Chip Variants

Specifications

Overview

The NVIDIA RTX Spark Superchip is an Arm-based system-on-chip announced by NVIDIA and Microsoft on May 31, 2026 at GTC Taipei during Computex. It fuses a Blackwell RTX GPU with a Grace CPU on a single die — an architecture previously reserved for data-center hardware — to power Windows-on-Arm laptops and compact desktop PCs.

The flagship configuration pairs a 20-core Grace CPU with a Blackwell RTX GPU featuring 6,144 CUDA cores and fifth-generation Tensor Cores with FP4 precision, linked by NVLink-C2C. NVIDIA rates the platform at up to 1 petaflop of FP4 AI performance with up to 128 GB of unified memory — enough to prototype, fine-tune, and run inference on large local models entirely offline.

MediaTek collaborated on the custom CPU design. Microsoft added Windows optimizations for unified memory, workload-profile scheduling, and power/thermal management via the Power and Thermal Framework, plus the Windows 11 Prism emulator for x86 applications. RTX Spark PCs join the Copilot+ PC category.

Initial systems ship Fall 2026 from Asus, Dell, HP, Lenovo, Microsoft, and MSI (with Acer and Gigabyte following), including the Microsoft Surface Laptop Ultra and a Surface RTX Spark Dev Box (128 GB unified memory, 100 W thermal envelope). It succeeds the DGX Spark / GB10 Grace Blackwell Superchip as NVIDIA’s flagship consumer AI compute platform.