Submission Review — Emerging & Advanced Tech Track Chair Kubecon Cloud Native Con North America 2025

With every conference, especially larger ones like KubeCon or Open Source Summit, there is an opportunity, if you have been around for some time in the community, to review conference proposals. For some tracks, this is especially interesting since everyone usually tries to come up with the "best new thing" to talk about. Reviewing these proposals is quite interesting, and sharing some of the trends and technologies mentioned will surely be interesting for others too.

The submissions reviewed for this article were part of the "Emerging and Advanced Technologies" track for KubeCon + CloudNativeCon North America 2025. This track represents a smaller, subset of the total submissions, usually about 10% of all talks submitted. So this does not represent the entire conference, there are dedicated tracks for Platform Engineering, AI, Security, etc. All submitted proposals undergo a multi-stage review process as described at length here.

Themes - Emerging and Advanced Technologies Track

Based on the talks submitted, a few themes have been identified. Most talks have a clear scope that falls into one category, but for some, it can be counted as two or three if multiple categories are connected within a submission.

Kubernetes and the ecosystem around it are still the main motivators for most submissions. This is to be expected at the conference. AI follows closely as the next major category, with talks focusing on MCP, Agent Networks, GPU, and LLM hosting. The other areas are kinda stable. The heatmap below shows where each

List - Tech Cherry Picking

A selection of some of the projects mentioned.

• Kueue is a Kubernetes-native job queueing controller that orchestrates batch, HPC, and AI/ML workloads, providing an essential mechanism for managing resource quotas to ensure execution in environments with constrained or fluctuating resources, such as limited hardware or variable energy availability (renewable energy…).

• MCP on Kubernetes using KAITO (Kubernetes AI Toolchain Operator) is an operator that automates the deployment of AI/ML workloads on Kubernetes by managing GPU provisioning and model configuration, with the ability to interoperate with standards like the Model Context Protocol (MCP).

• Numaflow is a Kubernetes-native platform for executing massively parallel real-time data and event processing workloads, often used as a foundational component for MLOps.

• Kubestellar provides another concept or solution to multi-cluster management, utilizing a logical control plane to deliver, single-cluster API experience with a non-invasive, pull-based architecture.

• Hyperlight is a lightweight, embeddable Virtual Machine Manager (VMM) for executing untrusted code in secure micro-VMs, which adopts a novel, WebAssembly-inspired approach requiring applications to be recompiled with its libraries to attain Wasm-like security and performance characteristics.

• Kagent transforms Kubernetes management by providing an agentic AI framework where agents can independently plan and execute operational tasks based on declarative config.