HPE expands NVIDIA-based AI infrastructure portfolio

Hewlett Packard Enterprise has expanded its NVIDIA-focused AI infrastructure portfolio, adding new servers, turnkey systems, networking and storage products, plus services and financing aimed at production deployments.

The updates fall under the NVIDIA AI Computing by HPE portfolio and are positioned around secure deployment, scaling and governance as organisations move from pilots to operational AI.

"The AI race is fundamentally about speed, scale, and trust," said Antonio Neri, President and CEO of HPE. "Our industry leadership across cloud, networking, and AI enables organisations to operationalise AI securely, efficiently, and at an unprecedented scale. Together with NVIDIA, HPE delivers turnkey AI factories and networks that transform AI ambitions into real enterprise value."

NVIDIA described the collaboration as a foundation for enterprise adoption. "NVIDIA and HPE are setting a new standard for enterprise AI infrastructure," said Jensen Huang, Founder and CEO of NVIDIA. "HPE's leadership across private cloud, networking, and secure on-prem systems uniquely positions them to make AI a core enterprise capability. Together, we are building AI factories and AI grids - foundational infrastructure to embed intelligence into every workflow."

Private Cloud AI

A central element is an expansion of HPE Private Cloud AI, a packaged system co-engineered with NVIDIA. HPE described it as a turnkey "AI factory" designed for enterprise inferencing, with updates intended to improve performance, scalability and deployment options.

New network expansion racks raise the maximum scale of an HPE Private Cloud AI deployment to 128 GPUs, allowing customers to run larger workloads while keeping the same operational experience.

An air-gapped configuration is also being added for the large HPE Private Cloud AI system, targeting environments that require full isolation or sovereign deployment with no exposure to external networks.

HPE ProLiant Compute DL380a Gen12 servers, along with HPE Private Cloud AI systems based on the DL380a, are being certified for Fortanix Confidential AI. The joint solution uses NVIDIA Confidential Computing and is positioned for on-premises processing of sensitive data and models.

CrowdStrike is part of the updated package, adding agentic security for HPE Private Cloud AI, including threat detection and response for AI infrastructure, models and AI agents.

The stack also includes NVIDIA AI Enterprise software and blueprints, including the updated NVIDIA AI-Q blueprint for AI agents and a new NVIDIA Omniverse blueprint for digital twins. HPE is updating HPE Private Cloud AI, its latest ProLiant servers and its AI factory systems to support NVIDIA Nemotron open models, part of the NVIDIA Agent Toolkit.

NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs are available across configurations of HPE Private Cloud AI and HPE's AI factory offerings.

Vertical packages

HPE also introduced new NVIDIA co-designed, multi-workload solutions for areas including retail, medical research and manufacturing. Cited use cases include autonomous edge intelligence, retail shopping assistance, video search and summarisation, and biomedical research.

The designs combine HPE ProLiant Compute servers with NVIDIA accelerated computing and NVIDIA Spectrum-X Ethernet networking. They also include BlueField data processing units and ConnectX network interface cards, based on NVIDIA Enterprise Reference Architectures.

The software layer includes NVIDIA AI Enterprise and NVIDIA CUDA-X libraries, along with blueprints and confidential computing. HPE also pointed to Multi-Instance GPU and virtual GPU technologies, its chip-to-cloud security approach, and management automation through HPE Compute Ops Management.

In addition, NVIDIA RTX PRO 4500 Blackwell Server Edition GPUs are being added to HPE ProLiant servers for edge deployments and data analytics workloads. HPE is developing systems built on RTX 4500 Blackwell GPUs, including an integration with the NVIDIA Retail Shopping Assistant Blueprint.

Networking and storage

Separate updates focus on networking and storage infrastructure for distributed AI. HPE is introducing networking solutions for service providers, sovereign customers and large enterprises to connect distributed AI deployments, using HPE Juniper Networking routers and coherent optics.

HPE has also expanded its at-scale and sovereign AI factory offerings for large organisations, and is planning a new generation of systems based on the NVIDIA Vera Rubin architecture for demanding AI workloads.

On the data side, HPE highlighted inference context and broader data pipelines as bottlenecks as AI moves into production. It is working with NVIDIA across the AI data lifecycle, from ingest and vectorisation through inference and recovery.

HPE said Alletra Storage MP X10000 is the first vendor platform to achieve NVIDIA-Certified Storage validation for object-based systems at the Foundation level. The validation covers loads of up to 128 GPUs and includes functional tests for availability and reliability.

HPE will also support NVIDIA's STX rack-scale reference architecture as part of new AI storage development, citing future systems that combine Vera Rubin, BlueField-4, Spectrum-X networking, ConnectX network interfaces and NVIDIA AI software.

Services and financing

HPE Services is launching an "agents hub" focused on structured adoption of agentic AI. It is developing and validating agents powered by NVIDIA Nemotron models, with reusable patterns for enterprise deployment.

HPE Services and Protopia AI are also working on a blueprint for multi-tenant AI factories in regulated and sovereign environments. HPE said the design focuses on using sensitive data without exposure.

HPE Financial Services announced a "90/9 Advantage" financing programme. The offer requires no payments for the first 90 days, followed by monthly lease payments of 1 percent for the next nine months, across networking, hybrid cloud and compute servers.

Support for NVIDIA RTX PRO 4500 Blackwell Server Edition GPUs across the HPE ProLiant Compute server portfolio is expected to roll out in the first half of 2026, with Fortanix support for HPE ProLiant DL380a Gen12 systems planned for later in 2026.