Karpenter Handles the Scaling. Here’s What It Leaves on the Table

2026/09/17 23:00 GMT

Karpenter gives teams a faster, more flexible way to scale nodes. But node autoscaling is only part of the optimization story.

Join Justin Garrison and the Cast AI team for a live discussion on what Karpenter does well, where cost and performance still leak through, and how to close the gap without ripping out the autoscaler you already run.

Why Attend?

Karpenter has earned its place as the default for Kubernetes node autoscaling – and if you run it, you already know why. But in enterprise environments, teams keep hitting the same walls, and none of them are configuration problems:

Nodes are provisioned perfectly, yet cluster utilization stays low – because pods request far more than they use, and no node autoscaler can fix requests
Consolidation gets switched off (or heavily restricted) because stateful and long-running workloads can’t survive the disruption
Spot adoption stalls after the first few interruptions hit production, so clusters quietly drift back to on-demand
Costs are visible at the node level, but nobody can say which team or workload the money actually goes to
Multi-cloud setups end up with different tooling and logic per cluster – Karpenter here, cluster autoscaler there

In this session, we’ll dig into why these limits exist, where the boundaries of node autoscaling really are, and what enterprise teams are doing above that layer to close the gap – without touching their Karpenter setup or changing how nodes get provisioned.

🎁 BONUS: All participants will have the chance to win exclusive Cast AI swag and AirPods Pro 3!

What You’ll Learn

Where Karpenter stops and workload-level optimization begins
How over-requested CPU and memory silently undermine even well-tuned autoscaling
Why node consolidation often fails without safe container migration
How Spot interruption prediction changes the risk/reward math
How Cast AI applies the same optimization engine across Karpenter and standard cluster autoscaler environments

Live Discussion Format

No slides-only walkthrough here. Justin Garrison will bring his latest thinking on Kubernetes autoscaling and the Cast AI team will respond with real-world data, architecture decisions, and product-level detail.

Panelists

Justin Garrison

Cloud Native Influencer, Author, and Practitioner

Steven Fraser

Staff Solutions Architect, Cast AI

Victor Viana

Senior Sales Engineer, Cast AI

Who Should Attend

DevOps & SREs running Karpenter or cluster autoscaler on Kubernetes
Platform Engineers responsible for cluster cost and performance
Cloud Engineers managing Spot instances, node consolidation, or rightsizing

Name

This field is for validation purposes and should be left unchanged.

Register for the webinar

First name(Required)

Last name(Required)

Work email(Required)

Company name(Required)

Job title(Required)

By submitting this form, you acknowledge and agree that Cast AI will process your personal information in accordance with the Privacy Policy.