From The Register, October 13:
Nvidia's GB10 workstations arrive with 1 petaFLOPS of compute, 128GB of VRAM, and a $3K+ price tag
Systems from Nvidia, Dell, and others available starting Oct. 15
Nvidia's tiniest Grace-Blackwell workstation is finally making its way to store shelves this week, the better part of a year after the GPU giant first teased the AI mini PC, then called Project Digits, at CES.
Since rebranded as the DGX Spark, the roughly NUC-sized system pairs a Blackwell GPU capable of delivering up to a petaFLOP of sparse FP4 performance with 128 GB of unified system memory and 200 Gbps of high-speed networking.
But with a price starting around $3,000, small doesn't mean cheap. Then again, it's not exactly aimed at mainstream PC buyers. The systems, which will also be available under various brand names from OEM partners, won't even come with Windows. A Copilot+ PC this is not. Instead, it ships with a custom spin of Ubuntu Linux.
Spark is actually intended for AI and robotics developers, data scientists, and machine learning researchers looking for a lower-cost workstation platform that's still capable of running models up to 200 billion parameters in size.
These kinds of workloads are incredibly memory-hungry, which makes running them on consumer graphics processors impractical. High-end workstation cards, like the RTX Pro 6000, can be had with up to 96 GB of speedy GDDR7, but a single card will set you back more than $8,000, and that's before you factor in the rest of the platform cost.
At the time of launch, the DGX Spark is technically Nvidia's highest capacity workstation GPU — at least until its Blackwell Ultra-based DGX Station makes its debut.
Honey, I shrunk the superchip
Powering the DGX Spark is the GB10 system-on-a-chip, which is essentially a miniaturized version of the Grace-Blackwell Superchips that power its flagship NVL72 rack systems.As we explored back at Hot Chips, the GB10 is composed of two compute dies connected at 600 GB/s via Nvidia's proprietary NVLink chip-to-chip interconnect tech. And, in case you're wondering, this same technology will eventually be used to mesh Nvidia's GPUs to Intel's future client CPUs as part of a tie-up between the two chip heavyweights.
The GPU tile is capable of delivering up to a petaFLOP of sparse FP4 or around 31 teraFLOPS at single precision (FP32) — putting it on par with an RTX 5070 in terms of raw performance. Yes, the $550 consumer card does offer more than twice the memory bandwidth, but with just 12 GB of GDDR7, you'll be fairly limited in terms of what models and AI workloads you can run....
....MUCH MORE
And there's the rub.
Granted there is a lot more value for money than a couple earlier Nvidia supercomputers that we got fired up about. Some backlinks from June 2018's "Nvidia debuts cloud server platform to unify AI and HPC (NVDA)":
....Before that they explained how to build your own (remember, this was pre-media-rapture, hence the explanatory tone) supercomputer, relayed in May 2015's:
Nvidia Wants to Be the Brains Of Your Autonomous Car (NVDA)
...Among the fastest processors in the business are the one's originally developed for video games and known as Graphics Processing Units or GPU's. Since Nvidia released their Tesla hardware in 2008 hobbyists (and others) have used GPU's to build personal supercomputers.
Here's NvidiasBuild your Ownpage. [link rotted, here is an Internet Archive capture]
Or have your tech guy build one for you....
NVIDIA really started blurring the lines with their mini-supercomputer for training AI in 2016:
Technology Review on NVIDIA's Pint-Sized Supercomputer (NVDA)
$129,000
In April 2018 the company had a new offering that stunned reviewers into caveman-speak:
UPDATED—NVIDIA Wants to Be the Brains Behind the Surveillance State (NVDA)
The company just rolled out a $399,000 two-petaflop supercomputer that every little totalitarian and his brother is going to lust after to run theirsurveillance-citysmart-city data slurping dreams.
The coming municipal data centers will end up matching the NSA in total storage capacity and NVIDIA wants to be the one sifting through it all. More on this down the road, for now here's the beast.
From Hot Hardware:
NVIDIA Unveils Beastly 2 Petaflop DGX-2 AI Supercomputer With 32GB Tesla V100 And NVSwitch Tech (Updated)...
Ummm, beast fast.
But of course back then the world was fresh and new.
Today's press release does have a nice mention of the earlier machines:
NVIDIA DGX Spark Arrives for World’s AI Developers
October 13, 2025
News Summary:
- NVIDIA founder and CEO Jensen Huang delivers DGX Spark to Elon Musk at SpaceX.
- This week, NVIDIA and its partners are shipping DGX Spark, the world’s smallest AI supercomputer, delivering NVIDIA’s AI stack in a compact desktop form factor.
- Acer, ASUS, Dell Technologies, GIGABYTE, HPI, Lenovo and MSI debut DGX Spark systems, expanding access to powerful AI computing.
- Built on the NVIDIA Grace Blackwell architecture, DGX Spark integrates NVIDIA GPUs, CPUs, networking, CUDA libraries and NVIDIA AI software, accelerating agentic and physical AI development.
NVIDIA today announced it will start shipping NVIDIA DGX Spark™, the world’s smallest AI supercomputer.
AI workloads are quickly outgrowing the memory and software capabilities of the PCs, workstations and laptops millions of developers rely on today — forcing teams to shift work to the cloud or local data centers.
As a new class of computer, DGX Spark delivers a petaflop of AI performance and 128GB of unified memory in a compact desktop form factor, giving developers the power to run inference on AI models with up to 200 billion parameters and fine-tune models of up to 70 billion parameters locally. In addition, DGX Spark lets developers create AI agents and run advanced software stacks locally.
“In 2016, we built DGX-1 to give AI researchers their own supercomputer. I hand-delivered the first system to Elon at a small startup called OpenAI — and from it came ChatGPT, kickstarting the AI revolution,” said Jensen Huang, founder and CEO of NVIDIA. “DGX-1 launched the era of AI supercomputers and unlocked the scaling laws that drive modern AI. With DGX Spark, we return to that mission — placing an AI computer in the hands of every developer to ignite the next wave of breakthroughs.”
DGX Spark brings together the full NVIDIA AI platform — including GPUs, CPUs, networking, CUDA® libraries and the NVIDIA AI software stack — into a system small enough for a lab or an office, yet powerful enough to accelerate agentic and physical AI development. By combining breakthrough performance with the reach of the NVIDIA ecosystem, DGX Spark transforms the desktop into an AI development platform.
....MUCH MORE
