A new kind of Windows PC is now on sale, and it’s designed around one idea: running AI on your own desk instead of in the cloud. On Wednesday, November 18, NVIDIA announced that the first RTX Spark compact desktops from ASUS, Dell and MSI are available, with more models from Acer, GIGABYTE, HP and Microsoft on the way.
Here is what these small machines are, what’s inside, what they’re good for and who should care.
What RTX Spark desktops are
According to the NVIDIA blog, RTX Spark compact desktops are dedicated Windows PCs for local AI models and always-on agents. They join RTX Spark laptops, which are already available.
The pitch is simple. A laptop goes where you go. A small desktop can stay home and keep working, running agents, scheduled jobs, renders or image generation while your laptop travels or does something else.
What’s inside
NVIDIA says the RTX Spark desktops include:
- A 20-core NVIDIA Grace CPU.
- An NVIDIA Blackwell RTX GPU with 6,144 CUDA cores.
- Up to 128GB of unified memory, shared between the processor and graphics.
- Up to 1 petaflop of AI performance.
- 10-gigabit Ethernet for moving big files quickly.
That much memory matters because large AI models need a lot of room. NVIDIA says these systems can run 120-billion-parameter language models locally, render 3D scenes over 90GB in real time and edit high-end 4:2:2 video smoothly.
Why “local AI” is the selling point
Most people use AI through cloud services. RTX Spark is built for running supported models on your own hardware. NVIDIA says that keeps locally processed code and files on the PC and avoids per-token cloud API charges for those tasks.
It’s not all-or-nothing, though. A new NVIDIA tool called Switchyard, now available in the Hermes agent, automatically routes tasks between local and cloud models. Everyday tasks can run on the device, while advanced reasoning can still use the cloud.
What creators get
NVIDIA and Adobe are expanding native Windows on Arm support across Adobe’s apps on RTX Spark:
- Adobe Premiere accelerates AI editing features such as Enhance Speech, Object Mask, Scene Edit Detection and Auto Reframe.
- Photoshop adds faster performance for the Remove Tool and native Camera Raw support.
- Illustrator and InDesign now have native Windows on Arm versions, with NVIDIA citing up to 30 percent higher performance in Illustrator compared with its emulated version.
- Substance 3D Painter 12.2 runs natively on Arm for the first time, in public beta.
Blender 5.3 also adds DLSS 4.5 Ray Reconstruction to its Cycles viewport, giving 3D artists near-final-quality lighting previews while they work.
What developers get
For developers, NVIDIA highlights native CUDA support, Linux tools through the Windows Subsystem for Linux and the ability to fine-tune supported models with tools like Unsloth. Microsoft also announced at Ignite that developers can run GGUF models from Hugging Face inside Windows ML using a llama.cpp backend.
Who it’s for
RTX Spark desktops are aimed at a specific crowd:
- Creators who generate images and video or work with large 3D scenes.
- Developers building and testing AI apps and agents.
- AI enthusiasts who want to experiment with bigger models at home.
If you mostly browse, stream and use AI chatbots in a web browser, a regular PC or laptop is still plenty. NVIDIA did not list prices in its announcement, so check the PC makers’ sites for current pricing and configurations.
Things to keep in mind
- These systems use Arm-based processors. Most major apps now run natively, but check that any specialized software you rely on supports Windows on Arm.
- “Local” only applies to models that actually run on the device. Cloud features still send data to the cloud.
- More models are on the way from Acer, GIGABYTE and HP, plus a Microsoft Surface RTX Spark Dev Box, so it may pay to compare options.
The wrap-up
RTX Spark desktops are NVIDIA’s bet that the next must-have PC is a small box that runs AI for you around the clock. For creators and developers who lean heavily on AI tools, that could mean faster work and more control over their data. For everyone else, it’s a glimpse of where home computing is heading.
Image: Notebook Computer on Desk by Image Catalog via Flickr, CC0 1.0