NVIDIA RTX Spark Compact Desktops Launch: Local AI Power on Your Desk
ASUS, Mechrevo, and MSI are first out of the gate with NVIDIA RTX Spark compact desktops, packing up to 128GB unified memory and 1 Petaflop of AI performance for local agents, creation, and dev work.
NVIDIA's RTX Spark family just got a whole lot more desk-friendly. ASUS, Mechrevo, and MSI have officially launched their RTX Spark compact desktop PCs, bringing local AI agents, image and video generation, and model fine-tuning to a dedicated Windows machine that can sit right next to your monitor.
So what's the big deal? These little boxes pack up to 128 GB of unified memory and the full NVIDIA AI software stack, which means there's plenty of room for larger local AI models. They can also run RTX-accelerated creative and development tools. The best part? Even if you grab your laptop and head out, or use it for other work, the desktop keeps chugging along running agents and scheduled tasks in the background.
Under the hood, you're looking at a 20-core NVIDIA Grace CPU, an NVIDIA Blackwell RTX GPU with 6,144 CUDA Cores, and up to 128 GB of unified memory, delivering up to 1 Petaflop of AI performance. There's also 10-gigabit Ethernet for fast transfers of large datasets, model files, and creative assets to and from network storage. That hardware translates to some seriously impressive capabilities: running local large language models with up to 120 billion parameters, real-time rendering of 3D scenes over 90 GB, and smooth real-time editing of 4:2:2 video.
The software side is expanding too. Adobe's native Windows on Arm apps, NVIDIA DLSS Ray Reconstruction in Blender, and ComfyUI local generation workflows are all coming to RTX Spark. Native CUDA support and Linux dev tools via Windows Subsystem for Linux (WSL) give developers a solid foundation to build and run local AI workflows. You can deploy agents with Hermes Agent or NVIDIA NemoClaw Blueprint, and fine-tune supported models with Unsloth. NVIDIA Switchyard is now available in Hermes too, automatically splitting agent tasks between local and cloud models so everyday stuff runs on-device while heavy reasoning taps the cloud, helping manage costs.
Creators aren't left out either. RTX Spark can run creative agents that connect to common apps and local AI tools, turning natural-language instructions into finished work. NVIDIA showed this off at Adobe MAX, where attendees used plain English to turn sketches into rendered images and animations, building a digital Miami landscape together. ComfyUI now supports RTX Video frame generation and RTX Video HDR too, so generated clips can be bumped to 60 FPS and converted to HDR for color grading, all within a single ComfyUI workflow on RTX Spark.
Adobe and NVIDIA are also deepening their partnership, expanding native Windows on Arm and RTX acceleration across Adobe's creative suite. Premiere gets TensorRT-RTX acceleration for AI features like Enhance Speech, Object Mask, Scene Edit Detection, and Auto Reframe, plus GPU acceleration for Film Impact transitions and effects. Support for pro video formats including Canon RAW, Red RAW, ARRI RAW, ARRICORE, ARRI HDE, MXF-wrapped H.264 and HEVC, ProRes RAW, and long GOP video means editors can cut turnaround times by over 50% on everyday H.264 delivery tasks. Photoshop adds TensorRT-RTX acceleration to the Remove Tool and native Camera Raw support on Arm-based RTX Spark, so photographers can zap unwanted objects faster without relying on translation layers. Adobe also teased a next-gen image processing engine at Adobe MAX with broad GPU acceleration, including interactive Gaussian Blur and Light Leak filters.
The ASUS, Mechrevo, and MSI compact desktops are available now, joining the already-shipping RTX Spark laptops. More are on the way soon from Acer, Dell, Gigabyte, HP, and Microsoft's Surface RTX Spark Dev Box. So, are you ready to give your local AI agents a permanent home on your desk? Let us know in the comments!
đź“° Related Articles
Nvidia's Driver Bug: What It Means for Your SSD and How to P…
This bug causes unnecessary wear on SSDs, potentially reducing their lifespan, and forces users to either revert to an older driver or apply a workaround, highlighting the importance of monitoring sys…
Did November's NVIDIA Driver Update Wear Out Your SSD? Here'…
Players on affected drivers should manually install the hotfix to stop excessive SSD writes, while a permanent fix remains pending.
Is Your SSD Dying? NVIDIA's Latest Drivers Might Be the Caus…
The bug can significantly reduce SSD lifespan due to excessive writes, forcing users to choose between driver rollback and losing features like G-Sync.
Gigabyte's AI TOP AERO with NVIDIA RTX Spark Hits Shelves
Gigabyte launches the AI TOP AERO, a compact Windows PC powered by NVIDIA RTX Spark for local AI workloads, content creation, and gaming.
GIGABYTE AI TOP AERO: A Tiny AI Powerhouse with NVIDIA RTX Spark
GIGABYTE's new AI TOP AERO packs an NVIDIA RTX Spark superchip, 20-core Grace CPU, and Blackwell RTX GPU into a roughly 1-liter chassis for local AI development and gaming.
AMD's Su Lisa Calls NVIDIA's Jensen Huang an 'Outstanding Leader'
At the TomorrowX summit in Austin, AMD CEO Lisa Su praised NVIDIA and departing CEO Jensen Huang as industry-driving talents, while steering focus back to AMD's own opportunities.