Nvidia Backs Startup Reactor as World Models Drive New AI Investment Wave

San Francisco-based startup Reactor announced on Monday that it has added Nvidia’s venture capital arm, NVentures, and Sapphire Ventures to its cap table, expanding its Series A round and bringing its total funding to $74 million. The investment comes amid mounting industry enthusiasm for world models, an emerging branch of artificial intelligence designed to help systems understand and interact with physical environments such as robotics and autonomous vehicles.\n\nReactor specializes in software designed to make video-based world models run faster and at a lower cost, alongside managing a dedicated cloud computing platform. 'Once the model exists, how do you actually run it?' co-founder and CEO Alberto Taiuti told Fortune, highlighting the operational hurdles the company addresses. Taiuti, who previously worked on Apple’s Vision Pro and co-founded Luma AI, launched the startup alongside co-founder Bryce Schmidtchen to bridge the gap between large models and practical infrastructure.\n\nUnlike traditional large language models centered on text, world models process extensive video and physical data to simulate real-world physics and environments. Recent momentum in the sector includes AMD’s planned $8.2 billion acquisition of World Labs and Runway’s release of an open-weight world model. While Reactor's platform is currently utilized by Hollywood studios and streaming platforms, robotics remains its central commercial target by significantly reducing the cost and time needed to configure complex 3D simulation scenes.\n\nReactor plans to deploy the newly secured capital toward securing expanded computing capacity, growing its engineering team, and acquiring robotic hardware for system testing. For Nvidia, which has built its own Cosmos world models and Isaac robotics simulation tools, the investment secures a strategic stake in foundational infrastructure necessary to deploy interactive physical AI fleets at commercial scale.