Nvidia’s RTX Spark is the commercial name for its new Arm-based “superchip” for Windows PCs, built on the flagship N1X CPU-GPU design. It’s an integrated SoC combining a 20-core Arm CPU, Blackwell RTX GPU, and up to 128 GB of unified LPDDR5X memory, designed for AI-heavy laptops and desktops running Windows on Arm.
Co-developed with MediaTek, the CPU uses a performance/efficiency core mix, while the GPU features 6,144 CUDA cores, with performance comparable to midrange desktop GPUs but only uses around 45–80W laptop power rating. The chip uses a 3nm TSMC process and supports up to 600GB/s memory bandwidth via a high-speed interconnect.
Nvidia positions RTX Spark as an “AI-first” platform capable of up to 1 petaFLOP of FP4 performance, enabling local execution of large AI models (up to 120B parameters), along with gaming at 1440p using DLSS and full RTX features. It also targets creators with advanced media and 3D capabilities.
N1X represents the top-tier Spark configuration, while standard N1 chips scale down in CPU cores, GPU size, power, and memory. All variants share the same core approach: Arm CPU, Blackwell GPU, and unified memory for AI-focused Windows PCs.
N1X chip:
20‑core Arm CPU in a 10+10 layout (10 Cortex‑X925 performance cores + 10 Cortex‑A725 efficiency cores)
Blackwell GPU with 48 SMs (6,144 CUDA cores), roughly in the same core-count class as a desktop RTX 5070
Up to 128GB LPDDR5X memory, unified between CPU and GPU
N1 chip:
12 CPU cores and 20 SMs (2,560 CUDA cores)
Up to 64 GB LPDDR5X
RTX Spark will launch in fall 2026 across devices from major OEMs like Asus, Dell, HP, Lenovo, and Microsoft, as part of Nvidia’s push into the “AI PC” market—bringing data center-class architecture into consumer laptops and desktops.
Here are the laptop and PC models slated to be released with NVIDIA RTX Spark:
* Asus ProArt P14/P15
* Dell XPS 16
* HP OmniBook X 14, OmniBook Ultra 16
* Lenovo Yoga Pro 9n
* Microsoft Surface Laptop Ultra
* MSI mini-PCs