I've documented over $310,000 USD in savings through FinOps and operational redesign, while holding 99.95% uptime for financial institutions. The pattern is predictable: when something breaks, the instinctive response is to scale infrastructure, because it's politically easier than saying the problem is in the code, the process, or the design.
Adding compute makes the symptom disappear from the dashboard, but it raises the cost and leaves the root cause intact. Real FinOps starts by measuring capacity against demand and finding where the waste is — before spending a single peso more on instances.