NVIDIA Donates OpenShell to CNCF, Unveils New AI Infrastructure Tools

News

NVIDIA deepens open-source efforts with OpenShell donation to CNCF and new AI tools for Kubernetes at KubeCon 2026.

NVIDIA has announced significant contributions to the open-source community at KubeCon + CloudNativeCon 2026, unveiling new tools and initiatives to enhance AI infrastructure on Kubernetes. Central to these efforts is the donation of NVIDIA OpenShell, an open-source runtime for securing and governing autonomous AI agents, to the Cloud Native Computing Foundation (CNCF). This move underscores NVIDIA’s commitment to fostering collaboration and innovation in the cloud-native ecosystem.

OpenShell: Enhancing Security for AI Agents

OpenShell provides deterministic controls over AI agent activity, enabling organizations to enforce policies as agents access inference services, execute workloads, and utilize tools. Its extensible architecture integrates with existing enterprise systems, allowing governance across diverse deployment environments. According to Jim Zemlin, CEO of the Linux Foundation, “NVIDIA’s contribution of OpenShell to CNCF gives developers a shared foundation to build on and an opportunity to shape how agent security evolves.” Since its launch, OpenShell has garnered widespread adoption, with hundreds of organizations utilizing it and tens of thousands of contributions made to the project.

Scaling Agent Workloads with Agent Substrate

In addition to OpenShell, NVIDIA is contributing to Agent Substrate, a Google-led CNCF Sandbox project designed to manage long-running agent workloads at scale on Kubernetes. This framework integrates with NVIDIA Dynamo for model inference and KAI Scheduler for resource allocation, enabling efficient scaling of agent environments. By suspending and resuming agent environments as needed, Agent Substrate optimizes compute usage, reducing idle resource consumption. NVIDIA and Google are also collaborating to integrate OpenShell with Agent Substrate, ensuring secure and governed agent operations across cloud and on-premises systems.

New AI Tools for Kubernetes

NVIDIA has also introduced three new components to its DSX AI Factory Platform, aimed at simplifying Kubernetes management and scaling AI workloads:

• NVIDIA Kubernetes Engine (NVKE) and NVIDIA Kubernetes Recipes (NVKR): These open-source tools automate cluster lifecycle management and provide signed, version-pinned images. Currently in public preview, they are supported by ecosystem partners such as Mirantis, Netris, and SK Telecom.
• NVIDIA Snapshot: This modular component accelerates inference scaling by restoring checkpointed replicas in seconds, eliminating delays caused by model reloads and warmups. In tests by Crusoe and Saturn Cloud, NVIDIA Snapshot reduced workload startup times from minutes to seconds.

These advancements aim to make Kubernetes a more robust foundation for AI factories, enabling developers to deploy safer and more efficient AI infrastructure at scale.