growing focus for us," Germanidis stated. "This technology makes training robotic policies that interact with the real world—whether in robotics or self-driving cars—far more scalable and cost-effective."

Germanidis noted that collaborating with robotics and autonomous vehicle companies wasn't part of Runway’s initial vision when it launched in 2018. It was only after companies from these and other industries reached out that Runway realized its models had a broader range of applications than originally anticipated.

Robotics companies are now utilizing Runway's technology for training simulations, Germanidis explained. He added that training robots and self-driving cars in real-world environments is expensive, time-consuming, and difficult to scale for most companies.

Although Runway acknowledges that its simulations won't replace real-world training entirely, Germanidis emphasized that companies can gain significant value by running simulations on Runway’s models, which can replicate highly specific scenarios.

Tech and VC heavyweights join the Disrupt 2025 agenda

Netflix, ElevenLabs, Wayve, Sequoia Capital, Elad Gil—these are just some of the top names joining the Disrupt 2025 agenda. They’re here to share insights that help startups grow and gain a competitive edge. Don’t miss the 20th anniversary of TechCrunch Disrupt—a unique opportunity to learn from leading voices in tech. Secure your ticket now and save over $600 before prices increase.

Tech and VC heavyweights join the Disrupt 2025 agenda

Netflix, ElevenLabs, Wayve, Sequoia Capital—just a few of the industry leaders joining the Disrupt 2025 agenda. They’ll share the insights that help startups grow and sharpen their competitive edge. Don’t miss the 20th anniversary of TechCrunch Disrupt—your chance to learn from the best in tech. Get your ticket today and save up to $675 before prices rise.

Unlike real-world training, using these models enables more straightforward testing of specific variables and situations without altering other aspects of the scenario, he added.

"You can take a step back and simulate the results of different actions," he explained. "For instance, if a vehicle takes one turn instead of another, or performs a certain maneuver, what will happen? Simulating these alternative outcomes from the same starting point is extremely challenging in the physical world—where keeping all environmental factors constant while testing a specific action is nearly impossible."

Runway isn't the only company exploring this field. For example, Nvidia recently introduced the latest iteration of its Cosmos world models, along with additional robot training infrastructure, earlier this month.

The company doesn't plan to create entirely separate model lines for its robotics and autonomous driving clients, Germanidis said. Instead, Runway intends to fine-tune its existing models to better serve these sectors. A dedicated robotics team is also in development.

Germanidis added that although these industries weren't part of the company's original pitch to investors, they now support this expansion. Runway has raised over $500 million from backers including Nvidia, Google, and General Atlantic, and is valued at $3 billion.

"We see the company as being built on a principle, not just focused on a market," Germanidis remarked. "That principle is the idea of simulation—creating increasingly accurate representations of the world. Once you develop such powerful models, they can be applied across diverse markets and industries. The industries we serve are already evolving, and they'll continue to transform as generative models grow even more capable."