Problem

Generative AI is powerful but difficult to deploy in applications that must respond at the speed of reality. Interactive video, gaming, media, commerce, advertising, and physical-AI simulations require millisecond-level latency, while conventional large-model training and inference can be too slow and computationally expensive. Decart’s core pain point is therefore the cost and latency of generating responsive, high-quality AI experiences at scale; its platform is designed to enable sub-35ms experiences and improve efficiency across large GPU clusters.

The clearest killer use case is real-time interactive worlds: turning video input and keyboard commands into an evolving game-like environment, or generating interactive video that responds instantly to users. This replaces pre-rendered or delayed content with environments that can be explored and controlled in real time, while also opening applications in robotics simulation and other physical-AI workflows.

Product / Service

Decart combines infrastructure software with real-time world models. Its DOS product is an AI optimization layer intended to make training and inference faster and more efficient, including across clusters of more than 1,000 NVIDIA H100 GPUs. On top of that infrastructure, Lucy generates real-time interactive video, while Oasis creates interactive environments for gaming, physical AI, and robotics simulations. Decart describes the overall stack as spanning hardware-level efficiency through world models.

The delivery model is a combination of enterprise technology partnerships and consumer-facing or developer-accessible AI experiences. Customers can use Decart’s technology for applications in media, commerce, advertising, and physical AI, while products such as Oasis demonstrate the user-facing experience: a video clip and keyboard input become an interactive generated environment. The benefit is lower-latency, more economical generative AI that can operate continuously and interactively rather than producing only static or batch outputs.

Market

Decart competes in the emerging market for real-time AI infrastructure and world models, at the intersection of generative AI infrastructure, interactive video, gaming, simulation, and physical AI. The available research does not identify a definitive list of direct competitors by name; instead, it positions Decart among leading players building systems that simulate and generate interactive digital environments. Its closest competitive set is therefore likely to include companies developing real-time generative media, world models, AI optimization infrastructure, and robotics simulation platforms, although the cited research does not establish specific head-to-head competitors.

Decart is demonstrably post-revenue and has substantial financing traction. The company reported contracts with major cloud providers, AI laboratories, and hyperscale companies, and Amazon is a strategic customer while NVIDIA is both an investor and business partner. In May 2026, Decart raised $300 million at an estimated $4 billion valuation, following a $100 million raise at a $3.1 billion valuation in August 2025; reported cumulative funding exceeded $450 million. The company’s partnerships and planned product expansion suggest it is scaling from an early product company into an enterprise infrastructure and platform provider, rather than operating as a pre-revenue startup.