Table of Contents
HPE Private Cloud AI is a purpose-built solution designed to provide fast and easy deployment of On-Prem AI applications with a focus on inferencing, Retrieval-Augmented Generation (RAG), and Fine-tuning. HPE Private Cloud AI is a co-developed HPE and NVIDIA® enterprise solution that includes a complete infrastructure and software portfolio.
Overview
At A Glance
HPE Private Cloud AI delivers a unique cloud experience designed to accelerate data science productivity and time to business value. It delivers instant AI productivity by arriving at a customer’s location ready to deploy in three clicks after hardware installation and software onboarding are complete. Once available, multiple personas have self-service access to a diverse set of NVIDIA technologies and open-source tools and models to increase productivity by 90% through an evergreen, cloud-managed experience.1
Notes: 1Source: HPE internal reports. Comparison between using GPT-4 via OpenAI API vs. self-hosted Llama3, assuming an enterprise account with 5,000 users, 5 chat sessions per day, 8,000 tokens per chat.
AI teams can innovate faster with built-in compliance and explainability to foster model trust, quickly detect model bias, diagnose and improve model performance, and remain compliant with industry regulations.
Built with enterprise-grade controls means organizations can fearlessly innovate, with a scalable platform — all controlled from a unified dashboard.
Future-proof your AI journey with HPE. Launch small, scale seamlessly, and invest confidently with our co-developed NVIDIA + HPE solution. One modular architecture protects customers’ investment by ensuring compatibility with future innovations from NVIDIA, HPE, and the open-source world.
- - Evergreen cloud experience with NVIDIA technologies and a rich ecosystem of open-source tools and models
- - Automated AI pipelines with clear data lineage and verifiable changes empower efficient, accountable development
- - Robust security, on-demand scalability, and compliance for data and AI models — all managed from a single dashboard
One of the challenges businesses face is getting AI pilots to production faster.
HPE Private Cloud for AI delivers instant AI productivity with a unique, private cloud experience that accelerates the productivity of data science teams and time to business value with NVIDIA AI Computing.
HPE Private Cloud for AI offers enterprise customers the ability to leverage NVIDIA AI Enterprise (NVAIE) portfolio, including NVIDIA Inferencing Microservices (NIM), and HPE portfolio of curated market adopted open-source AI tools and platforms with full private control of their data. The solution will enable enterprises to expedite their Machine Learning and AI initiatives starting from creating their private data lakehouses, to data pipeline, model development and fine-tuning, to operationalizing their GenAI workflows.
What’s New
- − HPE Private Cloud AI Air-Gapped Large deployment option
- − HPE Private Cloud AI Developer system featuring NVIDIA RTX Pro 6000 Blackwell Server Edition GPUs
- − HPE Private Cloud AI Large configuration featuring NVIDIA RTX Pro 6000 Blackwell Server Edition GPUs
As agentic AI simplifies automation and provides consistent profitability from AI workloads, the importance of rapid productivity, reliability, and data protection has never been greater. To accelerate the development and delivery of agentic AI across industries, Hewlett Packard Enterprise and NVIDIA® are introducing the Private Cloud AI Developer and Large configurations featuring the NVIDIA RTX Pro 6000 Blackwell Server Edition GPUs to their end-to-end enterprise AI platform with support for Physical AI and Visual Computing workloads HPE is also supporting secure deployments with the Private Cloud AI Large air-gapped solution. Ensure the protection of sensitive data with private AI model customization by eliminating exposure to external networks.
| Category | Description |
|---|---|
| Platform | - Server support: HPE ProLiant Compute Gen11 and Gen12 servers - AMD-based HPE ProLiant Compute DL325 Gen 11 Control Nodes - Intel-based HPE ProLiant Compute DL380a Gen 11 AI Worker Nodes (G1) - Intel-based HPE ProLiant Compute DL380a Gen 12 AI Worker Nodes (G2) - Storage support: HPE GreenLake for File with Object Storage enabled for Small/Medium /Large T-shirt size |
| Manageability | - Cloud-based setup and lifecycle management (single-click upgrades) |
| Analytics & Monitoring | - Cluster and VM capacity and performance, storage health status information |
| Support | - One call support experience with HPE Services |
- HPE Private Cloud AI Smart templates
- Availability of pre-configured Smart Templates with HPE ProLiant Compute Gen 12 Servers
HPE Private Cloud AI Family
| Feature | Developer System | Small | Medium | Large |
|---|---|---|---|---|
| Control Node Qty | 1 | 3 | 3 | 3 |
| Worker Node Qty | 1 | 1 or 2 | 2 | 2 |
| Worker Node Generation | HPE ProLiant Compute Gen12 | HPE ProLiant Gen12 | HPE ProLiant Gen12 | HPE ProLiant Gen12 |
| CPU Type / Qty (per node) | 2x Xeon 32 Core CPUs | 2x Xeon 86 Core CPUs | 2x Xeon 86 Core CPUs | 2x Xeon 86 Core CPUs |
| GPU Type / Qty | 2x RTX Pro 6000 | 4 or 8x RTX Pro 6000 | 8x H200 | 16x H200 or RTX Pro 6000 |
| Storage | 22 TB Internal File/Object | 109 TB GreenLake for File with Object Storage | 109 TB GreenLake for File with Object Storage | 217T B GreenLake for File with Object Storage |
| Networking Switches | N/A | NVIDIA 4700M Aruba 6300M (oobm) | NVIDIA 4700M Aruba 6300M (oobm) | NVIDIA 4700M Aruba 6300M (oobm) |
| NIC Speed (AI Network) | 200 Gb NICs | 400 Gb NICs | 400 Gb NICs | 400 Gb NICs |
| Rack / PDU | N/A | 1x 42U Rack with PDUs | 1x 42U Rack with PDUs | 1x 42U Rack with PDUs |
| Install Services Included | N/A | Yes | Yes | Yes |
| Sales Motion | Traditional or GreenLake | Traditional or GreenLake | Traditional or GreenLake | Traditional or GreenLake |
| Air-Gapped Deployment Option | No | No | Yes | Yes |
| Expansion Rack Support | No | Yes | Yes | Yes |
HPE Private Cloud AI Expansion Racks
| Feature | RTX Expansion Rack | H200 Expansion Rack |
|---|---|---|
| Control Node Qty | 0 | 0 |
| Worker Node QTY | 2 | 2 |
| Worker Node Generation | HPE ProLiant Gen12 | HPE ProLiant Gen12 |
| CPU Type / QTY (per node) | 2x Xeon 86 Core CPUs | 2x Xeon 86 Core CPUs |
| GPU Type / QTY (per solution) | 8x RTX Pro 6000 (Small) 16x RTX Pro 6000 (Large) | 16x H200 (Medium or Large) |
| Storage | N/A | N/A |
| Networking Switches | N/A | N/A |
| NIC Speed (AI Network) | N/A | N/A |
| Rack / PDU | 400Gb NICs | 400Gb NICs |
| Install Services Included | 1x 42U Rack with PDUs | 1x 42U Rack with PDUs |
| Sales Motion | Yes | Yes |
| Air-Gapped Deployment Option | Traditional or GreenLake | Traditional or GreenLake |
| Expansion Racks Supported | No | Yes |
Key Features and Benefits
HPE Private Cloud AI is turnkey, deployed in a few hours, cloud-managed, and ready to use by AI personas and IT operations teams and provides rapid productivity for AI initiatives while protecting data and IP. The key value propositions aligned to customer problems are:
The core feature set includes:
- - Instant AI productivity: HPE Private Cloud AI provides a unique, private cloud experience that accelerates data science productivity and time to business value with NVIDIA AI Computing. The solution is pre-integrated and ready to run out of the box in hours. It is not a reference architecture like other solutions in the market.
- - Unify access to all your data: Secure and Unified access to all your data: HPE simplifies data management and reduces cost and complexity by integrating, organizing, and governing enterprise data for seamless access, data integrity and compliance. Enterprise-grade confidence and control: HPE Private Cloud AI is managed through a simple control plane on HPE GreenLake. Users can easily provision, orchestrate, manage and monitor the private cloud environment and the hybrid cloud landscape it exists within. Comprehensive, multi-layered controls protect sensitive data and models and maintain high performance, reliability and utilization of AI infrastructure.
Cloud experience that keeps data private: HPE Private Cloud AI delivers a true cloud experience through HPE GreenLake. Deployed on-premises and designed for hybrid, HPE Private Cloud AI provides flexible and modular choices to expand and grow with AI demand. As business needs change, it’s easy for customers to grow the solution. And monthly subscription pricing allows customers to start small financially and grow as their projects prove ROI.
Service and Support
Support is included as part of the subscription for HPE Private Cloud AI. Included with the support is 24x7 telephone and email support for the arrays and hardware components for the chosen subscription term.
Refer to the HPE Private Cloud AI Data sheet https://www.hpe.com/psnow/doc/a50010051enw?section=Document%20Types for the service deliverables and the shared responsibility model as part of the subscription.
Configuration Information
Easy Configuration through Smart templates
There are pre-defined smart templates available that allow for quick and easy ways to quote:
- HPE Private Cloud AI
There are pre-defined smart templates that allow for quick and easy way to quote HPE Private Cloud AI.
Here is an example of a HPE Private Cloud AI Smart template
Config Name: PrivateCloudAI-Small-1Svr/4xL40S GPU-109TB File/Object-3Phase/NA-Jpn-PDU-1Rack-3yr
Description: HPE Private Cloud AI Small Single Node-4GPU Solution for AI Inference. 109TB File/Object Storage, 100GbE Networking, Single Rack and 3Phase PDU
The Smart templates contain the following attributes to choose,
- 1. T-Shirt Sizing – Developer System, Small, Medium, or Large Configurations
- 2. Workload Tier –
- a. AI Inference
- b. Retrieval Augmented Generation (RAG)
- c. Model Fine-Tuning
- d. Visual / Physical AI
| PCAI T-Shirt Size | Base | Expanded |
|---|---|---|
| Developer System | 2x RTX Pro 6000 GPUs and 22TB of integrated Storage for developing low-mid parameter model AI applications | N/A |
| Small | 4 or 8x RTX Pro 6000 GPUs and 109TB Storage for AI Inference and Visual Computing | 8 or 12 RTX Pro 6000 GPUs via 1x Expansion Rack |
| Medium | 8x H200 GPUs and 109TB Storage for AI Inference and RAG | 24x H200 GPUs via 1x Expansion Rack |
| Large | 16x H200 or RTX Pro 6000 GPUs and 217TB Storage for AI Inference, RAG, Fine-tuning, and Visual / Physical AI | Up to 64x H200 or RTX Pro 6000 GPUs via 3x Expansion Racks |
3. Network Configuration
| Network Configuration (Small/Medium/Large) | Detail |
|---|---|
| Networking equipment included | Two top-of-rack switches and out of band management switches are included along with all transceivers and signal cabling required for l in-rack solution (Customers are responsible for transceivers to connect to core network switches) |
| Network Configuration (developer system) | Detail |
|---|---|
| Networking equipment included | None- Requires customer furnished networking (100GbE or 200GbE recommended) |
4. Rack and power Configuration
| Rack Configuration (Small/Medium/Large) | Detail |
|---|---|
| Rack included | The solution will include a 42U Rack with integrated PDUs for HPE Private Cloud AI. |
| Rack Dimensions: 600mm (W), 1200mm (D) |
| Rack Configuration (developer system) | Detail |
|---|---|
| Rack included | None- Requires customer furnished rack and PDUs (~2400W Recommended) |
Resources and additional links
- - The networking requirements, best practices, supported technologies, and supported network topologies for HPE Private Cloud AI: https://psnow.ext.hpe.com/doc/a00114771enw
Shared Responsibility Model (SRM)
HPE Private Cloud AI subscription includes the necessary hardware, software, and services to deliver the service level specified. The service levels offered are based on a foundational shared responsibility model (SRM) depicted below:
HPE Private Cloud AI Small/Medium/Large Configurations
| Customer | HPE |
|---|---|
| Responsible for the connectivity to GreenLake Cloud Platform (GLCP), the administration, and the management of the data/objects | Responsible for the functionality of the service |
| Site Readiness including datacenter facilities and internet connectivity | Installation of hardware systems & activation of Service |
| Maintain connectivity to GreenLake Cloud Platform Data resilience and remote replication Data backup | Customer Orientation Access to software, firmware, and documentation updates Onsite hardware support |
| Applying recommended software updates & security patches Data Monitoring | |
| Initiating the order of additional capacity beyond total available capacity | Operational guidance through the platform |
| Red Hat Linux and Rocky Linux OS lifecycle management | |
| NVIDIA and Aruba Switch OS/Firmware lifecycle management | Operational insights and dashboard through the platform |
| HPE PDU Firmware lifecycle management | |
| NVIDIA GPU Firmware lifecycle management | |
HPE Private Cloud AI Medium Air-Gapped Configuration
| Customer | HPE |
|---|---|
| Site Readiness including datacenter facilities | Installation of hardware systems & activation of Service |
| Data resilience and remote replication Data backup | Customer Orientation Access to software, firmware, and documentation updates Onsite hardware support |
| Applying recommended software updates & security patches Data Monitoring | |
| Initiating the order of additional capacity beyond total available capacity | |
| NVIDIA and Aruba Switch OS/Firmware lifecycle management | |
| HPE PDU Firmware lifecycle management | |
| NVIDIA GPU Firmware lifecycle management | |
HPE Private Cloud AI developer system
| Customer | HPE |
|---|---|
| Responsible for the connectivity to GreenLake Cloud Platform (GLCP), the administration, and the management of the data/ objects | Responsible for the functionality of the service |
| Site Readiness including datacenter facilities and internet connectivity | |
| Maintain connectivity to GreenLake Cloud Platform Data resilience and remote replication Data backup | Access to software, firmware, and documentation updates Onsite hardware support |
| Applying recommended software updates & security patches Data Monitoring | |
| Initiating the order of additional capacity beyond total available capacity | Operational guidance through the platform |
| Red Hat Linux and Rocky Linux OS lifecycle management | |
| Installation of hardware systems & activation of Service | Operational insights and dashboard through the platform |
| NVIDIA GPU Firmware lifecycle management | |
Pre-requisite for HPE Private Cloud AI
As part of the shared responsibility model, the customer is expected to make appropriate decisions including but not limited to:
- - Rack Infrastructure
- Space
- Rails
- - Power Infrastructure
- PDU − Cables
Summary of Changes
| Date | Version History | Action | Description of Change |
|---|---|---|---|
| 02-Mar-2026 | Changed | The current version of HPE Private Cloud AI QuickSpecs reflects a major generational update focused on next‑generation NVIDIA Blackwell GPUs, simplification of configurations, and expanded secure (air‑gapped) deployment options. Legacy G1 configurations have been removed, while new Developer, Large, and disconnected (air‑gapped) options have been introduced to support modern GenAI, RAG, fine‑tuning, and Physical/Visual AI workloads.
| |
| 06-Oct-2025 | Changed | HPE Rebranding FY25 | |
| 02-Sep-2025 | Changed | Overview and Configuration Information sections were updated. - AI G2 Small configuration featuring NVIDIA® RTX Pro 6000 Blackwell Server Edition GPUs information was added. | |
| 04-Aug-2025 | Changed | Overview and Configuration Information sections were updated. - Added G2 Medium Air-Gapped configurations | |
| 21-Jul-2025 | Changed | Survey link updated. | |
| 02-Jun-2025 | Changed | Overview and Configuration Information sections were updated. | |
| 07-Apr-2025 | Changed | Overview and Configuration Information sections were updated. Updates for developer system | |
| 03-Mar-2025 | Changed | Overview section was updated | |
| 03-Sep-2024 | New | New QuickSpecs |
© Copyright 2026 Hewlett Packard Enterprise Development Company LP. The information contained herein is subject to change without notice. The only warranties for Hewlett Packard Enterprise products and services are set forth in the express warranty statements accompanying such products and services. Nothing herein should be construed as constituting an additional warranty. Hewlett Packard Enterprise shall not be liable for technical or editorial errors or omissions contained herein.
a50009216enw, 17248 - Worldwide - V9 -02-March-2026