NVIDIA and Microsoft Bring RTX Spark to Windows PCs, Enabling People to Run AI Agents Locally

NVIDIA and Microsoft are bringing a new generation of AI capabilities to Windows PCs with RTX Spark and new agent-focused software. The technology is designed to let people run powerful AI models locally – without relying entirely on the cloud. Together, the companies are reshaping the PC into a more capable, always-on tool for working with AI.

RTX Spark-powered laptops, including the Surface Laptop Ultra, are available for preorder from October 7, with sales beginning October 16, allowing people to run powerful AI models locally on Windows PCs, according to NVIDIA news.

NVIDIA and Microsoft are also building Windows around a new generation of AI agents. Microsoft announced the general availability of Microsoft Execution Containers (MXC), which allow AI agents to run safely and continuously in the background – while remaining under Windows’ control. The goal is to make Windows a secure platform where agents can work, be monitored and managed.

RTX Spark is at the center of the hardware side of this effort. The compact systems combine an NVIDIA Blackwell RTX GPU with a Grace CPU and up to 128GB of unified memory. NVIDIA says the systems can deliver up to one petaflop of AI performance – giving users enough power to run large AI models directly on their PCs instead of always using cloud services.

The new systems are designed for more than AI development. Creators can use them for video and graphics work, while gamers can take advantage of features such as DLSS, Reflex and G-SYNC. Developers can also use the same NVIDIA AI software and CUDA tools across RTX Spark and NVIDIA’s larger AI systems, making it easier to move models and workflows between different machines.

Microsoft also previewed NVIDIA DGX Station for Windows, a much more powerful desktop system aimed at developers, researchers and businesses. It uses NVIDIA’s GB300 Grace Blackwell Ultra Desktop Superchip and offers 748GB of coherent memory and up to 20 petaflops of FP4 AI performance. The system is designed to let teams run and develop large AI models and always-on agents locally – while continuing to use Windows applications and tools.

Together, the announcements show how NVIDIA and Microsoft see the Windows PC changing as AI becomes more important. Instead of sending every AI task to the cloud, future PCs could handle increasingly powerful models and agents directly on the device. With RTX Spark for everyday users and DGX Station for more demanding workloads, the companies are aiming to make local AI a bigger part of how people work, create and use their computers.