HPE tightens Nvidia ties in AI stack, integrates Juniper

HPE launched an updated AI factory stack on the latest Nvidia technologies including the Vera CPU, added features to HPE Private Cloud AI and integrated the Juniper networking portfolio as part of its self-driving network strategy.

The barrage landed as HPE Discover kicked off in Las Vegas. HPE delivered strong earnings results and is benefiting from a surge in AI infrastructure demand.

CEO Antonio Neri said HPE is looking to enable enterprises to create new architecture that can scale AI agents by delivering a full stack of infrastructure largely with Nvidia.

• HPE's Q2 rides AI infrastructure boom
• Nvidia's Vera CPU, DGX Station, Windows PCs all go to the same place: AI agents running locally

HPE announced the following:

• HPE Private Cloud AI enhancements that include capabilities that boost observability and efficiency of agentic AI deployments.
• HPE is adding Nvidia Agent Toolkit including Nvidia Nemotron open models, Nvidia NemoClaw and Nvidia OpenShell to HPE Private Cloud AI.
• The HPE Private Cloud AI stack will add HPE ProLiant Compute DL394 Gen12, HPE's latest server, with Nvidia's Vera CPU.
• HPE Zerto Software updates that allow customers to identify rough agent actions.
• Support for secure local agent registration where customers can approve AI models, skills and tools under centralized governance.
• Inference efficiency enhancements via HPE Alletra Storage MPX10000 storage systems. HPE claims the ability to automatically apply metadata and governance to AI applications cuts token response times by 20x.
• HPE Data Fabric Software supports agentic AI by extending MCP support to Apache Airflow.
• HPE said it will integrate Nvidia Confidential Computing into its HPE AI Factory.
• HPE AI Factory and HPE Sovereign AI Factory will be available with Nvidia's latest servers, networking and components based on Nvidia's reference architecture.

Juniper integration

HPE announced that it has integrated Juniper's networking portfolio with its AI data center lineup. HPE is looking to create a self-driving network that runs across AI data centers to edge computing.

Here's a rundown of the combination of HPE and Juniper's networking stack.

• HPE Mist will support HPE Networking CX Switches. HPE also added the Marvis framework into HPE Aruba Central. Both Mist and Marvis were acquired in the Juniper acquisition. Mist is an AI-native cloud infrastructure for enterprise networks and Marvis is the built-in AI assistant.
• HPE launched new HPE Juniper Networking QFX Switches designed for AI inferencing. The HPE Juniper Networking QFX5140 Switch is designed for inference clusters and edge AI use cases. HPE Juniper Networking QFX5252 Switch tray for AMD Helios is a scale-up module designed for AMD Helios AI systems.
• Juniper's data center switching portfolio will be integrated into HPE AI Data Center lineup.
• A new SASE platform will converge with networking and security operations.
• HPE's pre-integrated AI infrastructure will support Juniper networking and management.