The 2026 edition of KubeCon + CloudNativeCon North America returned to Salt Lake City for a four-day program beginning November 9, bringing together users and vendors of Kubernetes, Prometheus, Envoy, and dozens of other CNCF-graduated projects. Registration pages and talk abstracts began appearing on the CNCF site this week.

Red Hat's confirmed speaking sessions illustrate the current technical priorities: Jooho Lee and Johnu George will present "Beyond Model Serving: Running Production LLMs on Kubernetes with KServe," covering infrastructure for deploying large-language-model workloads inside managed clusters. Sohan Kunkerkar and Ayato Tokubi will discuss extending and accelerating CRI-O — Red Hat's lightweight Kubernetes container runtime — with new plug-in hooks designed to reduce startup latency for short-lived workloads including AI agent containers.

The CNCF itself is using the conference to advance its AI Conformance Program, which certifies cloud platforms for interoperable AI workload deployment. Google Cloud Platform (GKE) has already earned certification as an AI-conformant platform, and the program plans to expand scope later this year to include Sovereign AI standards addressing sandboxing and data residency requirements.

For enterprise cloud-native practitioners, the emphasis on model-serving infrastructure and runtime efficiency indicates that Kubernetes adoption is shifting past general-purpose microservices into production AI inference and training pipelines — a move that carries cost and reliability implications for organizations investing in GPU-accelerated clusters.

Source: Event details from the Linux Foundation / KubeCon portal and Red Hat event page.