We take cryptocurrency — so the coins you hold can buy real hardware.Pay in crypto, spend it on real hardware. Free insured shipping worldwide over $399. Plain, unbranded boxes, sent from Germany.Free insured shipping over $399. Your rate is locked for 30 minutes once checkout opens.Rate locked 30 minutes. Every item sealed, serial-checked and covered by a 24-month warranty.Sealed, 24-month warranty.

Saved items Sign in
Compact desktop AI system, matte dark finish, front port array

NVIDIA

NVIDIA 940-54242-0000-100 DGX Spark AI edge system 128GB LPDDR5x 4TB NVMe SSD WiFi 7

SKU CH-87ZF-8KZXB MPN 940-54242-0000-100

A compact AI edge system with 128GB LPDDR5x memory, 4TB NVMe SSD and WiFi 7 for local model development

  • Up to 1 petaFLOP of FP4 AI performance for local fine-tuning and inference
  • 128 GB unified LPDDR5x memory enables models up to 200 billion parameters at FP4
  • Compact Grace Blackwell superchip design brings enterprise-scale AI to the desktop
  • Preinstalled NVIDIA AI software stack lets you develop locally and deploy anywhere
  • 4 TB NVMe SSD and WiFi 7 provide fast storage and connectivity for AI workflows

DGX Spark AI System with 128GB Unified Memory

The NVIDIA DGX Spark is a desktop AI supercomputer powered by the GB10 Grace Blackwell Superchip. It delivers up to one petaFLOP of FP4 AI performance in a compact, energy-efficient form factor. The system includes 128GB of LPDDR5x unified memory and a 4TB NVMe SSD. It ships with a preinstalled NVIDIA AI software stack for immediate development.

Suited for local AI prototyping

This system targets developers and researchers who need to prototype, fine-tune, and run large reasoning models locally. It supports models up to 200 billion parameters at FP4 precision. It is not suited for workloads requiring more than 128GB of unified memory per unit. Buyers needing multi-node cluster scale or higher parameter counts should look at larger infrastructure.

128GB memory and 4TB storage capacity

The 128GB LPDDR5x unified memory provides headroom for large model inference and fine-tuning tasks directly on the desktop. The 4TB NVMe SSD offers substantial local storage for datasets and model checkpoints. Memory capacity limits the maximum model size to 200 billion parameters at FP4. Storage headroom depends on dataset size and versioning needs.

Highlights

  • Up to 1 petaFLOP of FP4 AI performance for local fine-tuning and inference
  • 128 GB unified LPDDR5x memory enables models up to 200 billion parameters at FP4
  • Compact Grace Blackwell superchip design brings enterprise-scale AI to the desktop
  • Preinstalled NVIDIA AI software stack lets you develop locally and deploy anywhere
  • 4 TB NVMe SSD and WiFi 7 provide fast storage and connectivity for AI workflows

Specifications

No published specifications for this item yet.

Questions about this item

What power supply is required for the DGX Spark to operate?

The system ships with a connecting cable; plug it into a standard AC outlet and the GB10 Grace Blackwell Superchip powers on without any external PSU.

Will the DGX Spark share memory with an existing GPU in my workstation?

The 128 GB of LPDDR5x unified memory is local to the Superchip and is not shared with any other graphics card in the host machine.

How can I verify the system is delivering the advertised AI performance after setup?

Run an NVIDIA AI software stack benchmark; the dashboard reports up to 1 petaFLOP of FP4 throughput and confirms 200 billion parameter model capacity.

Also in this aisle

People compared these

Same shelf, same checkout — eight coins and a 30-minute rate lock.