· via The Verge
First Nvidia RTX Spark mini PCs from Asus, Dell and MSI target local AI
The first desktops built on Nvidia's RTX Spark N1X chip are here from Asus, Dell and MSI, pairing a Blackwell RTX GPU with a Grace CPU for developers running AI models locally.

The first mini PCs built around Nvidia's RTX Spark N1X chip have gone on sale from Asus, Dell and MSI, The Verge reports. The desktops follow the RTX Spark laptops that arrived earlier this year, and they are aimed squarely at developers who want to run AI models on local hardware rather than through a cloud service.
What's under the hood
According to The Verge, the new machines pair a 6,144-core Nvidia Blackwell RTX GPU with a 20-core Grace CPU and support up to 128GB of memory. Nvidia claims the systems deliver "up to 1 petaflop of AI performance" and describes them as a "dedicated Windows PC for local models and always-on agents."
The software story matters as much as the silicon here. The systems offer native CUDA support, which means existing CUDA-based machine learning code and tooling can run without porting, and developers can reach Linux development environments through Windows Subsystem for Linux.
Flagship pricing for compact hardware
That hardware does not come cheap. The Verge notes that Asus's ProArt GR1X starts at $4,499 in a configuration with 64GB of memory and 1TB of storage, which puts these machines far above ordinary consumer mini PCs and in the same bracket as high-end professional workstations.
The Verge also points out that the systems can run games, and that Windows on Arm game support continues to improve, but at these prices gaming is not a realistic reason to buy one.
Positioned against the Mac mini and Mac Studio
The framing throughout The Verge's report is that these are alternatives to Apple's Mac mini and Mac Studio for developers working with local AI models and RTX-accelerated software. That extends a rivalry that began with the RTX Spark laptops, which compete with the MacBook Pro, and adds to Nvidia's own DGX Spark desktop launched last year.
On the application side, The Verge highlights AI-powered features in tools like Adobe Premiere and Photoshop, along with new DLSS 4.5 Ray Reconstruction support in Blender 5.3, as examples of the RTX-accelerated software these machines are built to serve.
More systems are on the way
The initial lineup is unlikely to remain this small. The Verge reports that additional RTX Spark desktops are planned from Acer, Gigabyte and HP, and that Microsoft's Surface RTX Spark Dev Box is scheduled to begin shipping this month.
Why it matters
This launch signals how quickly local AI has moved from an enthusiast experiment to a product category that every major PC maker wants a piece of. A compact desktop with native CUDA, a Grace CPU and up to 128GB of memory gives developers a dedicated, always-on machine for running models and agents without renting datacenter GPUs or sharing their main computer.
It also sharpens the competition with Apple, whose unified-memory Macs have become a default choice for local inference. Nvidia's advantage here is the CUDA ecosystem: researchers and developers with existing CUDA code can move to these systems with little friction, something Apple's platform cannot offer. The open question is price. At $4,499 and up, these are professional tools, and whether demand for local AI workloads is broad enough to sustain a whole category of machines at this cost will determine how many of the announced systems actually find buyers.
- #nvidia
- #mini-pc
- #local-ai
- #cuda
- #hardware