Hewlett Packard Enterprise and NVIDIA Join Forces to Supercharge AI Training

Hewlett Packard Enterprise (HPE) has unveiled a cutting-edge supercomputing solution, developed in collaboration with NVIDIA, specifically designed for generative AI applications. This turnkey solution targets large enterprises, research institutions, and government organizations aiming to enhance the training and tuning of artificial intelligence (AI) models using proprietary datasets.

The comprehensive solution encompasses a powerful software suite enabling the training and customization of models, fostering the development of AI applications. This innovation includes liquid-cooled supercomputers, accelerated compute capabilities, advanced networking, storage solutions, and a suite of services to expedite organizations’ journey to unlock the true value of AI.

Justin Hotard, EVP, and GM of HPC, AI & Labs at HPE, expressed the necessity of purpose-built solutions for effective generative AI support. The collaboration with NVIDIA aims to provide a turnkey AI-native solution that enables customers to significantly accelerate AI model training and enhance outcomes.

Key components of this supercomputing solution include software tools for building AI applications, customizing pre-built models, and coding modifications. Integrated with HPE Cray supercomputing technology and powered by NVIDIA Grace Hopper GH200 Superchips, the solution offers unparalleled scale and performance for substantial AI workloads.

Generative AI is evolving various industrial and scientific endeavors, and this collaboration is set to provide the performance required for groundbreaking initiatives. Ian Buck, VP of Hyperscale and HPC at NVIDIA, highlighted that the collaboration’s turnkey solution, driven by NVIDIA GH200 Grace Hopper Superchips, will empower customers to achieve breakthroughs in their generative AI initiatives.

The purpose-built supercomputing solution includes AI/ML acceleration software, designed for scale based on the HPE Cray EX2500, a network for real-time AI with HPE Slingshot Interconnect, and turnkey simplicity complemented by HPE Complete Care Services.

What sets this solution apart is its focus on sustainability. By 2028, the growth of AI workloads is anticipated to require about 20 gigawatts of power within data centers. HPE addresses this concern by delivering energy-efficient solutions with liquid-cooling capabilities, resulting in a substantial performance improvement and reduced power consumption.