Hewlett Packard Enterprise (HPE) announced a major expansion of its high-performance computing (HPC) and artificial intelligence (AI) infrastructure portfolio, unveiling solutions designed to meet the increasing demands of scientific research and sovereign AI initiatives. The lineup includes new HPE Cray Supercomputing EX systems and purpose-built AI servers optimized for large language model (LLM) training, natural language processing (NLP), and multi-modal workloads. HPE’s innovations feature a direct liquid cooling architecture, advanced compute nodes, and state-of-the-art accelerators from AMD, NVIDIA, and Intel.
The new offerings include the HPE Cray Supercomputing EX4252 Gen 2 Compute Blade, delivering unprecedented CPU density with up to 98,304 cores per cabinet, and the EX154n Accelerator Blade, accommodating up to 224 NVIDIA Blackwell GPUs. These solutions enable researchers and governments to accelerate discoveries and scale AI training with high energy efficiency. HPE also introduced the ProLiant Compute XD server family, designed for service providers and enterprises building their own AI models. This expansion highlights HPE’s commitment to supporting the intersection of traditional supercomputing and AI-driven innovation.
• HPE Cray Supercomputing EX4252 Gen 2 Compute Blade: Equipped with eight 5th Gen AMD EPYC™ processors, it supports up to 98,304 cores per cabinet, offering unmatched compute density for space-constrained environments. Available Spring 2025.
• HPE Cray Supercomputing EX154n Accelerator Blade: Houses 224 NVIDIA Blackwell GPUs in a single cabinet and integrates them with NVIDIA Grace CPUs via NVLink-C2C for ultra-fast communication. Availability expected by late 2025.
• HPE Slingshot Interconnect 400: HPE’s next-generation networking solution provides 400 Gbps speeds with automated congestion management and low tail latency, enabling more efficient workload processing. Available Fall 2025.
• HPE Cray Supercomputing Storage Systems E2000: Offers over double the I/O performance of its predecessor, reducing idle time and improving data flow during compute operations. Available early 2025.
• HPE ProLiant Compute XD680 Server: Features Intel Gaudi 3 AI accelerators in an air-cooled configuration optimized for cost-effective AI training and inferencing. Launches December 2024.
• HPE ProLiant Compute XD685 Server: Supports NVIDIA GPUs or AMD accelerators with liquid cooling for energy-efficient performance in AI workloads. Available early 2025.
HPE’s new portfolio aims to streamline AI deployments with solutions tailored for performance and efficiency. The HPE Cray Supercomputing User Services Software, available now, complements these systems by improving workload management, power efficiency, and overall system utilization.
“Service providers and nations investing in sovereign AI initiatives are increasingly turning to high-performance computing as the critical backbone enabling large-scale AI training that accelerates discovery and innovation,” said Trish Damkroger, Senior VP of HPC & AI Infrastructure Solutions at HPE. “Our customers turn to us to fast-track their AI system deployment to realize value faster and more efficiently by leveraging our HPC solutions and decades of experience.”
HPE will showcase these solutions at SC24 in Atlanta, November 17-22, offering attendees a closer look at the next generation of HPC and AI technologies.
