NVIDIA Donates OpenShell to CNCF and Expands AI Infrastructure Tools
NVIDIA announced the donation of its OpenShell runtime to the Cloud Native Computing Foundation and introduced new open-source infrastructure tools at KubeCon.
NVIDIA has announced the contribution of its OpenShell runtime to the Cloud Native Computing Foundation (CNCF) alongside new open-source components for the DSX AI Factory Platform at KubeCon + CloudNativeCon. The announcements affect developers, cloud-native communities, and organizations deploying large-scale GPU infrastructure and autonomous AI agents on Kubernetes.
The primary updates focus on open-source contributions for agent security, multi-tenant agent execution, and rapid cluster and model scaling. These tools aim to give practitioners deterministic governance over autonomous workflows and reduce operational friction when managing containerized GPU environments.
Securing Autonomous Agents with OpenShell
NVIDIA is contributing NVIDIA OpenShell, an open-source runtime for securing and governing autonomous agents, to the CNCF as a Sandbox project. OpenShell applies deterministic controls to agent activity, allowing teams to enforce security policies as agents invoke tools, access inference services, and execute workloads.
The architecture features extensible hooks enabling organizations to connect their own compute resources, inference providers, credential storage, and monitoring solutions. According to the company, OpenShell has seen adoption from hundreds of organizations and tens of thousands of contributions since its initial launch.
Scaling Agent Execution and Cluster Management
In addition to OpenShell, NVIDIA is contributing to Agent Substrate, a Google-originated open-source framework accepted into the CNCF Sandbox for running long-running agent workloads on Kubernetes. NVIDIA is integrating Agent Substrate with NVIDIA Dynamo for model inference and KAI Scheduler for resource allocation, while collaborating with Google on a unified integration between OpenShell and Agent Substrate.
For cluster and model deployment, NVIDIA introduced new components to the DSX AI Factory Platform:
- NVIDIA Kubernetes Engine (NVKE) and NVIDIA Kubernetes Recipes (NVKR): Available now in public preview, these open-source building blocks automate cluster lifecycle management and provide signed, version-pinned images for delivering managed Kubernetes on GPU infrastructure.
- NVIDIA Snapshot: A modular component designed to restore checkpointed replicas in seconds without model reloading or warm-up delays.
Availability and Integration Limits
NVKE and NVKR are currently available in public preview, backed by early ecosystem alignment from platform providers such as Mirantis, Netris, OpenNebula Systems, Rafay, Spectro Cloud, vCluster, Firmus, and SK Telecom. Meanwhile, NVIDIA Snapshot has been evaluated by design partners Crusoe and Saturn Cloud using NVIDIA HGX B300 systems, where initial testing demonstrated reduced workload readiness and vLLM startup times.
Source
Official announcement or documentation
Some links on this page may be affiliate links. If you buy through them we may earn a commission at no extra cost to you. See our affiliate disclosure.