Skip to content
The Daily Circuit
Technology

NVIDIA Begins Selling RTX Spark Compact Desktops for Local AI Work

NVIDIA has started selling RTX Spark compact desktops built by ASUS, Dell and MSI, positioning the systems as dedicated Windows PCs for local artificial intelligence agents, image and video generation, and model fine-tuning.

By The Daily Circuit AI Desk (automated, AI-generated)

4 min read

NVIDIA RTX Spark compact desktop for local AI workloads

AI-generated image (flux-1-schnell), not a photograph of the events described.

Key facts

  • RTX Spark compact desktops from ASUS, Dell and MSI are available starting November 18, 2026.[1]
  • The systems use a 20-core NVIDIA Grace CPU, Blackwell RTX GPU with 6,144 CUDA cores, up to 128GB unified memory, up to 1 petaflop AI performance, and ten-gigabit Ethernet.[1]
  • They can run 120-billion-parameter large language models locally, render 90GB+ 3D scenes in real time, and edit 4:2:2 video in real time.[1]
  • Mini PCs from Acer, GIGABYTE and HP and the Microsoft Surface RTX Spark Dev Box are coming soon.[1]
  • Running supported models locally avoids per-token cloud API charges for those tasks.[1]
  • Adobe Premiere, Photoshop, Substance 3D Painter, Illustrator and InDesign are receiving native Windows on Arm or RTX acceleration updates for RTX Spark.[1]
  • Blender 5.3 adds DLSS 4.5 Ray Reconstruction to the Cycles viewport.[1]
  • Windows Server 2028 support is coming to NVIDIA DGX Station for Windows.[1]

NVIDIA has started selling RTX Spark compact desktops built by ASUS, Dell and MSI, positioning the systems as dedicated Windows PCs for local artificial intelligence agents, image and video generation, and model fine-tuning. The announcement was made on November 18, 2026 [1].

NVIDIA has started selling RTX Spark compact desktops built by ASUS, Dell and MSI, positioning the systems as dedicated Windows PCs for local artificial intelligence agents, image and video generation, and model fine-tuning. The announcement was made on November 18, 2026 [1].

The desktops join RTX Spark laptops that are already available. Mini PCs from Acer, GIGABYTE and HP are coming soon, as is the Microsoft Surface RTX Spark Dev Box [1].

Each compact desktop is built around a 20-core NVIDIA Grace CPU, an NVIDIA Blackwell RTX GPU with 6,144 CUDA cores, and up to 128GB of unified memory. NVIDIA said the systems deliver up to 1 petaflop of AI performance and include ten-gigabit Ethernet for moving large datasets, model files and creative assets between the desktop and networked storage [1].

NVIDIA highlighted three example capabilities: running locally hosted large language models with as many as 120 billion parameters, rendering 3D scenes of 90 gigabytes or more in real time, and editing 4:2:2 video smoothly in real time. Because supported models run on the PC, locally processed code and files stay on the device, which helps users avoid per-token cloud application programming interface charges for those tasks [1].

Software ecosystem and creative workflows

The company described an expanding software ecosystem for RTX Spark laptops and compact desktops. Native Windows on Arm applications from Adobe are part of the rollout, along with the DLSS Ray Reconstruction feature in Blender and local generation workflows in ComfyUI [1].

Developers can use native CUDA support and Linux development tools through Windows Subsystem for Linux. Agent setup is supported through Hermes Agent or NVIDIA NemoClaw blueprints, and supported models can be fine-tuned using Unsloth. NVIDIA Switchyard, now available in Hermes, routes agent tasks automatically between local and cloud models so that everyday work can stay on device while advanced reasoning uses cloud resources [1].

Creative workflows can connect Photoshop and ComfyUI through Hermes Agent, letting creators use natural-language instructions to turn sketches into rendered images and animations or to animate an image entirely on the device. NVIDIA demonstrated this at Adobe MAX, where attendees created images and animations for a shared digital Miami landscape [1].

RTX Video Frame Generation and RTX Video HDR, already available in ComfyUI, can expand a generated clip to 60 frames per second and convert it to high dynamic range for color grading within a single ComfyUI workflow [1].

Adobe integration

NVIDIA and Adobe are expanding native Windows on Arm support and RTX acceleration across Adobe’s creative applications. Adobe Premiere accelerates AI-powered editing features through NVIDIA TensorRT-RTX, including Enhance Speech, Object Mask, Scene Edit Detection and Auto Reframe. GPU acceleration also covers Film Impact transition and effects plug-ins, and the update adds support for professional video formats such as Canon RAW, Red RAW, ARRI RAW, ARRICORE, ARRI HDE, H.264 and HEVC in MXF, ProRes RAW and long-GOP video. NVIDIA said the update improves turnaround by more than 50 percent on everyday H.264 video deliverables [1].

Adobe Photoshop gains TensorRT-RTX acceleration for the Remove Tool and Arm-native Camera Raw support. At Adobe MAX, Adobe previewed a next-generation image processing engine with extensive GPU acceleration, including interactive Gaussian blur and Light Leaks filters [1].

Adobe Substance 3D Painter 12.2 is now optimized for RTX Spark with native Arm support in public beta, meaning artists can run it without a compatibility layer. Adobe Illustrator and InDesign now offer native Windows on Arm versions optimized for RTX Spark; Illustrator delivers up to 30 percent higher performance than its emulated version [1].

Blender 5.3 adds DLSS 4.5 Ray Reconstruction to the Cycles viewport, giving artists near-final-quality ray-traced previews of lighting, reflections and materials in real time. The feature requires Blender 5.3 and the latest NVIDIA Game Ready Driver or NVIDIA Studio Driver [1].

Windows Server and Windows ML updates

Also announced at Microsoft Ignite, Windows Server 2028 is coming to NVIDIA DGX Station for Windows, which is powered by the GB300 Grace Blackwell Ultra Desktop Superchip. The expansion is intended to give organizations more hardware options for accelerated inference and local AI workloads [1].

Windows ML updates let developers run any GGUF model from Hugging Face within Windows ML using the llama.cpp backend. An experimental DxCGC backend lets game developers compile and run machine learning models through DirectX 12. Windows APIs also add GPU support for AION Instruct and VSR, and developers can build high-performance AI models for Windows ML using the command line [1].

Other local AI updates mentioned by NVIDIA include the llama.cpp v1.0.0 release, Ricursive Intelligence’s Ripple-65B Preview open model, and Runway’s Praxis-1, a 27-billion-parameter world action model that can run locally on GeForce RTX 5090, RTX PRO 6000 Blackwell, DGX Spark and DGX Station [1].

Context

The RTX Spark compact desktops are designed to keep agents and scheduled tasks running while a laptop travels or handles other work. NVIDIA said running supported models locally keeps locally processed code and files on the PC and avoids per-token cloud application programming interface charges for those tasks [1].

Why it matters

The launch expands NVIDIA’s local AI hardware portfolio beyond laptops into compact desktops, with mini PCs from Acer, GIGABYTE and HP and the Microsoft Surface RTX Spark Dev Box coming soon [1].

What could happen next

AI-generated analysis based on the sources below. Not a forecast or recommendation.

NVIDIA is positioning the RTX Spark desktops as dedicated systems for local AI models and always-on agents, with software support spanning Adobe creative tools, Blender, ComfyUI and Windows ML [1].

Risks and open questions

No pricing was disclosed in the announcement [1].

Sources

  1. 1.NVIDIA (primary source): NVIDIA RTX Spark Compact Desktops Out Now, 18 November 2026

How this article was produced. This story was researched and written by an automated system using the open model nvidia/nemotron-3.5-super-vl (via OpenRouter), working only from the sources listed above. Automated checks verified that every citation, quote and figure matches those sources before publication. It was not written by a human reporter. Editorial policy · Report an error

  • Blender
  • AI desktops
  • Adobe
  • NVIDIA
  • RTX Spark
  • local AI