AI Video Company Shifts to World Model, and Jensen Huang Invests Again
Runway, a video technology company that has made significant strides in recent years, recently announced the completion of its Series E financing, raising $315 million (approximately 2.17 billion yuan). This funding round, led by General Atlantic, will be directed towards the development of a next-generation world model, marking a pivotal shift in the company’s strategy.
Founded in 2018 by three art students from New York University’s School of the Arts—Cristóbal Valenzuela (CEO), Alejandro Matamala, and Anastasis Germanidis—Runway initially gained traction with innovative video editing tools. Its flagship feature, “Green Screen,” introduced in late 2020, allowed users to extract individuals from videos with a single click, rapidly positioning the company as a leader in the design field. Following this success, Runway secured an $8.5 million Series A funding round, followed by $35 million in Series B financing.
However, rather than continuing to develop video tools, Runway pivoted towards the generative AI space, specifically investing in the research and development of the text-to-image model Stable Diffusion. This strategic decision laid the foundations for its subsequent foray into generative AI, which gained momentum after the emergence of ChatGPT in late 2022.
In December of that year, Runway completed its Series C financing, raising $50 million, and subsequently launched its AI video model, Gen-1, which enabled video editing functions. The iteration continued with the release of Gen-2, which became the first large-scale commercially available text-to-video model, setting the stage for further advancements.
The company’s rapid iteration culminated in the introduction of Gen-4 in April 2025, which incorporated physical laws to enhance the model’s understanding of materials and gravity. Runway’s post-investment valuation surged past $1.5 billion during this time, elevating it to unicorn status. Following a successful Series D financing round of $308 million, led by General Atlantic with participation from NVIDIA, Runway’s growth trajectory remained robust.
In early December 2025, Runway unveiled its upgraded model, Gen-4.5, which significantly improved the realism of AI-generated videos. This was quickly followed by the release of its first world model, GWM-1 (General World Models – 1). This autoregressive model, capable of generating images frame by frame in real time, includes three independent variants: GWM Worlds for simulated environments, GWM Avatar for virtual characters, and GWM Robotics for synthetic data generation for robot training.
Runway’s transition to a world model approach has attracted significant investments, notably from NVIDIA, which signaled its optimism regarding this technology’s potential. The interest in world models is not isolated to Runway; NVIDIA has also invested in other companies leveraging similar technologies, such as Waabi and Wayve, both of which focus on autonomous driving applications.
Waabi has developed Waabi World, an AI closed-loop simulator for traffic-scenario training, while Wayve’s GAIA series generates video data to train AI drivers. NVIDIA itself entered the fray with the release of NVIDIA Cosmos at CES 2025, aimed at generating realistic videos that adhere to physical laws, further emphasizing the industry’s shift towards physical AI.
During ICCV 2025, industry leaders, including Tesla, discussed their own developments in world modeling. Tesla’s vice president, Ashok Elluswamy, revealed that the company has created a world simulator to enhance data collection for extreme driving scenarios, addressing limitations in its current data sets. Similarly, Ideal Auto showcased its integration of a cloud-based generative world model with its vehicle systems to create a training closed-loop.
The increasing involvement of major players has kept the world model concept at the forefront of technological innovation. WeRide, known as “the first Robotaxi stock,” recently launched WeRide GENESIS, which generates and adjusts various edge scenarios for testing purposes. Waymo is also leveraging world modeling, building a model based on Google Genie3 to generate driving scenarios and LiDAR point clouds.
As the interest in world models expands, Runway’s shift towards this approach signals a broader evolution within the AI video landscape. The potential applications are vast, from enhancing generative video capabilities to creating realistic simulations for various industries. As major companies invest and innovate in this domain, the vision of generating not just videos but entire realistic worlds may soon become reality.
See also
Investors Shift Strategies, Emphasizing Governance and Control in AI’s Evolving Landscape
Montage’s $902M IPO Soars 64%, Signaling Investment Surge in China’s AI Ecosystem
Heliware Secures Rs 4.5 Cr Funding to Expand AI-Powered Geospatial Solutions Globally
India’s AI Market to Surge to $131.31 Billion by 2032, Fueling Digital Transformation
Healwell AI Unifies Platforms, Targets 2026 Growth with Strategic AI Enhancements
