Nvidia donates OpenShell agent runtime to CNCF and previews GPU Kubernetes tooling
Nvidia has donated OpenShell, an open-source runtime that applies policy controls to autonomous AI agents, to the Cloud Native Computing Foundation. At KubeCon North America it also put new GPU Kubernetes tools and a fast inference-restore feature into public preview.
Nvidia announced on 10 November that OpenShell, its open-source runtime for securing and overseeing autonomous AI agents, is moving into the Cloud Native Computing Foundation as a Sandbox project. The announcement came at KubeCon + CloudNativeCon North America.
OpenShell applies fixed policy controls to what an agent can do, covering tool use, inference requests and workload execution. Teams can plug in their own compute, inference providers, credential stores and monitoring. Nvidia says hundreds of organisations already use it and that it has drawn tens of thousands of contributions in under a year.
Nvidia is also contributing to Agent Substrate, a Google-led CNCF Sandbox framework for running long-lived agent workloads on Kubernetes, and says the two companies are working to integrate it with OpenShell.
On the infrastructure side, Nvidia put NVIDIA Kubernetes Engine and Kubernetes Recipes, open-source tools for managed Kubernetes on GPU clusters, into public preview, alongside NVIDIA Snapshot, which restores checkpointed inference replicas in seconds without reloading a model. Nvidia cites design partners: Crusoe cut readiness time for a GLM-5.3 deployment from about eight minutes to 31 seconds, while Saturn Cloud reduced vLLM start-up from several minutes to under 35 seconds.
Linux Foundation chief executive Jim Zemlin said the contribution gives developers a shared foundation to build on. The performance figures are Nvidia's own and have not been independently verified.
Why it matters
Agent governance is shifting from vendor-specific features to neutral, community-run infrastructure. Placing the runtime under the CNCF makes it easier for enterprises wary of single-vendor lock-in to adopt.
Faster replica restore matters commercially: it lets operators scale inference up and down with demand rather than keeping expensive GPUs idle.
Indicops view
Indian government and enterprise teams piloting agents should ask a basic question of every platform: where are tool permissions and credentials enforced? A policy runtime built on Kubernetes standards is a sensible candidate to evaluate, especially for regulated workloads that must run on sovereign or on-premise clusters.
More in Open Source
Back to the newsroomHugging Face's chief vows to fight on for open source as Nvidia's $12.9 billion bid faces uncertainty
After Jensen Huang's exit as Nvidia CEO, Hugging Face chief Clement Delangue said the company will keep fighting for open-source AI harder than ever. Nvidia's planned $12.9 billion purchase of the platform is scheduled to close in the first half of 2027 and has drawn US Justice Department questions.
Baidu releases ERNIE 5.5 Flash Base, a 280-billion-parameter open-weight model with a restrictive licence
Baidu has published ERNIE 5.5 Flash Base, a 280B-parameter mixture-of-experts checkpoint with 14B active per token, trained on 33.5 trillion tokens. Weights are downloadable, but the licence excludes the EU, UK and South Korea and limits commercial use above $50 million revenue.
Ricursive releases Ripple-65B Preview, an Apache-licensed open-weight model built for agentic coding
Ricursive Intelligence has released Ripple-65B Preview, a 65-billion-parameter dense open-weight model pretrained for agentic coding. The company claims a top self-reported Terminal-Bench 2.1 score among US open-weight models, though the figures have not been independently reproduced and some reruns narrow the gap.