Hewlett Packard Enterprise has shipped its first NVIDIA GB200 NVL72, a rack-scale computing system intended for large AI clusters. This solution features direct liquid cooling to enhance efficiency and performance, catering to the needs of service providers and enterprises facing demands for rapid deployment.

Trish Damkroger, senior vice president at HPE, noted the growing pressure for scalability and performance in AI. She highlighted that HPE's expertise, including their placement of several systems on the Green500 list for energy efficiency, positions them to meet these challenges effectively.

The NVIDIA GB200 NVL72 incorporates shared-memory architecture with advanced GPU technology, supporting extremely large AI models. Bob Pette, vice president at NVIDIA, emphasized that this collaboration aims to assist enterprises in efficiently establishing large AI clusters.

HPE's extensive experience in liquid cooling spans five decades, which supports their ability to deliver systems designed for high-density data centers. Features of the GB200 NVL72 include 72 NVIDIA Blackwell GPUs and up to 13.5 TB total memory. HPE also offers tailored support services to aid customers through on-site engineering and performance engagement.

The NVIDIA GB200 NVL72 is part of a broader portfolio of high-performance computing (HPC) and supercomputing systems aimed at meeting various compute-intensive demands, including scientific research and generative AI.