HPE (NYSE: HPE) has introduced the latest expansion of its Cray supercomputing portfolio, bringing forward a new generation of high-density systems built to power large-scale artificial intelligence workloads and complex scientific research. The updated lineup features three advanced compute blades developed with key technology partners, a unified management platform, and next-generation interconnects that together establish one of the most capable supercomputing architectures available to global research institutions, national programs, and major enterprises accelerating their AI-driven innovation.
Trish Damkroger, senior vice president and general manager for HPC and AI Infrastructure at HPE, shared that organizations depending on supercomputers continue to seek stronger, more versatile performance across diverse workloads. She noted that the upgraded Cray platform blends AI and traditional high-performance computing into a single, high-density architecture designed to unlock new discoveries. According to her, the launch demonstrates HPE’s continuing commitment to pushing the boundaries of innovation and enabling scientific progress.
The expansion follows the recent debut of the Cray GX5000 system, engineered to support major breakthroughs at the intersection of AI and simulation. Complementing the compute platform is the new Cray Storage Systems K3000, the first production-ready storage system built with integrated DAOS technology to accelerate demanding data-intensive applications.
Momentum for the refreshed Cray platform is growing rapidly. The High-Performance Computing Center at the University of Stuttgart (HLRS) and the Leibniz Supercomputing Centre (LRZ) have selected the GX5000 as the foundation for their next national supercomputers, named Herder and Blue Lion. Leaders from both institutions emphasized that the new systems will significantly raise performance capabilities, support combined AI and simulation workflows, and advance ongoing efforts to improve energy efficiency. Their adoption further demonstrates the platform’s readiness to support world-class research.
HPE has also introduced three new liquid-cooled compute blades designed for exceptionally dense performance: the GX440n accelerated blade powered by NVIDIA Vera CPUs and Rubin GPUs, the GX350a accelerated blade featuring next-generation AMD EPYC “Venice” processors paired with AMD Instinct MI430X GPUs, and the GX250 compute blade built around eight next-generation AMD EPYC CPUs for double-precision scientific workloads. Each blade supports high-bandwidth networking through Slingshot 400 endpoints and optional onboard NVMe storage, and all can be combined within a single GX5000 rack to tailor configurations to specific workload requirements.
To bring unified control across these environments, HPE has introduced an upgraded system management platform that supports multi-tenant and containerized deployments, along with full power and energy oversight to help organizations manage consumption and operational costs. Strengthened security and improved governance features create a consistent, secure management experience throughout the system’s lifecycle.
Networking performance also advances with the availability of Slingshot 400 for the GX5000 platform. The redesigned, liquid-cooled switch blade offers 64 high-speed 400 Gbps ports and can be deployed in configurations of up to 2,048 total ports, enabling low-latency connectivity capable of supporting the heavy demands of converged AI and high-performance computing.
The Cray Storage Systems K3000 further enhances the ecosystem with strong compute performance, generous memory capacity, and rapid data transfer capabilities. With its DAOS-based architecture, it is tuned for input/output-heavy AI workloads and can be configured with a wide selection of NVMe drive counts, capacities, and memory tiers, along with support for Slingshot, InfiniBand NDR, and 400 GbE connectivity.
HPE backs these technologies with a comprehensive services portfolio that spans performance tuning, global deployment, and round-the-clock operational support. With decades of experience delivering large-scale, energy-efficient computing systems, HPE continues to help organizations push scientific and technological boundaries.
The GX440n, GX250, and GX350a blades, along with the updated management software and Slingshot 400 for GX5000 systems, will be available in early 2027. The Cray Storage Systems K3000 will be available beginning in early 2026.
Industry partners emphasized the significance of the launch. AMD’s Travis Karr highlighted the combined strength of EPYC CPUs and Instinct GPUs in solving large-scale scientific and AI challenges. NVIDIA’s Dion Harris noted that systems built on the Rubin architecture will accelerate analytics, simulation, and AI across disciplines. Hyperion Research CEO Dr. Earl Joseph added that the new systems will enable major advancements in product development, scientific discovery, and societal impact.
HPE (NYSE: HPE) is a global technology leader delivering solutions across AI, cloud, and networking to help organizations operate more intelligently, convert data into actionable insights, and maximize impact. The company’s long-standing expertise empowers customers worldwide to pursue bold goals and redefine what’s possible.
