White Paper · Cloud (AWS · Azure · GCP)

Cloud Contract Framework 2026

How AWS, Microsoft Azure, and Google Cloud enterprise commitments are engineered to maximise lock-in and spend — and the buyer-side tactics that turn committed-use discounts, egress, and renewal timing back into leverage.

By Atonement Licensing Advisory Former AWS, Azure & Google Cloud commercial executives Published Jan 2026 · Updated June 2026 ≈ 17 min read

Executive Summary

Enterprise cloud has quietly become the largest and least-governed line in most IT budgets. Unlike a perpetual software licence, a cloud commitment is a forward bet on consumption: the customer trades a multi-year minimum spend for a discount, and the vendor's entire commercial model is built to ensure that bet is sized in the vendor's favour. AWS Enterprise Discount Programmes (EDPs), Microsoft Azure commitments under the MCA and enterprise agreements, and Google Cloud committed-use discounts (CUDs) all share the same underlying mechanic — and the same negotiable seams.

This framework distils what former hyperscaler commercial practitioners know about how cloud agreements are structured and where buyers consistently leave money on the table: oversized commitments, ignored egress and support costs, private pricing left unrequested, and renewals negotiated under deadline pressure. The central message for buyers is simple: a cloud commitment should be sized to a credible consumption floor, never to the discount the vendor offers for committing more, and the contract terms around the discount matter more than the headline percentage itself.

15–40%
Incremental discount available on large committed-use deals beyond the public rate
20–35%
Typical share of committed spend that goes unconsumed on oversized deals
5–15%
Of a mature cloud bill is egress & inter-region transfer — most of it recoverable
180 days
Runway needed before renewal to negotiate from strength, not deadline

1. How Cloud Commercial Models Are Designed

A cloud enterprise agreement is not a licence — it is a consumption commitment wrapped in a discount. The customer agrees to spend a minimum amount over a term (typically one to five years), and in exchange the provider applies a discount to on-demand rates. Three design choices do most of the commercial work. First, the discount is tiered to the size of the commitment, which creates a constant gravitational pull toward committing more than the business can credibly consume. Second, the commitment is usually a floor, not a budget: unconsumed commitment is generally forfeited, so an oversized deal is pure margin for the vendor. Third, the agreement layers multiple billing constructs — on-demand, reserved instances or savings plans, committed-use discounts, marketplace spend, and support tiers — each with its own rules, so the effective rate a customer actually pays is far harder to audit than a simple percentage off list.

The practical consequence is that the gap between what a customer commits and what they can productively use is where most cloud overspend lives. Buyers who treat the vendor's recommended commitment as a technical sizing exercise concede the most value; buyers who treat it as a commercial proposal to be tested against their own consumption forecast keep it. Every hyperscaler's account team is compensated on committed and consumed spend growth — the recommended number is an opening position, never a neutral estimate.

Insider note

The "use it or lose it" nature of most commitments means an EDP or CUD sized 25% above real consumption does not save 25% — it can cost more than paying on-demand, because the forfeited commitment is spend with zero workload behind it. Always model the break-even consumption point before signing.

2. AWS Enterprise Discount Programmes and Savings Plans

The AWS Enterprise Discount Programme is a private, negotiated agreement under which a customer commits to a minimum spend over a term in return for a platform-wide percentage discount. Layered beneath it sit Savings Plans and Reserved Instances, which discount specific compute families in exchange for one- or three-year commitments. The two layers interact: an EDP discount typically applies on top of the rates a customer is already paying, including Savings Plan rates, so the effective discount stacks. The risk is that the EDP minimum is set against an optimistic growth curve, and Savings Plans are over-purchased against a baseline that later shrinks.

The table below shows how an EDP commitment behaves at different consumption outcomes against an illustrative three-year, $30M commitment with a 15% discount.

Table 1 — How an oversized AWS EDP commitment erodes its own discount
Scenario (3-yr, $30M commit, 15% discount)Actual consumptionEffective cost vs on-demandNet outcome
Consumption meets commitment$30M−15%Full discount realised