Runway built its name by giving filmmakers AI tools. Now it wants to prove that video generation can become something much bigger: a way to simulate the world.
Runway's next act is not really about making prettier clips. The New York startup is trying to use its creative AI business as the foundation for world models, systems that learn how environments behave and can predict what happens next. That puts a company with 155 employees in the same conversation as Google DeepMind, OpenAI, and other labs with far deeper compute budgets.
The timing matters. TechCrunch's May 15 profile frames Runway's push as a bet that video intelligence can become a broader AI platform, not just another tool for directors, marketers, and visual effects teams. The company was founded in 2018 by Anastasis Germanidis, Cristóbal Valenzuela, and Alejandro Matamala-Ortiz after they met at NYU's ITP program. It has since raised about $860 million, including a $315 million February Series E led by General Atlantic with backing from Nvidia, AMD Ventures, Adobe Ventures, Fidelity, AllianceBernstein, Felicis, and others. That round valued Runway at $5.3 billion.
Those numbers are large for a creative software company. They are smaller when the opponent is Google. But Runway also has something that matters in a market where many AI labs still burn cash before proving demand: one founder told TechCrunch the company added $40 million in annual recurring revenue in the second quarter of 2026.
That revenue gives Runway a practical advantage many research labs do not have. Its tools are used across film, advertising, marketing, and digital content. Lionsgate announced a partnership with Runway in 2024 to build a custom AI model around the studio's film and television library, while AMC Networks later moved to incorporate Runway's tools into marketing and television development work.
This is not a small detail. Paying users give Runway feedback, distribution, and a reason to keep improving the product in public. A filmmaker does not care whether a model has elegant internal representations. They care whether a character stays consistent, whether the camera move works, whether physics feel believable, and whether the clip can be used without wasting a day.
That pressure is useful. Runway's Gen-4.5 video model was released in December and later updated with native audio, long-form multi-shot generation, character consistency, and more advanced editing controls. Independent video benchmarks have placed it ahead of models from Google and OpenAI in some rankings, which helps explain why investors are willing to fund a startup taking on a much larger research race.
But benchmarks are only one part of the story. A video model that can produce a convincing shot is not automatically a model that understands the world. It may be learning enough structure to imitate motion, lighting, and cause and effect. Or it may be learning polished shortcuts. The difference matters if Runway wants to move from production workflows into robotics, drug discovery, climate modeling, and gaming.
Compute Is The Hard Question
Runway's argument is that video is a natural path to intelligence because the world arrives to humans as a stream of visual, physical, and temporal signals. If a model can learn from that, it may become better at simulation than systems trained mainly on text. That is the logic behind GWM-1, its first general world model, launched in December with applications for interactive worlds, robotics, and avatars.
Google is taking a similar question from the opposite direction. DeepMind has spent years working on agents, games, robotics, and simulated environments. Its Genie 3 work and Project Genie prototype are aimed at interactive worlds, while Veo 3.1 keeps pushing Google's video stack into creation tools like Flow, Vertex AI, and the Gemini API. Google also has what every AI startup wants most: direct access to enormous infrastructure.
Runway has tried to narrow that gap through partnerships and capital. TechCrunch previously reported that the company signed a deal with CoreWeave to expand compute capacity, and Nvidia and AMD's participation in its funding round is strategically important. Still, buying or renting enough compute to train frontier video and simulation models is a different burden for a startup than it is for Alphabet.
That is why Runway's move beyond Hollywood is both ambitious and necessary. The filmmaking market can support a serious company, but it may not justify the cost of training the next generation of world models on its own. Robotics companies need synthetic environments. Drug discovery teams need better simulation. Climate researchers need models that can reason across messy physical systems. Game studios want worlds that respond in real time. If Runway can serve even a few of those markets, its valuation starts to look less like a creative AI premium and more like a platform bet.
The risk is that video intelligence does not generalize cleanly. A model can generate a beautiful storm without understanding climate. It can show a robot arm picking up a cup without being reliable enough to train a real machine. It can render a molecule or a lab scene without helping discover a drug. The leap from visual plausibility to useful prediction is the whole challenge.
Runway does not need to beat Google everywhere. It needs to prove that a focused startup, grounded in real creative workflows and pushed by paying users, can build world models that are commercially useful before the largest labs turn the category into another infrastructure contest. That is what to watch next: not just whether Runway's videos look better, but whether its simulations start helping companies make decisions outside the editing room.
Also read: Samsung's AI chip boom is turning labor into a supply risk • Bill Ackman is betting Microsoft can outlast the AI spending scare • Lake Tahoe's power crunch shows AI's hidden infrastructure bill
