RTX Spark
#2 in KI-Inferenz-Hardwarenvidia · siet Angekündigt am 31. Mai 2026 (GTC Taipei/Computex 2026); Marktverfügbarkeit ab Herbst 2026 · 17× · tolest 29. Juni 2026
NVIDIA RTX Spark is a superchip (codenamed N1X) unveiled by NVIDIA and Microsoft in late May 2026, combining a 20-core Arm-based Grace CPU with a Blackwell RTX GPU featuring 6,144 CUDA cores and a shared unified memory pool on one package. It is designed for slim Windows laptops and compact desktop PCs, enabling local AI agents, creative workloads, and gaming without cloud dependency. It delivers up to 1 petaflop of AI compute (FP4) and up to 128GB of unified memory, and will ship starting fall 2026 in devices from manufacturers including ASUS, Dell, HP, Lenovo, Microsoft Surface, and MSI. NVIDIA has not officially disclosed pricing; figures circulating come from analyst estimates.
Features
| Manufacturing Process (nm) | TSMC 3nm process (Grace CPU and Blackwell GPU die) |
| License | Proprietary NVIDIA chip platform; only available as a pre-built OEM device, not purchasable separately as a processor |
| Platform | Windows-on-Arm PC platform (laptops & compact desktops), Copilot+ PC category |
| Price | No official NVIDIA pricing; analyst estimates (Morgan Stanley): N1X approx. $2,899, N1 approx. $1,799 |
| Compute Performance (FLOPS/TOPS) | Up to 1 petaflop of AI compute (FP4 with sparsity) |
| Release Date | Announced May 31, 2026; laptops/desktops available starting fall 2026 |
| Memory | Up to 128 GB unified LPDDR5X memory, approx. 300 GB/s bandwidth |
| Availability | From fall 2026 via ASUS, Dell, HP, Lenovo, Microsoft Surface, MSI; Acer/GIGABYTE to follow later |