As a new class of computer, DGX Spark delivers a petaflop of AI performance and 128GB of unified memory

The system delivers a petaflop of AI performance and 128GB of unified memory in a compact form. It enables developers to run inference on models with up to 200 billion parameters and fine-tune models of up to 70 billion parameters locally. Teams can also develop AI agents and run complex AI software stacks on-site.

DGX Spark integrates GPUs, CPUs, networking, CUDA libraries, and the full NVIDIA AI software stack into a system small enough for a lab or office but powerful enough to accelerate advanced AI development. Developers can start immediately, as the NVIDIA AI software stack is preinstalled. It also supports ecosystem tools such as models, libraries, and NVIDIA NIM microservices for local AI workflows, including customizing image generation models, building vision-language agents, or running optimized AI chatbots.

1.NVIDIA founder and CEO Jensen Huang delivers DGX Spark to Elon Musk at SpaceX.

2.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.

3.Acer, ASUS, Dell Technologies, GIGABYTE, HPI, Lenovo and MSI debut DGX Spark systems, expanding access to powerful AI computing.

4.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.

Create, edit, connect, and play like never before with devices for every budget. Let your Intel® Core™ Ultra processor-based AI PC help you improve and accelerate everyday tasks, such as:

Draft emails or organize your calendar.

Remix a song with a few clicks and text entries.

Generate images or music from text.

Enhance photos and videos in a few clicks.

Write, code, and create faster with AI assistants.

Upscale your games for stunning visuals without sacrificing power or battery.

Compared to AI you use in your web browser, AI PCs powered by Intel® Core™ Ultra processors are able to run AI directly on your PC, making them fast, reliable, and highly efficient. Plus, Intel® Core™ Ultra processors offer uncompromising AI software compatibility, supporting 99 percent of AI features tested and delivering the most capable AI PC.2 3

1. Takeaway One: The “AI Performance” Gap Is a Chasm

The most common metric used to measure a chip’s AI capability is TOPS, or Trillions of Operations Per Second. In simple terms, it’s a raw measure of how many calculations a processor can perform for AI-specific tasks. While the “AI PC” label is applied liberally, the actual TOPS performance reveals a vast and growing chasm between different classes of devices.

• The Hobbyist Board: At one end of the spectrum is the Orange Pi AIpro, a single-board computer for developers and makers. It’s a fantastic tool for learning and prototyping, delivering 8 TOPS of AI computing power.

The NPU: Understanding Dedicated AI Hardware

Unlike traditional laptops or desktop PCs, AI PCs have additional silicon for AI processing, usually built directly onto the processor die. On AMD, Intel, and Qualcomm systems, this is generically called the neural processing unit, or NPU. Apple has similar hardware capabilities built into its M-series chips with its Neural Engine.

In all cases, the NPU is built on a highly parallelized and optimized processing architecture designed to crunch many more algorithmic tasks simultaneously than standard CPU cores can. The regular processor cores still handle routine jobs on your machine—say, your everyday browsing and word processing. The differently structured NPU, meanwhile, can free up the CPU and the graphics-acceleration silicon to do their day jobs while it handles the AI stuff.

 By - Aaradhay Sharma