Nvidia RTX Spark

Arm-based system-on-chip by Nvidia From Wikipedia, the free encyclopedia

Nvidia RTX Spark is an Arm-based system on chip and computing platform developed by Nvidia for Windows laptops and compact desktop computers. Announced by Nvidia and Microsoft on May 31, 2026, RTX Spark combines a 20-core Nvidia Grace CPU with a Blackwell RTX GPU and unified memory, and is intended for local artificial intelligence, creative, and gaming workloads on Windows on Arm devices.[1][2]

Designed byNvidia
Cores
  • 20 CPU cores
Memory (RAM)
  • Up to 128 GB unified memory
GPUBlackwell RTX GPU
Quick facts General information, Designed by ...
Nvidia RTX Spark
General information
Designed byNvidia
Physical specifications
Cores
  • 20 CPU cores
Memory (RAM)
  • Up to 128 GB unified memory
GPUBlackwell RTX GPU
Co-processor5th-generation Tensor Cores
Architecture and classification
ApplicationWindows laptops and compact desktop PCs
MicroarchitectureGrace
Instruction setArm
Close

History

Nvidia had previously supplied an Arm-based Tegra system-on-chip for Microsoft's Surface RT.[3]

In October 2023, Reuters reported that Nvidia was developing Arm-based central processing units intended to run Windows, as part of Microsoft's effort to broaden the Windows on Arm hardware ecosystem. The same report stated that Microsoft's exclusivity arrangement with Qualcomm for Windows-compatible Arm chips was set to expire in 2024.[4] In January 2025, Tom's Hardware reported that Nvidia was developing Windows on Arm chips under the codenames N1 and N1X, with MediaTek involved in the project.[5]

At CES 2025, Nvidia announced Project DIGITS, later renamed DGX Spark, a compact AI-oriented computer based on the GB10 Grace Blackwell Superchip.[6] In September 2025, Tom's Hardware reported that Nvidia CEO Jensen Huang had described N1 as the Arm product used in DGX Spark and related products, connecting the previously reported N1/N1X project with GB10-based systems.[7]

Nvidia and Microsoft announced RTX Spark on May 31, 2026, at Nvidia GTC Taipei during Computex.[1][8] Nvidia stated that MediaTek collaborated on RTX Spark's custom CPU design.[1]

Specifications

RTX Spark combines a 20-core Nvidia Grace CPU with a Blackwell RTX GPU containing 6,144 CUDA cores and fifth-generation Tensor Cores with FP4 precision. The CPU and GPU are connected using Nvidia's NVLink-C2C chip-to-chip interconnect. Nvidia rates the platform at up to one petaflop of AI compute, with support for up to 128 GB of unified memory.[1][8]

Microsoft stated that Windows on RTX Spark includes optimizations for unified memory, heterogeneous scheduling, power and thermal management, and the Windows 11 Prism emulator for x86 and x86-64 applications. The company also said RTX Spark PCs would join the Copilot+ PC category and would include local AI processing through NPUs in addition to GPU acceleration.[2]

Software support

Microsoft and Nvidia stated that RTX Spark systems would support native Arm versions of creative and technical applications, including Adobe Photoshop, Adobe Premiere, Blender, DaVinci Resolve, Maxon Cinema4D, Maxon Redshift, Topaz Photo, CapCut, Cubase, Bitwig Studio, Affinity by Canva, and MATLAB through Prism. Microsoft also stated that Windows game compatibility on RTX Spark would be supported by native anti-cheat implementations from Epic's Easy Anti-Cheat and BattlEye, expanded Prism compatibility, and Xbox PC app support.[2]

Devices

Nvidia stated that RTX Spark laptops and compact desktops would be available in fall 2026 from Asus, Dell, HP, Lenovo, Microsoft Surface, and MSI, with models from Acer and Gigabyte to follow.[1] Microsoft announced the Surface Laptop Ultra as one of the first RTX Spark-based laptops.[2][9] Microsoft also announced the Surface RTX Spark Dev Box, a compact developer system using RTX Spark with 128 GB of unified memory and a 100-watt thermal envelope.[10]

References

Related Articles

Wikiwand AI