NVIDIA DGX Spark – A Personal AI Supercomputer for Local AI

The NVIDIA DGX Spark is a compact AI computing system designed to bring powerful Grace Blackwell AI acceleration directly to the desktop. Powered by the NVIDIA GB10 Grace Blackwell Superchip, it combines a 20-core Arm CPU, Blackwell GPU architecture, 128GB of coherent unified memory, high-speed networking and the NVIDIA AI software stack into a purpose-built platform for modern AI development.

Unlike a conventional desktop workstation, DGX Spark is engineered specifically for AI prototyping, inference, fine-tuning, generative AI, data science, autonomous agents, robotics and computer vision. NVIDIA positions the platform to run supported AI models with up to 200 billion parameters locally, while multiple DGX Spark systems can be connected for larger workloads.

Its compact 150 × 150 × 50.5mm chassis and approximately 1.2kg weight make it dramatically smaller than traditional AI workstations and servers, while still providing a complete NVIDIA AI development environment.

Product Specifications

Top Features

🚀 NVIDIA GB10 Grace Blackwell Superchip

Advanced Grace Blackwell architecture engineered specifically for modern AI workloads.

🧠 128GB Coherent Unified Memory

Large 128GB LPDDR5x unified memory pool designed for memory-intensive AI models.

⚡ Up to 1 PFLOP FP4 AI Performance

High-throughput AI acceleration for supported FP4 workloads with sparsity.

🎯 6,144 CUDA Cores

Massive parallel processing capability for CUDA-accelerated applications.

💾  4TB NVMe SSD

Large local storage capacity for models, datasets, containers and development projects.

🌐 ConnectX-7 at 200Gbps

Professional high-speed networking for demanding AI workloads and multi-DGX Spark configurations.

📡 Wi-Fi 7

Next-generation wireless connectivity for compatible networks.

🖥️ Multi-Display Support

HDMI 2.1a plus up to three DisplayPort connections through USB-C.

🔌 4× USB-C

High-speed USB-C connectivity for peripherals, displays and supported devices.

🛡️ Self-Encrypting NVMe

Supported storage configuration provides self-encryption for additional data protection.

📦 Ultra-Compact Design

Approximately 150 × 150 × 50.5mm and 1.2kg.

🔋 240W Power Supply

NVIDIA specifies a 240W external power supply for DGX Spark.

🛡️ 1 Year Warranty

This listing includes a 1 Year Warranty, providing additional confidence for professional AI development and business use.

Why Buy NVIDIA DGX Spark?

1. Purpose-Built for AI

DGX Spark isn’t simply a desktop PC with a powerful graphics card. It is designed as an integrated AI platform combining NVIDIA compute, unified memory, networking and AI software.

2. Run AI Models Locally

The combination of 128GB unified memory and Blackwell acceleration allows developers to work with substantially larger AI models locally than many conventional consumer systems. NVIDIA specifies support for models up to 200 billion parameters for local AI development and inference under its stated configurations.

3. Develop Without Building a Custom AI Server

Instead of sourcing a CPU, GPU, motherboard, memory, cooling system and networking hardware separately, DGX Spark provides a tightly integrated AI computing platform.

4. Compact Desktop Footprint

At approximately 1.2kg, DGX Spark is considerably easier to deploy than conventional rack-mounted AI infrastructure.

5. Designed to Scale

A single DGX Spark can serve as a powerful local AI development machine, while multiple systems can be interconnected for larger model workloads.

6. NVIDIA AI Ecosystem

Access to NVIDIA’s CUDA ecosystem, AI frameworks, libraries, NGC resources and NIM technologies gives developers a broad software foundation for building AI applications.