HPE Supports Operational Deployment of Agent-Based AI with NVIDIA – Delivering a Foundation with Security, Governance, Scalability, and Sovereignty

HPE Private Cloud AI supports the deployment of secure AI agents with new governance capabilities and promotes improved token economics through expanded AI data pipelines.

Addition of NVIDIA Vera CPU, NVIDIA Agent Toolkit, and NVIDIA Confidential Computing to HPE AI Factory with NVIDIA

*This press release is a Japanese summary based on the English release issued by HPE (headquartered in Houston, Texas, USA) at HPE Discover Las Vegas 2026 on June 16, 2026 (local time). Please refer to the original for full details.

HPE has announced new innovations to support the transformation toward agent-based enterprises and to advance the operational deployment of AI under stronger security, governance, and control. The new products transforming the HPE AI Factory with NVIDIA will accelerate the evolution toward next-generation AI, where intelligence adapts, evolves, and collaborates under control.

Antonio Neri, President and CEO of HPE, said:

"As AI becomes more autonomous, new architectures are required to support secure execution, responsible control, and cost-efficient scalability. HPE, through our full-stack AI solutions co-developed with NVIDIA across networking, servers, storage, and software, will underpin the agent-based enterprise and help our customers confidently transition from proof-of-concept to production."

Jensen Huang, Founder and CEO of NVIDIA, said:

"Every layer of the computing stack is being redefined for the era of AI agents. The AI factory we are building together with HPE leverages the NVIDIA Vera CPU, accelerated infrastructure, and secure AI software to help enterprises transform data into intelligent actions."

Enabling Enterprise-Grade Agent-Based AI Operations with HPE AI Factory with NVIDIA

Across all industries, organizations are exploring how to fully leverage AI agents in production environments to automate business processes and enable more accurate decision-making. As organizations move toward full-scale operational deployment of agent-based AI and optimize token usage (the unit of AI processing), HPE is providing technologies to simplify this process and support more secure and high-performance AI operations. The turnkey AI factory solution co-developed with NVIDIA, HPE Private Cloud AI, adds new capabilities that deliver greater control, observability, and efficiency to support reliable, enterprise-grade deployment of agent-based AI.

Secure and governed agent-based AI systems provide the control functions necessary to confidently transition agents from development to production. The NVIDIA Agent Toolkit software, including the NVIDIA Nemotron open models, NVIDIA NemoClaw, and NVIDIA OpenShell secure runtime, delivers an agent operating system that enables efficient inference, monitoring of agent behavior, policy enforcement, and reduced deployment risks. HPE Private Cloud AI expands its product lineup with the HPE ProLiant Compute DL394 Gen12 featuring the NVIDIA Vera CPU, a compute-optimized platform designed for agent-based AI and high-performance data processing, including security and management features. Additionally, new capabilities in HPE Zerto Software will enable identification of malicious agent actions and allow rollback to a clean state using continuous data protection. HPE Private Cloud AI also supports secure local agent registration and provides functionality to approve AI models, skills, and tools under centralized governance and security policies.

While data is essential to maximizing AI value, it can also become the biggest bottleneck. HPE Private Cloud AI leverages the HPE Alletra Storage MP X10000 to transform unstructured data into AI-ready pipelines within minutes, supporting improved inference efficiency. The HPE Alletra Storage MP X10000 incorporates intelligence that automatically applies metadata and governance policies during data preparation for AI applications, reducing token processing response times by up to 20x (Note 1). HPE Private Cloud AI supports optimization by improving prompt processing efficiency and increasing token throughput by up to 20% (Note 2). HPE Data Fabric Software extends MCP (Model Context Protocol) support to Apache Airflow and implements an enterprise AI inventory (ledger) that tags metadata across distributed data, making data required for agent-based workflows easier to discover and utilize. The standalone HPE Data Fabric appliance, available on HPE ProLiant Compute servers, supports simplified and accelerated deployment.

HPE Private Cloud AI contributes to optimizing AI investments by supporting token cost reduction, maximizing GPU utilization, and enabling long-term scalability. New features include an integrated model gateway to govern access to frontier models, workload prioritization for running tasks, and multi-node inference across up to 256 GPUs. Additionally, using NVIDIA NeMo, pre-trained AI models—including NVIDIA Nemotron open models—can be fine-tuned for use in agent-based AI while securely accessing existing enterprise data.

Enhancing Security for Large-Scale HPE AI Factory Deployments

The following new features will enhance large-scale HPE AI Factory and HPE Sovereign AI Factory deployments.

NVIDIA Confidential Computing for Large-Scale and Sovereign Architectures: HPE Services will lead the integration of NVIDIA Confidential Computing into HPE AI Factory. NVIDIA Confidential Computing