NVIDIA RTX Spark Compact Desktops Out Now

Glass Epoch DeskOriginal digestSource: NVIDIA Newsroom

AI-assisted digest, checked against the original source before publishing. How we work

NVIDIA's RTX Spark desktops from ASUS, Dell and MSI ship today, pairing up to 128GB unified memory with Windows AI tooling for local models and always-on agents.
NVIDIA RTX Spark Compact Desktops Out Now
Photo: BoliviaInteligente · free stock license

NVIDIA RTX Spark compact desktops from ASUS, Dell and MSI are available today, extending the RTX Spark family beyond laptops to a dedicated Windows PC for running local AI agents, image and video generation, and model fine-tuning. The company positions the machines as a persistent home for AI workloads that can keep going while a laptop travels or handles other tasks.

The hardware centers on a 20-core NVIDIA Grace CPU paired with an NVIDIA Blackwell RTX GPU carrying 6,144 CUDA cores and up to 128GB of unified memory, rated at up to 1 petaflop of AI performance. Ten-gigabit Ethernet is included for moving large datasets, model files and creative assets between desktops and networked storage. NVIDIA frames the memory footprint as room for larger local models running alongside RTX-accelerated creative and development tools.

For operators, the pitch combines locality with uptime: running supported models on the machine keeps processed code and files on the PC and sidesteps per-token cloud API charges for those tasks. NVIDIA cites the ASUS ProArt GR1X mini PC as an example built to run personal AI agents 24/7, with scheduled tasks, asset generation, long renders and other background work continuing beyond an active editing session.

The software ecosystem is expanding in parallel. Native Windows on Arm applications from Adobe, the NVIDIA DLSS Ray Reconstruction feature in Blender and local generation workflows in ComfyUI arrive on RTX Spark hardware. Developers also get native CUDA support and Linux development tools through Windows Subsystem for Linux, can configure agents with Hermes Agent or NVIDIA NemoClaw blueprints, and fine-tune supported models using Unsloth.

Availability will widen over time: mini PCs from Acer, GIGABYTE and HP are coming soon, as is the Microsoft Surface RTX Spark Dev Box, joining desktops now on sale from ASUS, Dell and MSI. What remains to be seen is whether developers and creators treat these boxes as genuinely always-on infrastructure — and how local throughput, model support breadth and the avoided API spend actually compare with the cloud workflows they are meant to complement or replace.

Key facts

  • • ASUS, Dell and MSI RTX Spark desktops go on sale today.
  • • Acer, GIGABYTE and HP mini PCs plus a Surface Dev Box follow later.
  • • 20-core Grace CPU, Blackwell GPU with 6,144 CUDA cores, 128GB unified memory.
  • • Rated up to 1 petaflop AI performance; 10-gigabit Ethernet onboard.
  • • Unsloth fine-tuning; agent blueprints via Hermes Agent and NVIDIA NemoClaw.

Why it matters

Builders get a fixed-cost, on-prem alternative to per-token API billing for agents, generation and fine-tuning, with enough unified memory to hold larger local models. Expect agent tooling and creative apps to be the wedge — the practical questions now are model coverage and sustained local throughput versus cloud baselines.

Spotted an error? Send a correction.

Get the day's AI news in one email

The Daily Epoch · 60-second read · unsubscribe anytime

Related