this article contains no affiliate links. All links are plain links, and we earn nothing if you click them. See our affiliate disclosure. How this article was prepared: AI helped gather and draft this research; a person checked the facts against the cited sources.
Update, November 19, 2026: the CNCF has welcomed OpenShell as a Sandbox project. OpenShell is adopting Cedar as its policy language, and the project is moving to the ai-openshell GitHub organization.
Quick answer: On November 10, 2026, NVIDIA contributed OpenShell, its open source runtime for securing and governing AI agents, to the Cloud Native Computing Foundation as a Sandbox project. It also open-sourced NVIDIA Snapshot and previewed Kubernetes Engine and Recipes. In one partner test, Snapshot cut inference ready time from about eight minutes to 31 seconds.
What OpenShell is
NVIDIA announced the move at KubeCon + CloudNativeCon in a blog post by Warren Barkley.

Source: NVIDIA blog — NVIDIA Expands Open Source Commitment With OpenShell Donation and New AI Infrastructure, captured Nov 11, 2026, about 13:20 MYT. Credit: NVIDIA. Cropped official blog header (headline, deck, date and author). Not a TechKit graphic. No NVIDIA product renders.
OpenShell is a rulebook layer between an AI agent and everything it touches. It applies deterministic controls to what an agent does, so a team can enforce policies as the agent uses tools, calls inference services and runs workloads.
NVIDIA says hundreds of organizations have put OpenShell to use, and that the project has received tens of thousands of contributions in less than a year since launch. Those are NVIDIA's figures.
Why handing it to CNCF matters
The CNCF is a nonprofit that's part of the Linux Foundation and hosts KubeCon. As a CNCF Sandbox project, OpenShell moves to the foundation's community-led home. Anyone can inspect and improve it.
Agent Substrate and the Google tie-in
NVIDIA is also contributing to Agent Substrate, an open source framework that originated at Google and is in the CNCF Sandbox. Agent Substrate preserves agent environments between tasks while releasing compute when agents are idle, so teams don't keep dedicated compute running for every session. NVIDIA is making it work with NVIDIA Dynamo for model inference and KAI Scheduler for sharing cluster resources.
Faster AI servers: NVKE, NVKR and Snapshot
Three new open source pieces:
– NVIDIA Kubernetes Engine (NVKE) and NVIDIA Kubernetes Recipes (NVKR), now in public preview, are building blocks for running managed Kubernetes on GPU servers. Mirantis, Netris, OpenNebula Systems, Rafay, Spectro Cloud and vCluster have aligned with NVIDIA on integration plans. Those are plans, not shipped products.
– NVIDIA Snapshot restores checkpointed replicas in seconds.
The speed numbers come from two design partners testing on NVIDIA HGX B300 systems, not from NVIDIA-wide benchmarks:
On a narrow screen, swipe the table sideways to see every column.
| Partner test | Before | With Snapshot | | — | — | — | | Crusoe ready time | roughly 8 minutes | 31 seconds | | Saturn Cloud, vLLM startup | 2 to 8.5 minutes | 6 to 34 seconds |
TechKit text card. Based on the NVIDIA blog, checked Nov 19, 2026 MYT. Not an official screenshot. Partner-test figures, not NVIDIA-wide benchmarks.
Who should care
If your team runs AI agents or inference on Kubernetes, OpenShell gives you a shared way to control what agents can do, and Snapshot targets slow cold starts. If you use AI tools through a chat app or an API and never touch servers, you can skip it.
For more agent infrastructure news, see AWS's new web search API for agents and today's trending AI agent repos.
Quick questions
What is NVIDIA OpenShell?
An open source runtime for securing and governing autonomous AI agents. It applies controls to what agents can do as they use tools, call inference and run workloads.
What does contributing to CNCF change?
OpenShell becomes a CNCF Sandbox project, moving to the foundation's community-led home.
What is Agent Substrate?
A Google-originated open source framework in the CNCF Sandbox for running long-running agents on Kubernetes. It preserves agent environments between tasks and frees compute when agents are idle.
What is NVIDIA Snapshot?
An open source component that restores checkpointed inference replicas in seconds.
Can I use it now?
OpenShell and Snapshot are open source. NVKE and NVKR are in public preview. NVIDIA's post doesn't give a general availability date.
Sources
NVIDIA Blog: NVIDIA Expands Open Source Commitment With OpenShell Donation and New AI Infrastructure (November 10, 2026) — https://blogs.nvidia.com/blog/open-source-openshell-infrastructure/ — November 10, 2026
CNCF Blog: Welcoming OpenShell to the CNCF: Open Governance to Secure Agentic AI (November 10, 2026) — https://www.cncf.io/blog/2026/11/10/welcoming-openshell-to-the-cncf-open-governance-to-secure-agentic-ai/ — November 10, 2026
Unite.AI: NVIDIA Donates OpenShell to CNCF and Adds Open-Source AI Factory Tools — https://www.unite.ai/nvidia-donates-openshell-to-cncf-and-adds-open-source-ai-factory-tools/ — November 10, 2026 (secondary, used for adoption figures, partner plans and the Dynamo/KAI tie-in)
Last checked: November 19, 2026, MYT.
1 thought on “NVIDIA OpenShell Joins CNCF: What It Means for AI Agents”