From film tools to world models: How New York art school startup Runway is taking on Google

Runway, the AI video startup behind tools used in Hollywood productions, is betting that the future of artificial intelligence lies not in language but in video and so-called world models. Rebecca Bellan reports for TechCrunch that the company, now valued at $5.3 billion, is pushing into territory that puts it in direct competition with Google and other well-funded rivals.

World models are AI systems trained to simulate environments and predict how they behave. Rather than learning from text on the internet, these models learn by observing the world directly. Runway co-founder and co-CEO Anastasis Germanidis argues this approach can go beyond the limits of language-based AI.

“Language models are trained on the entire internet, on message boards and social media, on textbooks, distilling the existing human knowledge,” Germanidis said. “But to get beyond that, we need to leverage less biased data.”

From art school to AI frontier

Runway was founded in 2018 by three people who met at NYU’s Tisch School of the Arts. Germanidis grew up in Athens and studied neuroscience and film. Co-CEO Cristóbal Valenzuela and chief innovation officer Alejandro Matamala Ortiz both came from Santiago, Chile. None of them followed the typical Silicon Valley path.

The company started with a simple goal: use AI to make everyone a filmmaker. Its tools, including the latest Gen-4.5 video model, now power production workflows at ad agencies and studios. Runway has signed deals with Lionsgate and AMC Networks. The company employs 155 people across offices in New York, London, San Francisco, Seattle, Tel Aviv, and Tokyo.

According to one of its founders, Runway added $40 million in annual recurring revenue in the second quarter of 2026.

A bigger ambition

Runway launched its first world model in December and plans to release another this year. The company has also started a robotics unit, which Germanidis says has already produced real-world deployments. He sees world models as scientific infrastructure with applications in drug discovery, climate modeling, and anti-aging research.

“If we can build a better scientist than human scientists, we can accelerate progress in how we understand the universe and how we solve problems,” Germanidis said.

The competition is formidable. Google’s Veo model competes with Runway’s video business, while Google’s Genie world model targets the same long-term space. Startups Luma AI and World Labs are on similar paths, having raised $900 million and $1.29 billion respectively. OpenAI, despite raising around $175 billion, shut down its video platform Sora in March after burning roughly one million dollars per day in compute costs.

Runway has raised $860 million to date, including a $315 million round in February backed by AMD Ventures and Nvidia. One open question is compute access. Training large AI models requires dedicated access to powerful computing clusters at scale. Kian Katanforoosh, CEO of Workera and a lecturer at Stanford, noted that no one has yet proven the leap from video intelligence to generalized reasoning through world models. He also raised concerns about whether Runway has the compute infrastructure needed for the task.

“How are you going to build a foundational model without a cluster?” Katanforoosh asked. Runway holds deals with CoreWeave and Nvidia but has not confirmed whether it has the kind of dedicated cluster access that frontier model training requires.

Katanforoosh still sees a path for Runway. He pointed to ElevenLabs, the AI audio startup that has outperformed OpenAI and Google on benchmarks despite having far fewer resources. Early Runway investor Michael Dempsey, managing partner at Compound, described the founders as people who move quickly and have been “right more often than not.”

Valenzuela, for his part, sees the company’s distance from Silicon Valley as a strength. “Rules are just rules they invented,” he said. “Scrub them all and start again.”