Nvidia announces RTX Spark, an AI-focused “superchip” intended for Windows laptops and desktop PCs. The company presents it as a step toward “reinventing the PC” for the AI era, aiming to run AI agents and other compute-heavy tasks locally rather than relying solely on cloud processing. Nvidia’s CEO Jensen Huang unveils the chip at Computex in Taipei, and multiple outlets report that the platform will debut in the fall on new systems from PC manufacturers including Dell, HP, ASUS, Lenovo, and MSI, with Windows for Arm in focus.
Several sources describe RTX Spark as Arm-based, combining an Nvidia Blackwell RTX-class GPU with an associated CPU/processing component and integrated AI capabilities. Nvidia claims peak AI performance around 1 petaflop, and media outlets cite a setup that includes thousands of Blackwell RTX GPU cores plus a built-in NPU. The unified memory architecture is described as configurable from 16GB to 128GB, with the chip drawing from low single-digit watts up to around 80W depending on configuration.
Nvidia and Microsoft describe co-optimization with Windows 11, including scheduling changes intended to balance CPU performance and efficiency for agent-like workloads, and the chip is positioned for tasks such as 3D rendering, video editing, and high-end gaming.
