HPE Unveils Next-Gen Supercomputing Solutions for AI and HPC Workloads

HPE Expands Supercomputing Portfolio with Advanced AI and HPC Solutions

Hewlett Packard Enterprise (HPE) has unveiled a new lineup of high-performance computing (HPC) and artificial intelligence (AI) infrastructure solutions, aimed at accelerating scientific research and AI development. The announcement showcases HPE's commitment to advancing supercomputing capabilities for both traditional HPC workloads and emerging AI applications 1

1

2

3

Next-Generation HPE Cray Supercomputing EX Systems

At the forefront of HPE's new offerings are the HPE Cray Supercomputing EX systems, designed for research institutions and government entities developing sovereign AI initiatives. These systems feature:

1. HPE Cray Supercomputing EX4252 Gen 2 Compute Blade: Set to launch in Spring 2025, this system can deliver up to 98,304 cores in a single cabinet, featuring eight 5th Gen AMD EPYC processors 1

   1

   2

   2

   .

2. HPE Cray Supercomputing EX154n Accelerator Blade: Expected to be available by the end of 2025, this blade can accommodate up to 224 NVIDIA Blackwell GPUs in a single cabinet, utilizing the NVIDIA GB200 Grace Blackwell NVL4 Superchip 1

   1

   3

   3

   .

Industry-First Liquid Cooling and Enhanced Networking

HPE is introducing the industry's first 100% fanless direct liquid cooling system architecture, spanning compute nodes, networking, and storage. This innovation is complemented by:

• Slingshot interconnect 400: Offering 400 gigabit-per-second speeds, doubling the line speed of the previous generation 1

  1

  2

  2

  .
• Cray Supercomputing Storage Systems E2000: More than doubles the I/O performance compared to the previous generation 1

  1

  2

  2

  .

AI-Optimized HPE ProLiant Compute XD Servers

For service providers and enterprises focused on AI model training and tuning, HPE is rolling out new ProLiant Compute XD servers:

• HPE ProLiant Compute XD688: Designed for high-performance AI training and inference, offering a choice between NVIDIA H200 SXM Tensor Core GPUs or Blackwell GPUs 2

  2

  .
• HPE ProLiant Compute XD685: Available with either NVIDIA GPUs or AMD Instinct MI325X accelerators, providing flexibility for different AI workloads 2

  2

  3

  3

  .

Related Stories

HPE expands AI portfolio with Nvidia Blackwell and Vera Rubin systems for enterprise deployment

Hewlett Packard Enterprise and NVIDIA Deepen Collaboration on AI Factory Solutions

HPE and NVIDIA Unveil New AI Factory Solutions with Blackwell GPUs

HPE Ships First NVIDIA Grace Blackwell GB200 NVL72 System with Advanced Liquid Cooling for AI

Enhancing User Experience and Management

To improve the overall user experience and system management, HPE is introducing:

• Cray Supercomputing User Services Software: A new offering to optimize system efficiency, regulate power consumption, and flexibly run diverse workloads 1

  1

  .
• HPE Integrated Lights-Out (iLO) management technology: Exclusive to HPE ProLiant Compute servers, enabling secure out-of-band remote control access 1

  1

  .

Market Impact and Availability

These new offerings position HPE at the forefront of the rapidly evolving AI and HPC landscape. The company's focus on liquid cooling, increased GPU density, and flexible configurations caters to the growing demand for powerful, energy-efficient computing solutions in research and enterprise settings 1

1

2

3

While some products, like the HPE ProLiant Compute XD688, are set to launch in early 2025, others, such as the Cray Supercomputing EX154n Accelerator Blade, are scheduled for release in late 2025. This staggered release allows HPE to align with the availability of next-generation GPUs and CPUs from partners like NVIDIA and AMD 1

1

2

3