The Pulse
NVIDIA contributes OpenShell to CNCF
NVIDIA is contributing its OpenShell agent runtime to the Cloud Native Computing Foundation as a Sandbox project. It also announced Kubernetes tools, including two projects in public preview and a component NVIDIA says can restore inference replicas in seconds.

OpenShell contribution
NVIDIA is contributing OpenShell, its runtime for securing and governing autonomous AI agents, to the Cloud Native Computing Foundation (CNCF). The contribution was announced on Nov. 10 alongside other Kubernetes infrastructure work at KubeCon + CloudNativeCon.
The contribution is designated as a CNCF Sandbox project, which NVIDIA says will be governed by organizations building OpenShell into their products and strengthened by security professionals. The CNCF is a nonprofit part of the Linux Foundation and hosts KubeCon.
Controls and agent workloads
OpenShell lets organizations set deterministic controls on agents as they use tools, access inference services and run workloads. Its architecture can connect to an organization’s compute, inference providers, credential storage and monitoring tools, allowing governance across deployment environments.
Hundreds of organizations have used OpenShell and the project has received tens of thousands of contributions in less than a year since launch, NVIDIA said. NVIDIA is also contributing to Agent Substrate, a Google-led open-source framework for running long-lived agent workloads on Kubernetes that has been accepted into the CNCF Sandbox.
Agent Substrate preserves agent environments between tasks while releasing compute when agents are idle. Suspending and resuming environments across a shared pool lets teams scale workloads without keeping dedicated compute running for every session.
NVIDIA and Google are collaborating on an integration of OpenShell and Agent Substrate, which NVIDIA says is intended to secure and govern agents across clouds and on-premises systems. The effort also connects agent execution with NVIDIA Dynamo for model inference and KAI Scheduler for shared cluster resource allocation.
Infrastructure projects
NVIDIA also announced three components of its DSX AI Factory Platform: NVIDIA Kubernetes Engine, NVIDIA Kubernetes Recipes and NVIDIA Snapshot. NVIDIA described the projects as open-source tools to help teams deploy, manage and scale GPU infrastructure on Kubernetes.
NVIDIA Kubernetes Engine and NVIDIA Kubernetes Recipes are now in public preview, offering building blocks for managed Kubernetes services on GPU infrastructure. They automate cluster lifecycle management and provide signed, version-pinned images.
NVIDIA said Mirantis, Netris, OpenNebula Systems, Rafay, Spectro Cloud, vCluster, Firmus and SK Telecom have aligned with it on integration plans. The company said the plans would bring the two projects’ capabilities into their offerings.
NVIDIA Snapshot is an open-source modular component that NVIDIA says restores checkpointed replicas in seconds without model reload or warmup. NVIDIA said it is intended to support faster startup, recovery and elasticity for inference and agent workloads on Kubernetes.
Partner evaluations
Design partners Crusoe and Saturn Cloud evaluated Snapshot on their workloads using NVIDIA HGX B300 systems. Crusoe, an upstream contributor to Snapshot, tested GLM-5.3 on Crusoe Cloud and reported that readiness time fell from roughly eight minutes to 31 seconds.
Saturn Cloud, which NVIDIA describes as a token factory platform, benchmarked Snapshot for scale-from-zero inference, with NVIDIA reporting that vLLM startup times fell from between two and 8.5 minutes to six and 34 seconds.