Hewlett Packard Enterprise (NYSE: HPE) expanded its HPE Cray supercomputing portfolio, introducing new compute blades, storage solutions, and interconnect technologies designed to deliver industry-leading compute density for artificial intelligence (AI) and high-performance computing (HPC) workloads. The next-generation portfolio is aimed at research laboratories, sovereign entities, and large enterprises seeking high-performance, scalable computing for simulations, AI model training, and scientific discovery.
The latest expansion builds on HPE’s recent launch of the Cray GX5000 platform, engineered for converged AI and HPC environments. The GX5000 combines multi-partner, multi-workload compute blades, high-performance interconnects, and unified management software, enabling one of the most powerful supercomputing architectures currently available.
“Global organizations relying on supercomputing are looking for better computing performance for all of their workloads,” said Trish Damkroger, senior vice president and general manager, HPC and AI Infrastructure Solutions at HPE. “Our new HPE Cray Supercomputing platform is the answer to customers’ needs for higher performance density with a unified AI and HPC architecture that is engineered for groundbreaking outcomes.”
The portfolio expansion also includes the HPE Cray Supercomputing Storage Systems K3000, the industry’s first factory-built storage system with embedded Distributed Asynchronous Object Storage (DAOS) open-source software for increased performance, enabling faster input/output for AI workloads.
Leading European supercomputing centers have already adopted the GX5000 platform. The High-Performance Computing Center of the University of Stuttgart (HLRS) and the Leibniz Supercomputing Centre (LRZ) of the Bavarian Academy of Sciences and Humanities have selected the platform for their next-generation systems, Herder and Blue Lion, respectively.
“As the builder of HLRS’s Hawk and Hunter supercomputers, HPE has for years been an excellent partner,” said Prof. Dr.-Ing. Michael Resch, director at HLRS. “The GX5000 platform will offer our users a major leap in performance for simulation and AI, as well as improved energy efficiency.”
LRZ Chairman Prof. Dr. Dieter Kranzlmüller highlighted the platform’s direct liquid cooling capabilities, which allow operation at up to 40°C and enable waste heat reuse across the research campus. He said the system’s performance, up to 30 times faster than current systems, will allow seamless integration of traditional modeling and AI-driven workflows.
The portfolio includes three new liquid-cooled compute blades. The GX440n Accelerated Blade supports NVIDIA Vera CPUs and Rubin GPUs for mixed-precision computing, delivering up to 192 GPUs per rack. The GX350a Accelerated Blade combines AMD EPYC “Venice” CPUs with AMD Instinct MI430X GPUs, offering up to 112 GPUs per rack. The GX250 Compute Blade provides CPU-only partitions with eight EPYC CPUs per blade, maximizing x86 core density.
All blades feature HPE Slingshot 400 Gb/s interconnect endpoints and optional NVMe SSDs, enabling customers to optimize configurations for AI, simulation, or mixed workloads. The platform also offers unified supercomputing management software, supporting multi-tenant, virtualized, and containerized environments, while providing system-wide monitoring of power, cooling, and security.
The HPE Slingshot 400 interconnect, now available for GX5000 clusters, features direct liquid-cooled switch blades with up to 2,048 ports per configuration, reducing latency and improving sustained bandwidth for large-scale AI and HPC workloads.
HPE’s DAOS-based K3000 storage system, built on ProLiant DL360 Gen12 servers, supports performance-optimized configurations with up to 16 NVMe drives or capacity-optimized setups with 20 NVMe drives, enabling high-throughput, low-latency storage for AI applications.
Industry partners emphasized the significance of the new platform. Travis Karr, corporate vice president at AMD, noted that HPE’s collaboration enables energy-efficient, high-density compute solutions combining EPYC CPUs and Instinct GPUs. Dion Harris, senior director at NVIDIA, said the GX5000 systems will accelerate simulations, analytics, and AI industrial adoption.
Dr. Earl Joseph, CEO at Hyperion Research, added, “HPE’s next generation of GX5000 supercomputers will help researchers and companies dramatically develop better products and scientific discoveries, while addressing critical societal goals.”
The new HPE Cray compute blades and management software are expected to be available in early 2027, while the K3000 storage system is slated for release in early 2026.
