NVIDIA moves RTX Spark into AI PCs

NVIDIA unveiled RTX Spark for Windows PCs at GTC Taipei 2026, moving from a graphics-card supplier toward a fuller platform role in personal computing. Microsoft backed the push by positioning Windows as “the native platform for local AI agents” and launching the Surface RTX Spark Dev Box, while Acer, Asus, Dell, Gigabyte, HP, Lenovo, and MSI lined up behind the chip.

The chip is presented as a complete SoC, combining Arm Grace CPU cores, CUDA cores, Tensor Cores, unified LPDDR5X memory, and local model execution. Jensen Huang framed the shift as a reinvention of the PC, where local agents can invoke tools, modify code, advance designs, and keep tasks running when users are away.

The move intensifies pressure on Intel, which had promoted AI PCs through Core Ultra and Intel 18A, while AMD, Qualcomm, and Apple are also pushing integrated AI computing. The broader change is toward AI SoC-centered systems, local inference, and developer ecosystems, with NVIDIA seeking to extend CUDA’s data-center flywheel onto laptops and desktops.