HPE Unveils Next-Generation AI Factory and Supercomputing Advancements with NVIDIA

The integration of AI into traditional high-performance computing (HPC) is accelerating, with research laboratories, sovereign entities, and large enterprises increasingly adopting these technologies. HPE's latest offerings include the HPE Cray Supercomputing GX5000, which unifies AI and HPC, providing organizations with robust tools to expedite scientific discovery.

HPE has introduced one of the first NVIDIA Vera CPU compute blades, featuring up to 16 NVIDIA Vera CPUs per blade. The HPE Cray Supercomputing GX240 Compute Blade is designed to support demanding AI workloads and boasts impressive scalability, accommodating up to 40 blades, which translates to 640 NVIDIA Vera CPUs and 56,320 NVIDIA Olympus Arm-compatible cores per rack.

To complement its supercomputing capabilities, HPE is now offering enhanced networking options, including the NVIDIA Quantum-X800 InfiniBand. This networking solution provides 144 ports of 800 Gb/s connectivity per port, ensuring power-efficient performance ideal for large-scale systems.

Trish Damkroger, Senior Vice President and General Manager of HPC & AI Infrastructure Solutions at HPE, commented, “Having built the three most powerful, exascale supercomputers in the world, HPE is at the forefront of innovation that brings together cutting-edge AI workloads with traditional HPC to accelerate scientific breakthroughs.”

HPE is also enhancing its AI Factory portfolio, which is co-engineered with NVIDIA, to cater to service providers, sovereign entities, and large enterprises.

The introduction of the NVIDIA Vera Rubin NVL72 by HPE marks a significant advancement in AI deployments for neo-cloud environments. This flagship AI system is engineered for frontier-scale models exceeding 1 trillion parameters and includes 36 NVIDIA Vera CPUs and 72 NVIDIA Rubin GPUs.

The new HPE Compute XD700 server, inspired by the Open Compute Project, is designed to maximize GPU density while minimizing space, power, and cooling costs. Each rack can support up to 128 Rubin GPUs, effectively doubling the GPU density compared to the previous generation.

The NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs are now available across all of HPE’s AI factories, providing customers with enhanced options for AI model training and inference.

HPE's updates extend beyond hardware, introducing software and services that facilitate faster AI deployment for large-scale projects.

HPE’s AI Factory portfolio is now endorsed by the NVIDIA Cloud Provider certification, streamlining the validation process for cloud service providers.

The enhanced AI Factory portfolio supports multi-tenancy models for virtual machines with GPU passthrough, enabling service providers to tailor their offerings through NVIDIA Multi-Instance GPU (MIG).

HPE's AI Factory portfolio now includes support for Red Hat Enterprise Linux and OpenShift, integrating seamlessly with NVIDIA AI Enterprise solutions.

HPE is set to launch its AI Factory systems with support for NVIDIA Mission Control software, which will streamline workload orchestration and enhance operational efficiency.

HPE has provided a timeline for the availability of its new systems:

HPE's latest press release illustrates the company's unwavering dedication to pushing the boundaries of AI and supercomputing. By leveraging NVIDIA's cutting-edge technology, HPE aims to provide its customers with the tools needed to accelerate innovation across various sectors, from healthcare to engineering. As enterprises and nations prepare for a future dominated by data and AI, HPE's advancements are poised to play a pivotal role in shaping the landscape of technology.