Nvidia’s RTX Spark superchip brings 1-PFLOP AI-agent power to Windows PCs

Nvidia has introduced RTX Spark, a new AI "superchip" designed specifically to run personal AI agents directly on Windows laptops and desktops. The chip combines an Arm-based 20‑core Nvidia Grace CPU with a Blackwell RTX GPU via NVLink-C2C, delivering up to 1 petaflop of AI compute and up to 128GB of unified memory. Nvidia positions RTX Spark as the foundation for a new class of "AI personal computers" that can run large language models with up to 120 billion parameters and 1 million-token contexts locally, while also supporting demanding graphics workloads such as rendering 90GB-plus 3D scenes, editing 12K 4:2:2 video, and playing ray-traced AAA games at 1440p and over 100 frames per second.

The first RTX Spark systems will be slim Windows laptops with all‑day battery life and compact desktop PCs launching this fall from ASUS, Dell, HP, Lenovo, Microsoft Surface, and MSI, with additional models from Acer and Gigabyte expected later. Nvidia and Microsoft are jointly developing new Windows security primitives and the Nvidia OpenShell runtime to let users run AI agents such as OpenClaw and Hermes Agent securely and privately on-device, using hardware sandboxes. The platform targets creators, AI developers, and gamers, and has broad ecosystem backing, with more than 100 software developers on board and Adobe rearchitecting Photoshop and Premiere to exploit RTX Spark for up to 2x faster AI and editing performance. Nvidia is initially planning more than 30 laptops and 10 desktops based on the chip, signaling a direct push into the consumer PC CPU market alongside established players.