Hewlett Packard Enterprise (HPE) has taken a significant leap forward in powering AI-intensive and high-performance computing (HPC) environments with the launch of its next-generation HPE Cray Supercomputing portfolio. Designed for research institutions, sovereign AI initiatives, and large enterprises, the platform introduces industry-leading compute density, direct liquid cooling, unified management software, and energy-efficient architecture—positioning HPE at the forefront of global supercomputing innovation.
“As organizations increasingly rely on supercomputing, they demand higher density, better energy efficiency and unified AI-HPC capabilities,” said Trish Damkroger, SVP and GM, HPC and AI Infrastructure Solutions at HPE. “Our enhanced Cray platform is purpose-built to deliver groundbreaking scientific, industrial and AI-driven outcomes.”
“AI workloads are no longer an add-on—they are redefining the architecture of modern supercomputers.”
— Trish Damkroger, HPE
The portfolio includes three new liquid-cooled blades—GX440n, GX350a, and GX250—optimised for GPU- and CPU-intensive workloads. These support flagship processors including NVIDIA Rubin GPUs, AMD Instinct MI430X accelerators and next-gen AMD EPYC “Venice” CPUs, offering unprecedented compute density—up to 192 GPUs or 320 CPU processors per rack.
Elite research institutions are among the early adopters. The High-Performance Computing Center at the University of Stuttgart (HLRS) has selected HPE’s Cray GX5000 platform for its upcoming Herder supercomputer. Prof. Michael Resch, Director, HLRS, said, “The GX5000 platform will deliver a major leap in simulation and AI performance, helping us accelerate scientific breakthroughs in a sustainable way.” Similarly, the Leibniz Supercomputing Centre (LRZ) is building its flagship Blue Lion system on the GX5000 architecture, powered by 100% liquid cooling and AI-optimized workflows. Prof. Dieter Kranzlmüller, Chairman at LRZ, emphasized its energy efficiency and ability to integrate traditional HPC and modern AI seamlessly.
Driving advanced data throughput and efficient system management, HPE also introduced Supercomputing Management Software for secure, multi-tenant operations, power-aware scheduling and full lifecycle governance. The new HPE Slingshot 400 interconnect enables 400 Gbps switching capacity per port, ensuring low-latency performance across large clusters. Meanwhile, the DAOS-based HPE Cray Storage Systems K3000 brings optimized data access for input/output-bound AI workloads.
“HPE’s next-generation AI supercomputers will supercharge scientific discovery,” said Dion Harris, Senior Director, NVIDIA. Travis Karr, Corporate VP, AMD, added, “By integrating high-density AMD CPU and GPU architectures with HPE infrastructure, we are enabling scalable, energy-efficient AI solutions.”
By blending HPC, AI, and sustainability, HPE’s Cray portfolio is shaping the future of AI-native supercomputing, empowering enterprises and research institutions to drive innovation at scale.