NVIDIA expands open source AI infrastructure push with OpenShell donation to CNCF

Traders Union
All News
Business & Wealth
Companies
NVIDIA pushes open AI via CNCF

Highlights

NVIDIA donates OpenShell, an open source runtime for securing autonomous AI agents, to the CNCF Sandbox, facilitating broader community governance and adoption.
NVIDIA introduces three new open source components—NVKE, NVKR, and Snapshot—for its DSX AI Factory Platform, aiming to accelerate managed Kubernetes deployment and inference elasticity on GPU systems.
Crusoe and Saturn Cloud, using NVIDIA HGX B300 systems, reduced workload startup times from eight minutes to 31 seconds and two to 8.5 minutes to six to 34 seconds respectively with NVIDIA Snapshot.

Open source tools for agent security and Kubernetes operations

According to NVIDIA Blog, the company is contributing NVIDIA OpenShell, an open source runtime for securing and governing autonomous agents, to the Cloud Native Computing Foundation, where it enters the CNCF Sandbox. The move places the project under community governance and is intended to broaden development input from product builders and security professionals.

OpenShell is designed to give organizations tighter control over what AI agents can do by applying deterministic controls to their actions as they use tools, access inference services and execute workloads. NVIDIA says its extensible architecture lets enterprises connect their own compute resources, inference providers, credential storage and monitoring systems, allowing governance policies to work across different deployment environments.

NVIDIA says hundreds of organizations are already using OpenShell, and that the project has received tens of thousands of contributions in less than a year since launch. Jim Zemlin, CEO of the Linux Foundation, says the contribution gives developers a shared base for building agent security infrastructure as cloud-native applications evolve.

NVIDIA is also contributing to Agent Substrate, a Google-led open source framework for running long-lived agent workloads at scale on Kubernetes that has also been accepted into the CNCF Sandbox. The company says it is aligning the framework with NVIDIA Dynamo for model inference and KAI Scheduler for shared cluster resource allocation, while also working with Google on integrating Agent Substrate with OpenShell across cloud and on-premises environments.

New DSX platform components target faster scaling

NVIDIA is also announcing three new open source components for its DSX AI Factory Platform, NVIDIA Kubernetes Engine, NVIDIA Kubernetes Recipes and NVIDIA Snapshot. The additions are aimed at helping AI infrastructure operators deliver managed Kubernetes services on GPU systems while improving deployment speed and inference elasticity.

NVIDIA Kubernetes Engine, or NVKE, and NVIDIA Kubernetes Recipes, or NVKR, are now in public preview. NVIDIA says the projects provide building blocks for managed Kubernetes services by automating cluster lifecycle management and supplying signed, version-pinned images.

The launch includes ecosystem support from Mirantis, Netris, OpenNebula Systems, Rafay, Spectro Cloud and vCluster, alongside Firmus and SK Telecom, which NVIDIA says are working on integration plans for the new capabilities. That support could extend the tools across a broader set of AI infrastructure platforms and cloud services.

NVIDIA Snapshot is intended to reduce delays in bringing additional inference capacity online by restoring checkpointed replicas in seconds without reloading models or warming up workloads. The company says the modular component supports faster startup, recovery and elasticity for inference and agentic workloads running on Kubernetes.

Design partners Crusoe and Saturn Cloud have evaluated Snapshot on their own workloads using NVIDIA HGX B300 systems. According to NVIDIA, Crusoe reduced the time for a GLM-5.3 workload on Crusoe Cloud to become ready from roughly eight minutes to 31 seconds, while Saturn Cloud cut vLLM startup times for scale-from-zero inference from between two and 8.5 minutes to six and 34 seconds.

As a reminder, our earlier report on the Cohere AISLE acquisition detailed plans to add AI-driven software vulnerability detection for enterprise security.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.