NVIDIA DGX Spark Blackwell 4TB AI Supercomputer
NVIDIA DGX Spark 4TB – Compact Grace Blackwell Personal AI Supercomputer Designed for Local LLM Development, Generative AI, Model Fine-Tuning, Inference, Agentic AI, Data Science and Advanced Research Workflows
- Model
- DGX Spark 4TB
- MPN
- 940-54242-0008-000
- UPC
- 810152850398
Prices change daily - ask for today's price.
Key features
- NVIDIA GB10 Grace Blackwell Superchip
- 20-core Arm CPU with 10 Cortex-X925 and 10 Cortex-A725 cores
- NVIDIA Blackwell GPU with 5th Generation Tensor Cores and 4th Generation RT Cores
- Up to 1 PFLOP of FP4 AI performance with sparsity
- 128GB LPDDR5x coherent unified system memory
- 273GB/s memory bandwidth with 256-bit memory interface
- 4TB NVMe M.2 SSD with self-encryption
- Run inference and testing with AI models up to 200 billion parameters
- Fine-tune AI models up to 70 billion parameters locally
- NVIDIA ConnectX-7 NIC at up to 200Gbps
- 10GbE LAN | WiFi 7 | Bluetooth 5.4
- 4 x USB Type-C | HDMI 2.1a | Up to 3 x DisplayPort through USB-C
- NVIDIA DGX OS with the NVIDIA AI software stack preinstalled
- 240W external power supply included
- Compact dimensions: 150 x 150 x 50.5mm | Weight: 1.2kg
- 1 Year Warranty
NVIDIA DGX Spark 4TB Personal AI Supercomputer
NVIDIA DGX Spark brings Grace Blackwell AI computing to the desktop in a compact, power-efficient form factor. Built for AI developers, researchers, startups, data scientists and advanced AI labs, it enables local prototyping, fine-tuning, inference, agentic AI development and model testing without depending only on cloud infrastructure.
Desktop AI Supercomputer for Local AI Development
NVIDIA DGX Spark is designed for developers who want to build and run advanced AI models locally. Powered by the NVIDIA GB10 Grace Blackwell Superchip, it delivers up to 1 PFLOP of FP4 AI performance with 128GB of coherent unified system memory and 4TB NVMe M.2 storage. This makes it ideal for model prototyping, local inference, autonomous agents, AI research, data science and edge AI application development.
GB10 Grace Blackwell
Built on NVIDIA Grace Blackwell architecture for next-generation AI development and reasoning workloads.
1 PFLOP FP4 AI Performance
Up to 1 petaFLOP of theoretical FP4 AI performance for accelerated inference and model workflows.
128GB Unified Memory
Large coherent unified memory helps run and test AI models with up to 200 billion parameters locally.
4TB NVMe Storage
High-capacity 4TB NVMe M.2 storage with self-encryption for large models, datasets and AI projects.
Built for Agentic AI, LLMs and Generative AI Workflows
DGX Spark is made for modern AI teams working on autonomous agents, local LLM testing, model evaluation, AI-assisted applications, computer vision, robotics, data science and edge AI development. It includes the NVIDIA AI software stack, helping developers start AI projects faster with optimized NVIDIA tools, libraries, frameworks and pre-trained model support.
AI Prototyping
Create, test and validate AI models and AI applications directly from your desktop setup.
Fine-Tuning
Fine-tune AI models up to 70 billion parameters for domain-specific business and research needs.
Inference
Run inference and validation with AI models up to 200 billion parameters locally.
Edge AI Development
Develop robotics, computer vision, smart city and edge AI applications using NVIDIA frameworks.
Grace Blackwell AI Power in a Compact Desktop Design
DGX Spark brings powerful NVIDIA AI computing into a compact 150mm x 150mm x 50.5mm chassis. It is built for users who need serious AI capability but do not want to depend on a full rack server or cloud GPU instance for every development cycle.
ConnectX-7 Networking for Scalable AI Projects
NVIDIA DGX Spark includes high-performance NVIDIA ConnectX-7 networking with up to 200Gbps capability. Two DGX Spark systems can be connected to work with larger AI models and extended development workloads, making it a serious platform for researchers and AI labs that need a compact but scalable local AI setup.
Ideal Use Cases for NVIDIA DGX Spark 4TB
NVIDIA DGX Spark is not a normal gaming desktop or regular office PC. It is a specialized AI development system built for local model development, inference, research, experimentation and production-oriented AI workflows before moving to data center or cloud infrastructure.
AI Developers
Build local AI agents, test model pipelines, run inference and validate applications before deployment.
Researchers & Labs
Run AI experiments, model evaluation, computer vision and machine learning workloads from a compact desktop system.
Startups & AI Teams
Prototype AI products locally, reduce cloud dependency and accelerate proof-of-concept development cycles.
Data Science
Accelerate data analytics, ML workflows and computationally complex model testing from your workspace.
Robotics & Edge AI
Develop robotics, smart city, vision AI and edge applications using NVIDIA AI frameworks.
Local LLM Workflows
Run, test and validate large language models locally with the benefit of 128GB unified system memory.
Product Gallery
NVIDIA DGX Spark 4TB Specifications
| Product Name | NVIDIA DGX Spark 4TB |
|---|---|
| Product Type | Personal AI Supercomputer / Desktop AI Development System |
| Architecture | NVIDIA Grace Blackwell |
| Superchip | NVIDIA GB10 Grace Blackwell Superchip |
| GPU | Blackwell Architecture |
| CPU | 20-core Arm CPU: 10 Cortex-X925 + 10 Cortex-A725 Arm |
| CUDA Cores | Blackwell Generation |
| Tensor Cores | 5th Generation |
| RT Cores | 4th Generation |
| Tensor Performance | Up to 1 PFLOP FP4 |
| System Memory | 128GB LPDDR5x coherent unified system memory |
| Memory Interface | 256-bit |
| Memory Bandwidth | 273GB/s |
| Storage | 4TB NVMe M.2 with self-encryption |
| USB | 4 x USB Type-C |
| Ethernet | 1 x RJ-45 connector, 10GbE |
| NIC | NVIDIA ConnectX-7 NIC @ 200Gbps |
| Wi-Fi | Wi-Fi 7 |
| Bluetooth | Bluetooth 5.4 |
| Display Output | 1 x HDMI 2.1a, up to 3 x DisplayPort over USB-C with DP Alt Mode |
| Audio Output | HDMI multichannel audio output |
| NVENC / NVDEC | 1 x NVENC / 1 x NVDEC |
| Operating System | NVIDIA DGX OS |
| Power Supply | 240W |
| GB10 TDP | 140W |
| Dimensions | 150mm L x 150mm W x 50.5mm H |
| Weight | 1.2kg |
| Best For | AI development, local LLM inference, fine-tuning, agentic AI, data science, robotics, computer vision and edge AI workflows. |
Buy NVIDIA DGX Spark 4TB from NationalPC
Bring Grace Blackwell AI performance to your desk with NVIDIA DGX Spark 4TB. Ideal for AI developers, researchers, startups, data scientists and advanced labs looking for local AI model development, fine-tuning, inference and agentic AI workflows.
Buy from NationalPC| Manufacturer Warranty | 1 Year |
|---|
| RAM Type | LPDDR5X |
|---|---|
| RAM Capacity | 128GB |
| CPU Cores | 20 Cores |
|---|---|
| CPU Threads | 20 Threads |
| Processor Generation | Grace Blackwell |
| Processor Model | NVIDIA GB10 Grace Blackwell |
| Integrated GPU | NVIDIA Blackwell (Integrated) |
| Storage Type | M.2 NVMe |
|---|---|
| Storage Capacity | 4TB |
| Product Segment | AI Supercomputer |
|---|
No reviews yet. Be the first to review this product.