HPE introduced new NVIDIA-backed AI Factory features to help enterprises deploy secure agentic AI with governance and scalable data pipelines.

Key Highlights

  • HPE Private Cloud AI features will become available from July 2026.
  • Token response times can improve by up to 20 times using HPE Alletra Storage MP X10000.
  • Token throughput can increase by up to 20% through prompt-processing improvements.
  • HPE AI Factory now supports NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs.

Hewlett Packard Enterprise Co. (NYSE:HPE) announced new AI infrastructure features on Tuesday designed to help enterprises move agentic AI systems into production with stronger security, governance and operational controls.

The updates were introduced at HPE Discover Las Vegas 2026 and expand the HPE AI Factory with NVIDIA. The company said the new offerings are aimed at enterprises seeking to deploy autonomous AI agents across business processes while maintaining visibility over data, models and workloads.

HPE Private Cloud AI, co-engineered with NVIDIA, is adding capabilities for secure agent deployment, agent monitoring, policy enforcement and local agent registration. The platform will support approved AI models, skills and tools under centralized governance rules.

The company said NVIDIA Agent Toolkit software, including open models, secure runtime tools and agent operating capabilities, will be used to support enterprise agentic AI deployments. HPE is also adding ProLiant Compute DL394 Gen12 systems with NVIDIA Vera CPUs for agentic AI and high-performance data processing.

HPE Zerto Software will add tools to detect rogue agent actions and restore systems through continuous data protection. These capabilities are expected to become available in the fourth quarter of 2026.

Data management is also a central part of the update. HPE said its Alletra Storage MP X10000 can apply metadata and governance policies to unstructured data and prepare it for AI applications.

The company said benchmark testing showed up to a 20 times improvement in time to first token using inference optimisation. HPE also said Private Cloud AI can improve prompt processing efficiency and raise token throughput by up to 20%.

HPE Data Fabric Software will extend model context protocol support to Apache Airflow and introduce an enterprise AI inventory for distributed data. The software is scheduled for availability in October 2026.

HPE is also adding NVIDIA Confidential Computing to AI Factory deployments. The feature is designed to protect models and private data during execution for on-premises and sovereign AI environments.

The broader HPE AI Factory platform now supports NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs, Spectrum-X Ethernet, BlueField-3 DPUs and ConnectX-8 SuperNICs. Those systems are available now.