HPE Expands Cray supercomputing lineup to power next wave of AI and scientific innovation

Hewlett Packard Enterprise has unveiled new additions to its next-generation HPE Cray supercomputing portfolio, introducing advanced compute blades, unified management software, and high-performance interconnects designed to meet escalating global demands for AI, simulation, and large-scale scientific workloads.

Hewlett Packard Enterprise (HPE) has announced a significant expansion of its next-generation HPE Cray supercomputing portfolio, introducing new AI-ready compute blades, enhanced management software, and next-generation interconnect technology. The additions aim to deliver unprecedented performance density for research institutions, sovereign AI initiatives, and large enterprises embracing high-performance computing (HPC) and artificial intelligence at scale.

Responding to Rising AI and HPC Demands

The expanded Cray lineup is engineered to serve organizations that require more powerful infrastructure to handle increasingly complex computing workloads. Trish Damkroger, Senior Vice President and General Manager of HPC and AI Infrastructure Solutions at HPE, said the new platform blends unified AI and HPC capabilities to drive breakthrough results. She highlighted HPE’s commitment to enabling scientific discovery and innovation through energy-efficient, high-performance systems.

This announcement follows last month’s launch of the HPE Cray Supercomputing GX5000 and the Cray K3000 storage system—the first factory-built solution with DAOS open-source technology to support AI-intensive applications with higher throughput and scalability.

Strong Industry Momentum with European Research Centers

The new Cray systems are already seeing rapid adoption in Europe. The High-Performance Computing Center of the University of Stuttgart (HLRS) and the Leibniz Supercomputing Centre (LRZ) have selected the GX5000 platform for their upcoming systems, Herder and Blue Lion. Both centers expect major advances in AI and simulation performance, while also improving energy efficiency through direct liquid cooling and heat reuse.

Leaders at HLRS and LRZ said the new systems will allow researchers to merge traditional computational science with emerging AI methodologies, unlocking new possibilities in climate research, engineering, molecular modeling, and more.

New Liquid-Cooled Compute Blades for AI and Scientific Workloads

At the heart of HPE’s new portfolio are three liquid-cooled compute blades supporting next-generation CPUs and GPUs from NVIDIA and AMD. These include: 

·         GX440n Accelerated Blade featuring NVIDIA Vera CPUs and Rubin GPUs for mixed-precision AI workloads

·         GX350a Accelerated Blade with AMD EPYC “Venice” CPUs and Instinct MI430X GPUs for sovereign AI and HPC

·         GX250 Compute Blade offering CPU-optimized performance for double-precision scientific modeling

These blades offer high compute density, support HPE Slingshot 400 interconnects, and can be mixed within a single GX5000 rack to create workload-specific configurations.

Unified Management, Slingshot 400, and Enhanced Storage

HPE is also introducing new Supercomputing Management Software designed for multi-tenant operations, containerized workflows, and improved energy optimization. Additionally, the Slingshot 400 interconnect—optimized for large-scale AI—delivers lower latency and higher bandwidth for demanding workloads.

The DAOS-based Cray K3000 storage system will be available in 2026, offering high-speed NVMe-based data throughput tailored for I/O-heavy AI applications.

HPE plans to make the new compute blades, management software, and interconnect technologies available starting in early 2027, reinforcing the company’s position as a global leader in AI-driven supercomputing solutions.