Runway, an AI video-generation startup based in New York, is positioning itself as a challenger to tech giants like Google and OpenAI by betting that the future of artificial intelligence lies in video rather than language. Founded in 2018 by two Chileans and one Greek who met at NYU's Tisch School of the Arts, Runway represents a departure from typical Silicon Valley AI companies — none of its founders attended Stanford, worked at Google, or raised a nine-figure seed round. The company has built its reputation on video-generation models, most recently Gen-4.5, which allow filmmakers and creators to turn text prompts into editable, cinematic content. Co-founder and co-CEO Anastasis Germanidis argues that the prevailing assumption in the AI industry — that intelligence lives in language — is fundamentally limited. While OpenAI and Anthropic have built powerful large language models trained on internet text, textbooks, and social media, Runway is investing in what it calls "world models" that learn directly from observational video data about how the physical world actually works. "To get beyond that, we need to leverage less biased data," Germanidis told TechCrunch from Runway's headquarters near Union Square. "Language models are trained on the entire internet... But to get beyond that, we need to leverage less biased data." The company's approach has already attracted significant commercial validation. Runway's technology now powers production workflows at major studios and advertising agencies, with formal partnerships signed with Lionsgate and AMC Networks. The company's tools were even used in the production of the Academy Award-winning film "Everything Everywhere All At Once." Financially, Runway is valued at $5.3 billion and reported adding $40 million in annual recurring revenue during the second quarter of 2026 alone. The stakes of Runway's bet extend far beyond entertainment. If video-based world models prove to be the next frontier of AI development, the implications could reach into fields like drug discovery and scientific research. However, the company faces formidable competition from well-resourced players who have invested heavily in language-based approaches. Runway's founders believe their observational approach to training AI could unlock capabilities that text-based systems simply cannot achieve, but whether that hypothesis proves correct remains to be seen.