Karpenter makes it easier for Kubernetes clusters to scale nodes based on workload demand, but many teams are unsure what to expect once they enable it in real environments.

This talk explains how Karpenter fits into Kubernetes autoscaling at a high level and what problems it is designed to solve. It walks through common expectations, typical behaviors teams observe after adoption, and early pitfalls such as unexpected scaling, cost surprises, and disruption during scale-down events.

Designed for engineers new to Karpenter or Kubernetes autoscaling, the session focuses on building intuition rather than configuration details, helping attendees understand when Karpenter helps, when it does not, and what questions to ask before using it in production.