One signed file installs and upgrades the cluster, and AI agents work with it within limits the operator sets
CIQ, the enterprise software company behind Rocky Linux, today introduced CIQ Stratum Kubernetes Engine (SKE), open source Kubernetes for the AI era. Kubernetes runs in production at 82% of organizations that use containers, and the Cloud Native Computing Foundation (CNCF) calls it the de facto operating system for AI. Running it still means assembling about ten upstream projects that each release on their own schedule. SKE ships them as one signed file that installs, joins, and upgrades the cluster.
Built from an unmodified upstream source by the team behind Rocky Linux, Fuzzball, Warewulf, and Apptainer, CIQ Stratum Kubernetes Engine (SKE) lets system administrators, developers, and platform engineers run Kubernetes next to their existing Linux, HPC, and GPU infrastructure.
AI raises the cost of that upkeep. Training and inference spread across many systems, and capacity grows and shrinks with demand. Two thirds of organizations that host generative AI models run some or all inference on Kubernetes, according to the CNCF January 2026 survey. Those clusters sit beside air-gapped sites, compliance rules, and general-purpose nodes, often across several Linux distributions. In a 2026 industry survey of 785 Kubernetes decision makers, 62% said complexity was slowing business initiatives, and 65% ran air-gapped clusters. Teams want speed without vendor lock-in or the cost of integrating every piece themselves.
Also Read: CIO Influence Interview with John Elliott, Cybersecurity Author Fellow at Pluralsight
A built-in MCP server lets AI agents work with the cluster under rules the operator sets. Its 37 tools come in read, write, and destructive tiers, and tools above the enabled tier are never registered. Kubernetes access controls still bound every call, and secrets stay off limits. An AGENTS.md file guides agent-driven installs, and the documentation is published in llms.txt format. Operators get a built-in web dashboard with pods, logs, and shell access.
The hard part of Kubernetes starts after the first install, when those projects drift apart. CIQ Stratum Kubernetes Engine (SKE) pins and tests every component as a set. Rolling upgrades check node health at each step and roll back a node that fails. etcd snapshots run every 12 hours by default, and certificates renew before they expire. Every release is signed and ships a software bill of materials. What a team learns on its first SKE cluster carries to the tenth.
CIQ Stratum Kubernetes Engine (SKE) is free and open source and runs a single cluster with an included web UI. CIQ Stratum Pro, the commercial version, manages every SKE cluster, from GPU clusters to the general fleet, through the same web interface under one sign-on. Stratum Pro adds an app catalog, an optional CIS hardening profile with a per-control check, and support from CIQ. Its fleet agent connects outbound only, so clusters need no inbound ports. Cluster API provisioning follows in a later release.
“Kubernetes has become the place AI runs, and the people running it still spend their weeks on upgrades and integration work. We built CIQ Stratum Kubernetes Engine (SKE) so every cluster installs and upgrades the same way, and so the agents those teams put to work can only do what the operator allows,” said Gregory Kurtzer, CEO of CIQ and founder of Rocky Linux, Apptainer, and Warewulf.
“Anyone who has stood up a cluster by hand knows the install sets the tone for everything after it, so we built CIQ Stratum Kubernetes Engine (SKE) around getting that step right. Build something on your laptop that works, and the same binary runs it on a cloud VM. What is easy for you to set up is just as easy for your agents to deploy, prototype, and test with,” said Peter Nelson, Chief Technology Officer, CIQ.
CIQ Stratum Kubernetes Engine (SKE) runs on any Enterprise Linux 9 or later distribution, so teams keep the operating system they already trust and avoid a vendor’s own node OS. That includes Rocky Linux, RLC Pro AI, and RLC Pro Hardened, on amd64 and arm64. It works alongside Warewulf Pro for node provisioning, Fuzzball for HPC and AI workflows, and Ascender Pro for automation, and each product also runs on its own.
Catch more CIO Insights: How Are CIOs Aligning Technology with Workforce Agility?
[To share your insights with us, please write to psen@itechseries.com ]
