The Default Price
When you first spin up a virtual machine (an EC2 instance) in the cloud, you are typically using the On-Demand pricing model.
This is the most flexible model. You pay only for the seconds or hours you use the resource, and you can turn it off at any time with no further obligation. It's like staying in a hotel and paying the rack rate night-by-night — convenient, but expensive.
The Problem
While flexible, On-Demand is also the most expensive way to buy cloud capacity.
For workloads that run 24/7 — like your production databases or core application servers — paying the on-demand rate is financially irresponsible. It's like living in that hotel for a full year but still paying the nightly rate instead of signing a lease. The hotel analogy breaks down quickly: cloud discounts for committed usage can reach 70–90%, meaning the financial stakes are enormous at scale.
A mature cloud financial strategy involves moving away from On-Demand and using financial instruments offered by the cloud provider to trade a little bit of flexibility for a significant discount.
The three main levers we can pull are:
- Commitments — Promise to use a certain amount of compute for 1 or 3 years in exchange for a discount.
- Spare capacity — Use idle cloud infrastructure at steep discounts, accepting the risk of interruption.
- Negotiated agreements — Enterprise-level contracts that apply discounts across your entire cloud bill.
Most organizations run a blended strategy: Savings Plans or RIs cover the predictable baseline, Spot covers flexible batch work, and On-Demand handles unpredictable spikes. The goal is to minimize the percentage of your bill that is pure On-Demand.