Hewlett Packard Enterprise has integrated the Nvidia OpenShell framework, part of the Nvidia Agent Toolkit, into its Private Cloud AI platform. The announcement was made on June 16 at the HPE Discover event in Las Vegas.
The integration functions as an agent operating system, providing behavior monitoring, policy enforcement, and action isolation through sandboxing. It also includes rollback capabilities powered by HPE’s Zerto software.
OpenShell is an open-source framework that allows for control enforcement without requiring modifications to the underlying code of AI agents. The platform also features new detection capabilities for rogue agents.
HPE has added support for the Nvidia Vera CPU to its ProLiant Compute DL394 Gen12 servers. The infrastructure now supports up to 256 GPUs in multi-node inference configurations.
The phased rollout of these features is scheduled to begin in July 2026, with full support for the Nvidia Agent Toolkit expected by the fourth quarter of 2026.
