Runway's bold AI pivot: Filmmaker origins to Google rivalry.

Runway, an AI video generation startup, defies the conventional Silicon Valley narrative. Its founders, hailing from Chile (two) and Greece (one), converged at NYU’s Tisch School of the Arts and established the company in New York, notably without the typical pedigree of Stanford or ex-Google alumni, nor the luxury of a nine-figure seed round that might defer revenue focus.

Many consider Runway to be among the most pivotal AI companies currently, not solely for its existing innovations but more significantly for its ambitious future endeavors.

The AI industry has predominantly pursued intelligence through language-based models over recent years, a strategy evident in leading large language models such as OpenAI’s ChatGPT and Anthropic’s Claude.

Runway, however, is charting an alternative course, positing that the next evolution of AI intelligence will emerge not from text, but from video and sophisticated 'world models' designed to comprehend the mechanics of reality itself, rather than merely human descriptions. While this distinction may appear academic, its potential impact is far-reaching.

According to Anastasis Germanidis, Runway’s co-Founder and co-CEO, the subsequent frontier in AI involves training models directly on real-world observational data. He contends that the pioneers in this domain will not be those who have merely refined language-based AI.

From Runway’s inviting, sun-drenched headquarters near Union Square, Germanidis explained to TechCrunch, "We’re basically bound by our own understanding of reality."

He further elaborated, "Language models are trained on the entire internet, on message boards and social media, on textbooks — distilling the existing human knowledge. But to get beyond that, we need to leverage less biased data."

Established in 2018, Runway has built its reputation on advanced video-generation models, such as its recent Gen-4.5, and AI tools that enable users to transform text prompts into editable, high-quality cinematic content.

Currently, Runway’s technology is integral to production workflows for filmmakers and advertising agencies, having secured partnerships with prominent media entities like Lionsgate and AMC Networks. Its tools notably contributed to films such as “Everything Everywhere All At Once.”

Runway is presently valued at $5.3 billion, and one of its founders indicated an addition of $40 million in annual recurring revenue during the second quarter of 2026.

Should Runway’s hypothesis that video generation is the conduit to world models prove successful, its ramifications could span industries from entertainment to pharmaceutical research. Conversely, failure risks Runway being overshadowed by rivals with significantly greater financial resources, most notably Google.

Over the past six months, the startup has actively pursued its strategy, moving beyond video generation to unveil its inaugural world model in December, with a second planned for release this year. World models are defined as AI systems capable of simulating environments with sufficient accuracy to forecast their behavior.

Runway is not alone in its endeavor to evolve physics-aware video models into comprehensive world models, which hold immediate applications in interactive entertainment, gaming, and robotics training. Startups like Luma and World Labs are pursuing similar paths, and Google’s Genie world model also aligns with this objective.

Ultimately, the collective ambition is to develop AI capable of addressing humanity’s most intractable challenges. This aspiration represents a significant evolution from Runway’s initial product, driven by both the burgeoning capabilities of the technology and founders inclined to explore its full potential.

Germanidis envisions world models as foundational scientific infrastructure. He posits that by training a single model on an increasing volume of sensory data and observations, one approaches a functional digital replica of the universe, enabling experiments at a pace far exceeding traditional laboratories. He highlights that a substantial portion of the scientific process involves waiting for results; compressing this wait, he argues, would inherently accelerate progress.

Germanidis stated, "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's passion for programming began at age 11 in Athens; he later moved to the U.S. at 18 to study neuroscience and film, eventually returning to computer science after a stint in Silicon Valley tech firms where he grew disillusioned with the culture. Co-CEO Cristóbal Valenzuela, a Santiago native, pursued economics before transitioning into film and software. Alejandro Matamala-Ortiz, also from Santiago and now Chief Innovation Officer, studied advertising and managed a design firm.

The trio met in 2016 at NYU’s ITP (Interactive Communications Program), a graduate course Valenzuela characterized as an "art school for engineers."

Matamala-Ortiz noted that all co-founders had, at various stages, harbored aspirations of becoming filmmakers. This shared ambition led Runway to commence with a straightforward mission: to leverage AI to empower everyone to be a filmmaker.

Following the release of their initial video generation model in February 2023 — which, by today’s standards, Matamala-Ortiz admits was "staggeringly unimpressive" — the mission evolved: could they enable everyone to be a *great* filmmaker?

Achieving this required expanding the team to its current size of 155 employees, distributed across offices in New York, London, San Francisco, Seattle, Tel Aviv, and Tokyo. Matamala-Ortiz further explained, "But throughout this process, we learned that these models can understand how the world works, and if you scale them, they can be useful for many other different things."

These "other things" include fields such as robotics, drug discovery, and climate modeling—complex problems that have challenged researchers for decades. Last year, Runway established a robotics unit, which Germanidis confirms has already yielded real-world testing and deployments.

Germanidis, among others, foresees the AI field progressing towards training singular models on diverse modalities—encompassing text, video, voice, and other sensor data—believing the resultant compounding effect is crucial.

His personal ambitious goal for Runway’s technology, provided sufficient time and resources, lies in developing biological world models and advancing anti-aging research.

Whether Runway can successfully translate its leadership in video generation into the realm of world models remains uncertain, as competition is intense. While Runway pioneered AI video generation, the pursuit of world models involves a distinct race with well-funded and highly regarded contenders, including Google, former Meta chief scientist Yann LeCun, AI ‘godmother’ Fei-Fei Li, and numerous emerging startups, all striving for the same objective.

Kian Katanforoosh, CEO of AI skills benchmarking firm Workera and a Stanford lecturer, observed that the connection between video intelligence and generalized reasoning through world models has yet to be definitively proven, though he doesn’t deem it impossible. He advised that for Runway to realize its world model ambitions, it must continuously acquire resources, with computational power being paramount.

While Runway maintains agreements with CoreWeave and Nvidia, the company has not confirmed whether it possesses dedicated cluster access—the assured, large-scale computational power essential for training cutting-edge models.

Katanforoosh questioned, "How are you going to build a foundational model without a cluster? I don’t think anybody can do that."

To date, Runway has secured $860 million in funding, including a $315 million round in February from strategic investors such as AMD Ventures and Nvidia. This capital places it broadly in line with immediate competitors like Luma AI ($900 million raised) and World Labs ($1.29 billion raised), according to PitchBook data.

However, Runway also faces formidable incumbents such as OpenAI, which CEO Sam Altman states has raised approximately $175 billion, and the tech titan Google, whose parent company Alphabet boasts a market capitalization of $4.86 trillion. Google, in particular, poses Runway’s most significant threat, with its Veo model directly competing in video generation and its Genie world model aiming for the same long-term world model objectives Runway is pursuing.

Katanforoosh referenced OpenAI’s decision to discontinue its video platform, Sora, in March, reportedly incurring compute costs of around $1 million daily against estimated revenues of only $2.1 million. His observation underscored that ample resources alone do not ensure success or survival, a point equally relevant for Runway.

Despite the challenges, Katanforoosh remains optimistic about Runway’s prospects, citing AI audio startup ElevenLabs as an example. ElevenLabs has surpassed OpenAI and Google on their respective benchmarks, despite having fewer resources and less pedigree. Katanforoosh suggests Runway could adopt a similar strategy.

Runway’s founders are acutely aware of this comparison. Valenzuela believes the startup’s absence of Bay Area "standardization" provides a distinct advantage, fostering not only diversity of thought but also necessitating a more resourceful, ‘scrappy’ approach. Without the substantial funding often available to Silicon Valley peers, Runway was compelled to prioritize early revenue generation.

Michelle Kwon, Runway’s Chief Operating Officer, affirmed that the company is not currently pressed to secure additional funding, even as computational requirements escalate with growth.

Early investor Michael Dempsey, managing partner at Compound, commented to TechCrunch, "Their background has led them to be early, to be right more often than not, and to build a culture that moves incredibly quickly."

For Valenzuela, this distinctive culture originates from his fundamental worldview. In his limited free time—a scarce commodity as co-CEO and a new father—he engages with literature, notably the Chilean poet Nicanor Parra. Valenzuela describes Parra as the philosophical antithesis of Pablo Neruda, advocating for a less formal, less academic poetry that belongs to the people rather than being governed by rigid rules.

Valenzuela articulated, "Rules are just rules they invented. That’s a driving force of how we do things at Runway. They say Silicon Valley is here and that’s where the startups are. Why? Those are just made up rules. Scrub them all and start again."