Tech

Nvidia RTX Spark Mini PCs Launch: Asus, Dell and MSI Put 120B-Parameter AI on Your Desk

Nvidia has put its most powerful AI chip for the desk into a box the size of a lunchbox. The first Nvidia RTX Spark mini PCs from Asus, Dell and MSI went on sale on Wednesday 18 November 2026, and they are built to run artificial intelligence models with up to 120 billion parameters entirely on the machine, without sending the work to the cloud.

The launch lands a week into the most turbulent stretch in Nvidia’s recent history, and it is also the company’s clearest answer yet to a simple question: what happens when the AI models everyone is racing to build need somewhere cheaper and more private to run than a data centre? The first answer costs $4,499.

Nvidia RTX Spark Mini PCs at a Glance

ItemDetail
ProductNvidia RTX Spark compact desktops
On sale fromWednesday 18 November 2026
First makersAsus, Dell and MSI
ChipNvidia N1X superchip: 20-core Grace CPU, Blackwell RTX GPU with 6,144 CUDA cores
MemoryUp to 128GB unified LPDDR5X, shared between CPU and GPU
AI performanceUp to 1 petaflop (FP4 precision)
Local modelsUp to 120-billion-parameter models on-device
Entry priceAsus ProArt GR1X from $4,499 (64GB memory, 1TB storage)
MSI pricingEdgeMesa N AI+ listed from $4,699 (64GB); 128GB version $7,999
Power140W desktop chip, against 45–80W in the RTX Spark laptops
SizeAbout 1.1 litres (Asus ProArt GR1X)
Coming nextAcer, Gigabyte and HP systems; Microsoft Surface RTX Spark Dev Box
LaptopsRTX Spark laptops launched 16 October 2026 from $2,599

What the RTX Spark N1X Chip Actually Packs

The heart of every one of these machines is the same N1X superchip, and it is the reason a box this small can claim this much power. Nvidia pairs a 20-core Grace processor, co-designed with MediaTek, with a Blackwell graphics processor carrying 6,144 CUDA cores, and both share a single pool of up to 128GB of LPDDR5X memory.

That shared pool is the trick. In a normal PC the processor and the graphics card each have their own memory, and large AI models have to be shuffled between the two. Here the whole 128GB is available to the model at once, which is what lets a 120-billion-parameter model sit on the desk instead of in a cloud data centre.

Nvidia rates the chip at up to one petaflop of AI performance at FP4 precision, and the Asus ProArt GR1X quotes memory bandwidth of up to 300GB/s. The desktop version of the chip runs at 140W, well above the 45–80W envelope of the same silicon in the RTX Spark laptops, so the mini PCs should sustain higher speeds for longer jobs.

Asus, Dell and MSI: the First Three Boxes and Their Prices

Asus leads with the ProArt GR1X, the most fully specified of the launch machines, starting at $4,499 for 64GB of memory and 1TB of storage. MSI’s EdgeMesa N AI+ Series is listed on the company’s United States store from $4,699 for the 64GB model, rising to $7,999 for the 128GB version, though some outlets report entry prices from $4,399, so the exact street price depends on the retailer and the configuration.

Dell is named by Nvidia as one of the first three makers, with its own system on sale alongside the Asus and MSI boxes. The company has not published the same level of pricing detail as its two rivals, and buyers should check Dell’s store before assuming a figure. Acer, Gigabyte and HP are expected to follow with their own RTX Spark desktops, and Microsoft’s Surface RTX Spark Dev Box is also due to begin shipping.

The pricing tells its own story. At $4,499 and above, these are not living-room PCs, and they are not aimed at gamers. They are workstations for people whose daily work is AI, and Nvidia is pricing them like it.

A Desktop That Runs a 120-Billion-Parameter Model Without the Cloud

The pitch for running models locally is threefold: privacy, latency and cost. A developer working with sensitive data never has to upload it, a creator gets answers without waiting on a network round trip, and a team running the same model all day stops paying a cloud provider by the token.

That is why the 128GB figure matters more than the petaflop figure. The largest models most developers actually use today sit in the 70- to 120-billion-parameter range, and they need roughly that much memory to run at usable speed. A 128GB RTX Spark box clears the bar; a 64GB box runs smaller models comfortably and the biggest ones only with compromises.

There is a catch the launch coverage has been quick to flag. Reviewers note that stock of the first machines looks thin, and that the jump from 64GB to 128GB costs $3,300 on the MSI box alone. Buyers who need the full 128GB are paying a steep premium for it, and the value of the machine depends heavily on whether their models actually need that much memory.

Why Nvidia Is Betting on the Desk After a Week of Boardroom Upheaval

The timing is impossible to ignore. Jensen Huang, who co-founded Nvidia in 1993 and led it for 33 years, resigned as chief executive on 12 November 2026, and interim chief executive Colette Kress, who keeps her role as chief financial officer, led the company’s earnings call this week. The company’s shares have fallen sharply since the departure, and its market value has dropped below $4 trillion for the first time since July 2025.

Against that backdrop, a product launch is also a message. The RTX Spark desktops show that the product pipeline Huang built is still shipping on schedule, and they push Nvidia’s chips into a market it has never really owned: the desk of the individual developer. The company has spent two years selling AI computing by the data centre; now it wants to sell it by the box.

The wider AI race is moving the same way. Google launched Gemini 4 Neon this week, its newest AI model, and every new model raises the same question for buyers: where will it run, and who pays for the computing? Nvidia’s answer is that more of it will run locally, on hardware it sells, and the RTX Spark is the cheapest way it has yet offered to do that at serious scale.

Who Should Buy One — and Who Should Wait

The honest answer is that very few people need one today. A developer fine-tuning open models, a small studio running local AI agents around the clock, or a researcher who cannot put data in the cloud will get real value from a 128GB box. Everyone else, including most gamers and general users, will get far more for their money from a conventional PC.

The wait-and-see case is strong too. Acer, Gigabyte and HP have yet to show their versions, Microsoft’s Surface RTX Spark Dev Box is still to come, and competition between six makers should push prices down from the launch premiums. Stock is thin, the 128GB premium is steep, and the software ecosystem for local agents is still maturing.

What is not in doubt is the direction of travel. Nvidia’s earnings this week showed a company whose data-centre business remains enormous, but the RTX Spark launch shows where it thinks the next growth comes from: not just bigger data centres, but millions of desks. The first boxes are expensive, small and aimed at specialists, and that is exactly how most computing revolutions have started.

Frequently Asked Questions

Q: When did Nvidia RTX Spark mini PCs go on sale?

A: The first Nvidia RTX Spark compact desktops from Asus, Dell and MSI went on sale on Wednesday 18 November 2026, about a month after the RTX Spark laptops launched on 16 October 2026.

Q: How much do Nvidia RTX Spark mini PCs cost?

A: The Asus ProArt GR1X starts at $4,499 for 64GB of memory and 1TB of storage. MSI’s EdgeMesa N AI+ is listed from $4,699 for 64GB, rising to $7,999 for 128GB, while some outlets report entry prices from $4,399.

Q: What is the Nvidia N1X chip?

A: The N1X is the superchip inside RTX Spark machines. It combines a 20-core Grace CPU with a Blackwell RTX GPU carrying 6,144 CUDA cores and up to 128GB of shared LPDDR5X memory, rated at up to one petaflop of AI performance.

Q: Can RTX Spark mini PCs really run a 120-billion-parameter model?

A: Nvidia says the 128GB configurations can run models with up to 120 billion parameters entirely on-device. The 64GB models run smaller models comfortably but face compromises with the largest ones.

Q: Which companies make RTX Spark mini PCs?

A: Asus, Dell and MSI are the first three makers, with the Asus ProArt GR1X and MSI EdgeMesa N AI+ Series named at launch. Acer, Gigabyte and HP are expected to follow, along with Microsoft’s Surface RTX Spark Dev Box.

Julian Vance

Julian Vance is a senior technology analyst and engineering correspondent covering the evolution of consumer electronics and industrial tech. His reporting cuts through corporate jargon to deliver data-backed hardware and software evaluations.

Related Articles

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to top button
×

Stay Updated!

Get our latest news and updates first on Google.

Add as a preferred source on Google