ChipsThe story, in brief

NVIDIA Announces DGX Spark 64GB: A 1-PetaFLOP Grace Blackwell Desktop for Local AI Agents, Fine-Tuning, and Inference

1 petaFLOP on your desk. NVIDIA's DGX Spark 64GB lets developers prototype local agents without cloud.

Illustration of independent geometric mechanisms passing paper tasks along branching amber tracks.
AI agents and the coordination of work.AI illustration by KeyNews
The KeyNews take

Why it matters

NVIDIA is expanding its Grace Blackwell desktop AI system into a 64GB configuration available from multiple OEMs, targeting developers who want to run and fine-tune local models and agents before clustering. This lowers the barrier to on-prem agentic AI development and shifts some workloads away from cloud inference.

The key facts

7 to know
  1. DGX Spark 64GB configuration announced

  2. Grace Blackwell (GB10) GPU-based desktop system

  3. 1 petaFLOP performance stated

  4. Available from Acer, ASUS, Dell, Gigabyte, HP, MSI

  5. Supports clustering two 64GB units to 128GB memory

  6. Positioned for local model inference, fine-tuning, and agent workloads

  7. Published October 2, 2026

The story so far

Earlier coverage of this storyline

  1. NVIDIA DGX Spark 64GB Gives Developers More Ways to Build and Scale Local AINVIDIA Blog
  2. This story

Go to the source

MarkTechPostmarktechpost.com

Publisher excerpt: NVIDIA announced a new 64GB configuration of DGX Spark — from Acer, ASUS, Dell, Gigabyte, HP and MSI — its GB10-powered desktop AI system. It gives developers a way to start with one system for local models and agents, then cluster two 64GB units for 128GB of memory across the cluster and more…
Read original report
Back to today's editionMore chips news

Keep reading

Related stories

More from Chips