Over the past seven years, the runway has built visual generation tools for the creative industry. Now, it sees new opportunities for its technology: robotics.
The New York-based runway is known for its video and photo generation world models, or to create a simulated version of the real world model. Recently, the company released Gen-4, its video generation model, In March and Runway Aleph, its video editing modelin July.
As the world model of the runway begins to improve and become more realistic – the company is beginning to gain aggressive interest from robotics and self-driving car companies.
“We think this ability to simulate the world is beyond entertainment, even though entertainment is an increasingly field for us,” Gerlisidis said. “This makes training more scalable and cost-effective [robotic] Policies to interact with the real world, both in robotics and in self-driving. ”
GerliDis said the collaboration between robotics and self-driving car companies was not originally envisioned when the runway was launched in 2018. It wasn’t until robotics and other companies in other industries reached out that the company realized that their models were wider than initially thought.
Germisidis said robotics companies are using the technology of the runway for training simulations. He added that in reality, training only robots and self-driving cars is expensive for companies, takes a long time, and is difficult to scale.
While the runway knows this won’t replace real-world training in any way, the Germans say companies can get a lot of value running simulations on runway models because they have incredibly specific features.
TechCrunch Events
San Francisco
|
October 27-29, 2025
He added that unlike real-world training, using these models makes it easier to test specific variables and situations without changing anything else in the scheme.
“You can take a step back and simulate the effects of different actions,” he said. “If the car turned that, or did this, what would it be? Creating these rollouts from the same context, which is really hard to do in the physical world, basically keeping all other aspects of the environment the same and only testing the effects of the specific action you want to take.”
Runways aren’t the only company that wants to solve this problem. For example, NVIDIA released its latest version of Model of the universeAmong other robot training infrastructure, earlier this month.
Germisidis said the company will not release “completely independent models” for its robotics and self-driving car customers. Instead, the runway will fine-tune its existing models to better serve these industries. The company is also building a dedicated robotics team.
GerliDis added that although these industries are not in the company’s initial investment in investors, they are still making this expansion. There is a runway Raised over $500 million Investors from Nvidia, Google and General Atlantic are valued at $3 billion.
“Our perception of companies is really based on principles, not in the market.” “This principle is a simulation idea that can build more and more world representation. Once you have these truly powerful models, you can use them in a variety of different markets, including a variety of different industries. [The] We expect industries that already exist, and they will change even more due to the power of generative models. ”
