Reported News
NVIDIA donates OpenShell agent runtime to CNCF, open-sources Kubernetes Engine and Snapshot
On November 10, 2026, NVIDIA said it is contributing OpenShell, "an open source runtime for securing and governing autonomous agents," to the Cloud Native Computing Foundation as a Sandbox project. In the same post it reports a design partner cut a GLM-5.3 workload's time to ready from roughly eight minutes to 31 seconds with its new open source Snapshot component.
- Changes reported
- 5
- listed below, each from a source we read
- We tested
- Not tested yet
- reported from sources; we did not run it ourselves
- Gaps we flag
- 4
- things we could not verify
- Sources
- 1
- 1 primary, linked below
What changed
| # | What the source says | Status |
|---|---|---|
| 01 | OpenShell, NVIDIA's runtime that applies deterministic controls to agent activity as agents use tools, call inference services and run workloads, moves to the CNCF as a Sandbox project, governed by the organizations building it into their products. | Reported |
| 02 | NVIDIA says hundreds of organizations use OpenShell and that the project has received tens of thousands of contributions in less than a year since launch. | Reported |
| 03 | NVIDIA is contributing to Agent Substrate, a Google-originated CNCF Sandbox framework that suspends and resumes long-running agent environments across a shared pool, wiring it to NVIDIA Dynamo and KAI Scheduler; the two companies are also integrating it with OpenShell. | Reported |
| 04 | NVIDIA Kubernetes Engine (NVKE) and NVIDIA Kubernetes Recipes (NVKR) are in public preview as open source building blocks for managed Kubernetes on GPU infrastructure, with signed, version-pinned images. | Reported |
| 05 | NVIDIA Snapshot restores checkpointed inference replicas without model reload or warmup. The post reports one partner reducing vLLM startup from between two and 8.5 minutes to six and 34 seconds on HGX B300 systems. | Reported |
reported by the sourcemeasured or run by usnot verified
Primary source: NVIDIA blog: NVIDIA Expands Open Source Commitment With OpenShell Donation and New AI Infrastructure. We are the fast, independent index; the vendor’s own page is the source of record for this release.
Why the donation matters
Policy enforcement for agents has mostly lived inside individual vendors' products. Putting OpenShell under neutral CNCF governance makes it easier for platform teams to adopt one runtime across clouds without depending on a single vendor's roadmap, though a Sandbox project is the CNCF's earliest maturity stage, not a graduated standard.
The Snapshot figures are the most concrete numbers in the announcement, but they come from design partners chosen by NVIDIA on NVIDIA hardware. Teams running inference on Kubernetes should treat them as best-case examples until independent results appear.
In the source’s words
“an open source runtime for securing and governing autonomous agents”
“OpenShell gives organizations control over what AI agents can do.”
What we have not verified
- We did not run OpenShell, NVKE, NVKR or Snapshot ourselves.
- The adoption figures (hundreds of organizations, tens of thousands of contributions) are NVIDIA's own claims; we did not check them against the project's repository.
- The Snapshot startup reductions are partner-reported results published by NVIDIA; we did not reproduce them or confirm which hardware other than HGX B300 is supported.
- We did not confirm the OpenShell acceptance on the CNCF's own Sandbox project list.
This brief is Reported: written from the sources below, not from running the release ourselves.
Sources
- NVIDIA blog: NVIDIA Expands Open Source Commitment With OpenShell Donation and New AI Infrastructureprimary source · fetched 2026-11-11
Related
Spotted an error? See our corrections policy or email the address on the contact page. Markdown version for agents: /news/nvidia-openshell-cncf-donation-kubernetes-snapshot.md
Want these briefs in your Google Top Stories? Add Agentic Weekly as a preferred source on Google.
Newsletter
The Weekly Verdict. Free, every Monday.
A free email every Monday: the week's few big AI releases, then 5 short use cases for the new models, each with a prompt you can paste.
Free. Double opt-in. One click to unsubscribe.