HPE adds 2U server for agentic AI workloads

The system uses NVIDIA Vera CPU and LPDDR5X memory to deliver 1.2 TB/s bandwidth and is planned for fall 2026 availability.

HPE announced the expansion of its industry-leading server portfolio with the introduction of the HPE ProLiant Compute DL394 Gen12, powered by NVIDIA Vera CPU. This next-generation server is engineered specifically to address the compute demands of emerging high-performance AI and data processing workloads delivering industry-leading agentic AI CPU performance, memory bandwidth, and low latency. Built on a foundation of advanced architecture, the platform also features HPE’s enterprise-grade Integrated Lights-Out (iLO) as security, while HPE Compute Ops Management provides a unified dashboard to manage and automate server environments.

The HPE ProLiant Compute DL394 Gen12, based on NVIDIA Vera CPU, was launched at COMPUTEX 2026 as part of a new collaboration with NVIDIA and Redpanda that is being explored by the New York Stock Exchange (NYSE). The new server features technology optimized for the specific demands of agentic AI, across data storage and processing, monitoring, management and security capabilities, ensuring smooth and secure operation for agentic AI.

The HPE ProLiant Compute DL394 Gen12, based on NVIDIA Vera CPU, was launched at COMPUTEX 2026 as part of a new collaboration with NVIDIA and Redpanda that is being explored by the New York Stock Exchange (NYSE). The new server features technology optimized for the specific demands of agentic AI, across data storage and processing, monitoring, management and security capabilities, ensuring smooth and secure operation for agentic AI.

Solving the memory challenge

HPE ProLiant Compute DL394 Gen12 is a 2U server that helps customers advance the adoption of AI computing in their organization. The server leverages NVIDIA Vera CPUs with a monolithic design, unlike traditional high-core-count chiplet architectures that suffer from non-uniform memory access (NUMA) issues, which results in variable latencies and non-deterministic performance can develop latency in multi-processor systems. By leveraging low-power double data rate 5X (LPDDR5X), a highly efficient form of dynamic random-access memory (DRAM), the HPE server achieves 1.2 TB/s aggregate bandwidth—up to 14 GB/s per core—enabling the HPE ProLiant Compute DL394 Gen12 to ingest and process data at high speed. This architecture enables NVIDIA Vera CPU to act as a high-speed orchestrator, ensuring resources are efficiently balanced to meet the most demanding AI workloads, while reducing capacity waste.

Embedded security and AI-driven operations with HPE ProLiant

HPE ProLiant Compute DL394 Gen12 embeds security at every level with Silicon Root of Trust, HPE’s firmware technology. Additionally, HPE ProLiant Compute servers with iLO 7—like the HPE ProLiant Compute DL394 Gen12, are enabled by the secure enclave, an HPE innovation that safeguards servers at every phase of its lifecycle. These next-generation servers are the first to meet NIST’s quantum computing resistant security requirements, giving organizations a more secure and future-ready platform for protecting sensitive workloads and regulated environments.

To enable AI-driven insights and enhance operational agility, the server is also equipped with HPE Compute Ops Management which provides a single, unified solution for overseeing distributed environments. This management layer provides customers with AI-driven operations that can reduce server management time and minimize the risk of revenue loss due to downtime.

Availability

The HPE ProLiant Compute DL394 Gen12 will be available in fall 2026, as a new addition to the NVIDIA AI Computing by HPE portfolio.

This server can also be acquired through HPE Financial Service’s 90/9 Advantage program that offers no payments for 90 days and an additional nine months at one percent.

For more information, visit hpe.com.