Trending: On-device modelsSearch
iHeartGeek
iTECH

NVIDIA RTX Spark desktops bring local AI PCs to ASUS, Dell and MSI

NVIDIA says RTX Spark compact desktops from ASUS, Dell and MSI are now available, bringing up to 128GB of unified memory to local AI workloads.

ASUS and NVIDIA RTX Spark compact desktop PCs, shown as white and black mini desktops

NVIDIA says new RTX Spark compact desktops from ASUS, Dell and MSI are now available to buy, bringing the company’s local-AI hardware platform to a desktop form factor.1 The machines are aimed at developers, creators and AI enthusiasts who want to run models, agents and demanding creative workloads on a dedicated Windows PC rather than relying entirely on cloud services.

The headline specification is up to 128GB of unified memory, paired with a 20-core NVIDIA Grace CPU and a Blackwell RTX GPU with 6,144 CUDA cores. NVIDIA claims up to 1 petaflop of AI performance and says the systems can run supported 120-billion-parameter models locally.1 Those figures are vendor claims, not independent benchmark results, so they should be read as an indication of the platform’s target workloads rather than a guarantee for every application.

The RTX Spark desktops also include 10-gigabit Ethernet for moving large model files, datasets and creative assets across a local network. NVIDIA says the machines support native CUDA and Linux development tools through Windows Subsystem for Linux, while local processing can keep code and files on the PC and reduce cloud API costs for suitable tasks.1

The first desktop systems come from ASUS, Dell and MSI. NVIDIA’s own launch material shows compact systems including the ASUS ProArt GR1X and MSI EdgeMesa N AI+ Series. Acer, GIGABYTE and HP systems are expected to follow, while Microsoft is planning a Surface RTX Spark Dev Box.1 NVIDIA has not supplied a complete UK price or regional availability breakdown in the announcement, so buyers outside the US should check local listings before treating the machines as immediately orderable.

The desktop launch arrives alongside software support aimed at making the hardware useful beyond model demos. NVIDIA highlights Windows-on-Arm support across Adobe applications, CUDA-accelerated creative workflows, ComfyUI integrations and Blender 5.3’s DLSS 4.5 Ray Reconstruction support in the Cycles viewport.1 The company also says developers can use local agents and fine-tuning tools, including workflows built around Hermes Agent, NVIDIA NeMoClaw blueprints and Unsloth.

Our opinion

RTX Spark desktops make a sensible case for local AI as an always-on workload: a compact machine can keep agents, renders and model jobs running without tying up a travelling laptop. The more important question is value. Until ASUS, Dell and MSI publish clear UK pricing and independent tests show how these systems perform against conventional desktops and cloud costs, the platform is promising hardware rather than an obvious recommendation. The availability announcement is still significant because it moves RTX Spark from an unusually capable laptop proposition into a broader desktop family.