The NVIDIA DGX-2 is a high-end AI supercomputing system designed for large-scale machine learning, deep learning, and data analytics workloads. It features 16 NVIDIA Tesla V100 GPUs connected via NVSwitch technology, delivering up to 2 petaflops of AI performance. This system is equipped with 512 GB of GPU memory, dual 24-core Intel Xeon CPUs, 1.5 TB of DDR4 system memory, and 30.72 TB of SSD storage, all housed in a 10U rackmount chassis designed to handle massive datasets and complex AI models efficiently.
Regarding its price, the NVIDIA DGX-2 is positioned at the premium end of the AI hardware market due to its unmatched computational power and enterprise-grade features. As of the latest market estimates, a new DGX-2 system typically costs between $300,000 and $400,000 USD. This price range depends on the specific configuration, vendor, and regional availability. The cost reflects not only the hardware but also the integrated AI software stack, optimized deep learning frameworks, and enterprise-level support offered by NVIDIA.
The DGX-2's price is significantly higher than more recent NVIDIA AI systems like the DGX A100, which ranges between $149,000 and $199,000 for an 8-GPU setup with A100 80GB GPUs in 2025. The DGX-2's 16 V100 GPUs and massive memory capacity justify its premium price, especially for organizations tackling the most demanding AI projects requiring extensive parallel processing and memory bandwidth.
In summary, the NVIDIA DGX-2 price reflects its status as a powerful, enterprise-grade AI platform capable of accelerating breakthroughs in AI research and development. Its cost is justified by the combination of cutting-edge GPU technology, large memory pools, high-speed NVSwitch interconnects, and comprehensive software and support services, making it a strategic investment for institutions and enterprises focused on advanced AI workloads.