NVIDIA Launches OpenShell Reference Architecture for Safe AI Agents
NVIDIA announced the OpenShell Reference Architecture at GTC Berlin on October 21, 2026, enabling building, containing, observing and enforcing safe agents on enterprise infrastructure and edge systems. The architecture combines NVIDIA Vera Rubin infrastructure, NVIDIA Vera CPUs, NVIDIA BlueField-4 DPUs running NVIDIA Sentry software, and the NVIDIA OpenShell secure runtime. It is now available and supported by a broad ecosystem including Anthropic, AWS, Cisco, CrowdStrike, Dell Technologies, Google Cloud, HPE,...
Key takeaways
NVIDIA OpenShell Reference Architecture is now available for enterprise and edge deployment of safe AI agents.
The architecture combines NVIDIA Vera Rubin infrastructure, NVIDIA Vera CPUs, NVIDIA BlueField-4 DPUs with NVIDIA Sentry, and NVIDIA OpenShell secure runtime software.
For agentic workloads, NVIDIA Vera Rubin NVL72 systems deliver 30x higher throughput per megawatt.
NVIDIA Vera CPUs deliver 2x the performance of x86 processors with greater energy efficiency.
NVIDIA Sentry on BlueField-4 boosts agent capacity by up to 20% and provides in-silicon policy enforcement that quarantines agents attempting to move outside their software boundary in milliseconds.
Edge extension is provided through NVIDIA RTX Spark, NVIDIA DGX Spark and NVIDIA DGX Station systems.
The architecture runs on Canonical, SUSE, Red Hat, Microsoft Windows with Microsoft Execution Containers and Kubernetes, and across AWS, Google Cloud and Microsoft Azure operating systems.
Key metrics
Primary source
TickerStar generated this brief from an official company, investor-relations, earnings-call, or regulatory source. Review the primary material before making an investment decision.
Open Investor relationsFollow NVDA in real time
Get material earnings, filing, IR, and earnings-call alerts by email.