Skip to content
STIRUP

TECHNOLOGY INTELLIGENCE

Understand what's next.

NVIDIA Contributes OpenShell to CNCF for Agent Security

NVIDIA is contributing its OpenShell runtime to the Cloud Native Computing Foundation (CNCF) to provide a shared foundation for securing and governing autonomous agents across deployment environments.

Share
AI-generated conceptual illustration accompanying: NVIDIA Contributes OpenShell to CNCF for Agent Security
AI-generated editorial illustration of the article topic; not a photograph, product screenshot or measured result.

NVIDIA's OpenShell Contribution to CNCF

NVIDIA is contributing its OpenShell runtime to the Cloud Native Computing Foundation (CNCF), a nonprofit under the Linux Foundation that hosts KubeCon. OpenShell is an open source tool designed to secure and govern autonomous agents, enabling developers to build, inspect, and improve security infrastructure collaboratively. According to NVIDIA, this contribution provides a shared foundation for developers to shape the evolution of agent security. Organizations can use OpenShell to apply governance across various deployment environments while integrating it with existing enterprise systems. Additionally, NVIDIA and Google are collaborating on an integration of OpenShell with Agent Substrate, allowing enterprises to secure and govern agents across cloud and on-premises systems.

Collaboration on Agent Substrate and Integration with NVIDIA Tools

NVIDIA is expanding its open source contributions by participating in the development of Agent Substrate, a framework initiated by Google and now hosted in the CNCF Sandbox. This collaboration aims to enhance the framework’s compatibility with NVIDIA tools such as Dynamo for model inference and KAI Scheduler for managing shared cluster resources. Additionally, NVIDIA and Google are working together to integrate OpenShell with Agent Substrate, enabling enterprises to secure and govern agents across both cloud and on-premises environments. These efforts reflect NVIDIA’s commitment to fostering open collaboration in the development of scalable agent workloads on Kubernetes. The integration is expected to provide organizations with greater flexibility and control over their AI and machine learning operations. The contributions and integrations are being developed through joint efforts between NVIDIA and Google, as noted in their respective statements.

New Open Source Projects for Kubernetes and AI Workloads

NVIDIA has expanded its open source commitment with the introduction of three new projects aimed at enhancing Kubernetes and AI workloads. These projects extend the DSX AI Factory Platform, making Kubernetes easier to run and inference faster to scale. The NVIDIA Kubernetes Engine (NVKE) and NVIDIA Kubernetes Recipes (NVKR) are now available in public preview, offering open source building blocks for delivering managed Kubernetes services on GPU infrastructure. Additionally, several companies, including Mirantis, Netris, OpenNebula Systems, Rafay, Spectro Cloud, and vCluster, along with Firmus and SK Telecom, have aligned with NVIDIA on integration plans to bring NVKE and NVKR capabilities into their offerings. An open source, modular component is also available, which restores checkpointed replicas in seconds without requiring model reload or warmup, supporting faster startup, recovery, and elasticity for inference and agentic workloads on Kubernetes.

Performance Improvements with Snapshot and Partner Evaluations

NVIDIA has demonstrated performance improvements through Snapshot, a tool contributed to the open source community. Design partners Crusoe and Saturn Cloud have evaluated Snapshot on their own workloads using NVIDIA HGX B300 systems. Crusoe, an upstream contributor to Snapshot, tested GLM-5.3 on Crusoe Cloud and reduced the time for the workload to become ready from roughly eight minutes to 31 seconds. Saturn Cloud, a token factory platform, benchmarked Snapshot for scale-from-zero inference and reduced vLLM startup times from between two and 8.5 minutes to six and 34 seconds.

Sources and reporting notes

Based on published source material. No independent hands-on testing is claimed.

Read more