HPE Expands Its AI Factory Portfolio for Deploying Agentic AI

During the HPE Discover 2026 event, Hewlett Packard Enterprise announced a series of new enhancements to its AI portfolio aimed at enabling easier deployment of agentic AI solutions in real production environments. The updates expand the capabilities of HPE AI Factory and NVIDIA AI Computing by HPE, with a focus on management, security, data preparation, and scaling of AI infrastructure.

HPE Private Cloud AI gains new capabilities for Agentic AI

One of the key updates is the expansion of HPE Private Cloud AI – the platform developed jointly with NVIDIA.

The system now supports:

• NVIDIA Agent Toolkit
• NVIDIA Nemotron models
• NVIDIA NemoClaw
• NVIDIA OpenShell runtime environment
  
  

These technologies provide mechanisms for AI reasoning, policy-based security control, agent behavior monitoring, and operational management.

HPE also introduced the new HPE ProLiant Compute DL394 Gen12 with support for NVIDIA Vera processors, optimized for agentic AI workloads and high-performance data processing.

Better control and recovery of AI environments

The company is expanding the capabilities of HPE Zerto by adding monitoring of AI agent actions and detection of potentially risky or unauthorized behavior.

With continuous data protection features, organizations will be able to restore their environments to a previously validated state when needed.

Additionally, local registration of AI agents is introduced, enabling approval of models, tools, and AI skills through centralized security and governance policies.

Data preparation and more efficient inference

Data preparation remains one of the biggest challenges in deploying AI solutions.

To address this, HPE is adding intelligent capabilities to HPE Alletra Storage MP X10000, which automatically tag metadata and apply governance policies to unstructured data.

According to the company, this enables:

• faster creation of AI-ready datasets
• up to 20x faster response time for token requests
• up to 20% improvement in prompt processing
• up to 20% increase in overall inference throughput
  
  

Development of HPE Data Fabric HPE Data Fabric also receives new features for agentic AI environments.

These include:

• support for Model Context Protocol (MCP) for Apache Airflow
• a centralized AI inventory catalog
• enrichment of distributed data with metadata
• improved data discoverability and management
  
  

The company will also offer a standalone HPE Data Fabric appliance solution on HPE ProLiant servers.

Improved resource and cost management

HPE is introducing a new unified model gateway that provides centralized and controlled management of multiple AI models through a single interface.

The platform also adds:

• workload prioritization
• multi-node inference support
• scaling up to 256 GPU accelerators
• fine-tuning capabilities for pre-trained models using NVIDIA NeMo
  
  

The goal is to help organizations use GPU resources more efficiently and optimize AI infrastructure costs.

Confidential Computing for AI environments

For large enterprise and sovereign AI environments, HPE is integrating NVIDIA Confidential Computing technologies into its AI Factory portfolio.

The technology protects AI models and sensitive data during runtime through:

• hardware-based protection
• encryption
• cryptographic attestation
  
  

The solution is designed to support compliance with regulatory, industry, and sovereign requirements.

Additional protection is provided through:

• NVIDIA BlueField-3 DPUs
• NVIDIA DOCA software
• zero-trust policies
• runtime threat detection
• encrypted network connections
  
  

Expanded integration with next-generation NVIDIA technologies

HPE also announced broader support for the latest NVIDIA AI technologies, including:

• NVIDIA RTX PRO 6000 Blackwell Server Edition
• NVIDIA Spectrum-X
• NVIDIA BlueField-3 DPU
• NVIDIA ConnectX-8 SuperNIC
  
  

The solutions are based on NVIDIA reference architectures and are designed to support the full lifecycle of AI projects – from training and development to production deployment.

Availability

HPE plans a phased rollout of the new features:

• Unified Model Gateway and part of the new HPE Private Cloud AI capabilities – July 2026
• Updates for HPE Data Fabric Software – October 2026
• Agentic observability, Data Intelligence Services, integration with Alletra MP X10000,
• NVIDIA Agent Toolkit and NemoClaw – Q4 2026
• HPE Zerto Agent Monitoring and NVIDIA Confidential Computing – Q4 2026
  
  

Support for NVIDIA RTX PRO 6000 Blackwell Server Edition, Spectrum-X, BlueField-3, and ConnectX-8 is available immediately.