Nvidia's venture capital arm has invested in Reactor, a San Francisco startup that develops software to make world models run faster and more cheaply, the company said Monday. Sapphire Ventures also joined the round, adding to a Series A announced in May and bringing Reactor's total funding to $74 million.
The investment gives Nvidia a stake in the infrastructure layer of a technology that is drawing rapidly growing attention and capital from across the tech industry. World models are typically built on video data rather than the text-based large language models behind products such as OpenAI's GPT and Anthropic's Claude. Expectations are rising that the approach will help move artificial intelligence beyond screens and into cars, robots and other physical systems.
The deal lands during a burst of activity in the sector. In just the past week, AMD announced plans to acquire World Labs for $8.2 billion in stock, and video AI startup Runway unveiled the first open-weight version of its world model. The momentum echoes the wave of companies that emerged to train and operate large language models as generative AI took hold.
Reactor emerged from stealth in May with a specific focus: not building world models, but making them practical to use. The company sells a cloud service that manages computing capacity for running the models. «Once the model exists, how do you actually run it?» co-founder and CEO Alberto Taiuti said, describing the question that led him and co-founder Bryce Schmidtchen to start the company.
Both founders previously worked on Apple's Vision Pro headset. Taiuti also co-founded the AI video generator startup Luma AI, where he served as chief technology officer. That background in interactive media and video generation informs Reactor's approach to the infrastructure problem.
Reactor says Hollywood studios, advertising platforms and streaming services have active projects on its platform. But the company sees robotics as a major area of focus. Building conventional simulations for robot testing can be time-consuming and costly, Taiuti said, because developers must create three-dimensional objects and configure scenes by hand. World models could reduce some of that work by generating environments from prompts.
Reactor says it can run world models that respond to robot actions, allowing developers to test their software in those generated environments. Running the models in the cloud can give developers access to more computing power without adding that hardware to each robot. It also lets them upgrade the computers running the models without replacing hardware across an entire fleet, Taiuti said.
He said the company can adapt technology developed for interactive media to robotics because «similar models and similar workloads can also be run on the same core engine.» That overlap could allow Reactor to serve entertainment and industrial customers with the same underlying infrastructure.
Reactor plans to use the new investment to secure more computing capacity, expand its team and buy robot hardware to test its systems, Taiuti said. The spending priorities reflect the capital-intensive nature of the sector, where access to chips and cloud capacity is a competitive advantage.
Nvidia invested in Reactor through its NVentures venture arm. The chipmaker has developed its own Cosmos world models, alongside Isaac tools for robot simulation and Jetson computers that run AI on robots. Its investment in Reactor gives it a stake in the infrastructure for running interactive models, a layer that could grow in importance as more companies build and deploy world models.
The funding round underscores how quickly the world model ecosystem is taking shape, with chipmakers, cloud providers and startups jockeying for position in what could become the next major phase of AI development.