HPE has introduced two new rack-scale products aimed at cutting the cost of running large AI and high-performance computing clusters, including a hybrid storage system the company says delivers up to 41 per cent more throughput than its closest rival and costs up to 82 per cent less than an all-flash alternative.
Key Takeaways
- HPE’s new hybrid E2000 storage delivers up to 41% more data throughput per rack than rival DDN, and costs up to 82% less than a comparable all-flash system.
- Performance reaches up to 116 GB/s read and 92 GB/s write throughput per rack, with up to 32% more usable capacity than HPE’s current hybrid E2000 systems.
- A new Rack Management Controller adds real-time leak detection for liquid-cooled racks, arriving for HPE Cray GX systems in March 2027 and AMD/Nvidia systems in June 2027.
Hybrid storage built for AI superclusters
The new configurations of the HPE Cray Supercomputing Storage Systems E2000 combine solid-state drives with new 32-terabyte SAS-4 hard disk drives, housed in enclosures that hold 74 or 100 drives. HPE is calling it the world’s fastest hybrid supercomputing storage system, citing throughput of up to 116 GB/s for reads and 92 GB/s for writes per rack.
HPE said the system delivers up to 41 per cent more data throughput per storage rack than its nearest competitor, DDN, and costs up to 82 per cent less than a comparable all-flash system. That comparison is based on 15 petabytes of usable capacity meeting Nvidia’s performance specification for a 1,024-GPU cluster. The hybrid configuration also packs in up to 32 per cent more usable storage capacity per rack than HPE’s currently shipping hybrid E2000 systems.
“When making long-term investments in environments where advanced workloads are rapidly evolving, our customers want scalable, dense, and flexible systems to help them address dynamic shifts in requirements. Our comprehensive portfolio, which now includes a world-leading hybrid supercomputing storage option, delivers powerful capabilities to scale, while maximizing capacity and optimizing costs,” said Trish Damkroger, senior vice president and general manager of HPC and AI Infrastructure Solutions at HPE.
The hybrid E2000 systems are due to ship in early 2027.
Rack-level monitoring for liquid-cooled systems
HPE also introduced the Rack Management Controller, built for monitoring and controlling high-density, liquid-cooled HPC and AI environments. It adds real-time monitoring, leak detection and automated control at the rack level, and integrates with HPE’s Supercomputing Management Software suite, which covers Foundation, Composable and Observability.
The controller initially supports HPE Cray Supercomputing GX5000 systems, with HPE planning to extend support to AMD and Nvidia rack-scale systems later. Support for HPE Cray Supercomputing GX systems is due from March 2027, followed by AMD and Nvidia systems from June 2027.



