💡 WHY RILEY IS RIGHT
The expensive line is rarely the compute itself; it is the control plane, the node pool sized for headroom, the managed add-ons, and the engineer hours spent upgrading all of it. When a cluster exists to run one scheduled job, replacing it with cron on a small VM removes a bill, a pager rotation and an upgrade calendar in one afternoon. Wins like that need no business case and no migration plan, and they hide in every organization that adopted orchestration before it had workloads.