Hewlett Packard Enterprise (HPE) has announced an exciting expansion of its high performance computing (HPC) and artificial intelligence (AI) infrastructure portfolio. This move includes the launch of HPE Cray Supercomputing EX solutions, along with two systems tailored for optimizing large language model training, natural language processing, and multi-modal model training. These innovative solutions aim to assist global customers in accelerating scientific research and innovation.
Trish Damkroger, HPE’s Senior Vice President and General Manager of HPC & AI Infrastructure Solutions, emphasized the increasing reliance on high-performance computing within the AI sector. As she noted, “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.” This underscores the importance of high-performance computing for rapid AI system deployment and operational efficiency.
The HPE Cray Supercomputing EX series is particularly noteworthy. It features a unique 100% fanless direct liquid cooling system architecture, providing diverse offerings aimed at research institutions and government bodies. The Cray Supercomputing EX4252 Gen 2 Compute Blade, which can deliver up to 98,304 cores in a single cabinet, is expected by Spring 2025. In parallel, the EX154n Accelerator Blade, supporting up to 224 NVIDIA Blackwell GPUs, promises to enhance supercomputing workload efficiency by the end of 2025.
Further bolstering the new offerings, the HPE Slingshot interconnect 400 delivers network interface controllers, cables, and switches with 400 gigabit-per-second speeds, doubling its predecessor’s speed. These components will be available for HPE Cray Supercomputing EX systems by Fall 2025. Completing the portfolio are the HPE Cray Supercomputing Storage Systems E2000 and the HPE Cray Supercomputing User Services Software.
Addressing AI model training and tuning, HPE introduced the HPE ProLiant Compute XD server family. This innovation focuses on simplifying AI cluster deployment for service providers and large enterprises. Among these, the ProLiant Compute XD680, an air-cooled server housing Intel Gaudi 3 AI accelerators, will launch in December 2024. Meanwhile, the XD685 server, integrating NVIDIA GPUs with liquid cooling technology, is slated for early 2025 release.
