Revolutionize Computing with 256 NVIDIA Superchips, 18,432 Cores, and 144TB GPU Memory for 1 exaFLOPS Performance. Includes Comprehensive Networking, AI Enterprise, and Three-Year Support. Pricing Available Upon Request.
$410,000
Experience Boundless AI Potential with the NVIDIA DGX™ GH200: A Supercomputer that Connects 256 Grace Hopper™ Superchips to Handle Terabyte-Class Models in Recommender Systems, Generative AI, and Graph Analytics. It Offers 144TB of Shared Memory with Linear Scalability for Massive AI Models.
DOWNLOAD DATASHEET
The DGX GH200 achieves its remarkable scalability by seamlessly extending the power of NVIDIA® NVLink® technology. This groundbreaking approach involves connecting every Grace Hopper Superchip within the system using the sophisticated NVLink Switch System.
This strategic interconnection allows the DGX GH200 to effectively harness the combined capabilities of all 256 GPUs, creating a supercomputing environment that can tackle the most demanding AI and high-performance computing workloads with unparalleled efficiency and speed. The result is a harmonious orchestration of computational power, enabling organizations to push the boundaries of what's possible in the world of AI and data analytics.
The NVIDIA DGX platform is purpose-built for enterprise AI, representing a holistic approach that seamlessly integrates the finest aspects of NVIDIA software, infrastructure, and expertise. This results in a cutting-edge, unified AI development and training solution designed to meet the evolving needs of modern enterprises.