HPE Increases Storage Capacity for Larger AI Workloads

HPE announced it is expanding its storage options to support larger AI workloads—introducing Alletra Storage MP X10000 Release 4, now generally available.

According to HPE, the platform expands the platform's ability to support larger AI deployments, accelerate inference, and reach more enterprise and government opportunities, with meaningful advances across scale, NVIDIA ecosystem integration, data intelligence, security, and operations.

Release 4 is organized around the three things that matter most to the customers deploying it: unified storage at greater scale, AI efficiency, and operational resilience.

Release 4 doubles single-cluster scale. A single X10000 now supports up to 16 nodes and 16 JBOFs, delivering roughly twice the performance and twice the capacity of the prior release and scaling to approximately 23 PB of raw capacity in one system. That is the headroom to consolidate AI training, inference, analytics, and large unstructured data repositories on one platform, rather than requiring siloed architectures or disruptive upgrades, HPE said.

File and object now live natively on the same platform. Release 4 adds native-namespace NFS alongside object storage, so customers get file and object as first-class protocols without translation overhead or a second silo to manage.

Customers can grow into far larger AI and data lake environments on a single, unified foundation, and address more of their unstructured data estate without fragmenting it.

Release 4 deepens HPE’s NVIDIA integration, extending its RDMA acceleration from object to file RDMA support for file workloads, GPUDirect Storage enablement, and readiness for NVIDIA AI Enterprise certifications.

“This builds on a proof point the industry has already recognized. The X10000 is the first object storage platform to achieve NVIDIA-Certified Storage validation, and independent testing of the RDMA-accelerated X10000 for KV cache offload showed up to 20x faster time to first token and up to 17x higher effective inference throughput, improving GPU utilization and lowering cost per token,” HPE said.

The X10000 Data Intelligence platform prepares data for AI directly inside the storage environment, and Release 4 advances it with NVIDIA NIM-based multimodal AI capabilities, multinode data intelligence deployments, and support for customer-developed AI functions.

Faster data access across NVIDIA-powered environments, higher infrastructure utilization, and a shorter path from raw data to AI-ready data, without standing up a separate data preparation stack.

Release 4 strengthens the platform for the most security-conscious and sovereign environments, including expanded support for disconnected and air-gapped deployments.

These capabilities open new conversations with government, regulated industries, and enterprises that require stronger controls and operational isolation

Customers also gain more flexible SaaS subscription options, including 1-, 6-, and 7-year terms.

Customers can consolidate unstructured data, enrich it with intelligence, accelerate access to GPUs, and serve trusted data to inference and agentic AI workflows, all on one platform, said the vendor.

HPE Alletra Storage MP X10000 Release 4 is available now.

For more information about this news, visit www.hpe.com.