NVIDIA DGX Spark – Personal AI Supercomputer for Developers & AI Workloads

Experience next-generation AI computing at your desk with the NVIDIA DGX Spark, a compact yet powerful AI supercomputer built on the advanced Grace Blackwell architecture. Designed for developers, researchers, and AI startups, DGX Spark delivers up to 1 PFLOP of AI performance, enabling seamless training, fine-tuning, and inference of large-scale models.

With 128GB of high-bandwidth unified memory shared between CPU and GPU, DGX Spark eliminates traditional bottlenecks and allows you to run and optimize AI models up to 200 billion parameters locally. Whether you're building generative AI applications, LLM pipelines, computer vision systems, or edge AI solutions, DGX Spark provides enterprise-grade performance in a compact desktop form factor.

Powered by a 20-core ARM CPU and next-gen Blackwell GPU with 5th Gen Tensor Cores, it delivers unmatched efficiency and scalability. Advanced connectivity options including 10Gb Ethernet, Wi-Fi 7, and NVIDIA ConnectX-7 enable high-speed networking and clustering for even larger workloads.

Key Features:

• NVIDIA Grace Blackwell Superchip (CPU + GPU integrated)
• Up to 1 PFLOP AI performance (FP4 precision)
• 128GB LPDDR5X unified memory for large AI models
• Supports AI models up to 200B parameters (405B with dual systems)
• 20-core ARM CPU (high-performance + efficiency cores)
• Up to 4TB NVMe SSD storage
• Compact desktop design (ideal for personal AI labs)
• High-speed networking: 10GbE + ConnectX-7 (200Gb/s)

Perfect For:

• AI developers & startups
• LLM fine-tuning & inference
• Generative AI (text, image, video)
• Data science & machine learning
• Edge AI & robotics development