I grew up in the high desert of the American Southwest, where my grandmother hauled water by hand to a ranch with no running line, and every drop was accounted for. That upbringing shapes how I see AI’s fastest-growing cost today: a resource footprint that’s already pushing Kubernetes clusters to run AI workloads at single-digit GPU utilization, water and energy spent on capacity nobody is using.

Grounded in CNCF’s own data and TAG Environmental Sustainability and Cloud Native AI Working Group’s draft framework currently under community review, we’ll uncover where that waste lives and walk through simple, measurable actions to take starting this week, using open source tools from the CNCF landscape likely already included in your stack.

This talk is an invitation, not a warning. Attendees will leave knowing exactly where to find resource waste, what to measure, and first steps to running more sustainable workloads. We built Kubernetes together. We can run it sustainably together too.