A new server configuration, the HPE ProLiant Compute DL394 Gen12 equipped with the NVIDIA Vera CPU, is scheduled to become available with HPE Private Cloud AI in 2027. Designed specifically for agent-focused workloads, the processor targets functions such as orchestration, tool interactions and real-time data handling. The New York Stock Exchange is already evaluating the technology through a collaboration involving Redpanda and HPE. HPE also plans to offer the NVIDIA Vera Rubin NVL72 rack-scale system and introduce the HPE Compute XD700 platform, which supports as many as 128 Rubin GPUs within a rack.

The software stack is also expanding. NVIDIA Agent Toolkit, which includes NVIDIA Nemotron open models, NVIDIA OpenShell secure runtime and NVIDIA NemoClaw blueprints, will be integrated with HPE Private Cloud AI. These technologies are intended to help organizations supervise agent behavior, apply governance requirements and operate autonomous multi-agent systems. HPE is also introducing local agent registration features that allow organizations to approve models, tools and skills against centralized governance and security controls before deployment.

Additional data protection and storage enhancements accompany the software updates. New capabilities in HPE Zerto Software can identify unauthorized agent actions and restore environments to a previously clean condition using continuous data protection. Meanwhile, HPE Alletra Storage MP X10000, which reached the foundation level of NVIDIA-Certified Storage, automatically applies metadata and governance policies to prepare unstructured information for AI workflows and improve token throughput.

Security remains a major focus across the portfolio. NVIDIA Confidential Computing is now supported through HPE AI Factory at Scale, HPE Sovereign AI Factory and HPE Private Cloud AI. The technology is designed to protect sensitive data and models during execution through encryption and cryptographic attestation. HPE ProLiant Compute DL380a also earned certification within the NVIDIA Confidential Computing program, while NVIDIA BlueField DPUs and NVIDIA DOCA provide security functions including zero-trust policy enforcement, threat detection and encrypted networking across AI workloads and agents.

The infrastructure expansion extends throughout the broader portfolio. HPE AI Factory at Scale, HPE Sovereign AI Factory and HPE Private Cloud AI can now be configured with NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs, Spectrum-X Ethernet networking, BlueField-3 DPUs and ConnectX-8 SuperNICs. Future Vera Rubin NVL72 deployments will include integrated NVIDIA networking technologies, while HPE is also expanding networking choices with NVIDIA InfiniBand options for large-scale and sovereign AI deployments. The company additionally expanded its Unleash AI ecosystem by adding nearly a dozen software partners, including Aizen, BridgeTEK, deepset, Deliverance, Faclon Labs, Gallop, Rocket, Supervity, Thales, Trustwise and Vortiqx.

📊 What This Means (Our Analysis)

The announcement reflects a broader shift from experimental AI projects toward operational environments that require governance, security, infrastructure integration and workload management. By extending hardware, software, storage and networking capabilities within a single portfolio, HPE is positioning its AI Factory offerings to address multiple requirements involved in running agent-based systems at scale.

What stands out is the emphasis on trust, oversight and operational control. The additions focus not only on computing performance but also on governance enforcement, confidential computing, agent monitoring and data protection. Together, those capabilities indicate that enterprises are increasingly prioritizing secure and manageable AI deployments rather than concentrating solely on model development.

📌 Our Take: As agentic AI moves deeper into production environments, the supporting infrastructure around security, governance and operational reliability is becoming just as important as raw computing power.