Negotiating Cloud Commitment Deals: Azure, AWS and GCP
The hyperscalers want one thing above all from a large customer: a big, multi year spend commitment. Get the size and the shape of that commitment right and you save a great deal. Get it wrong and you have prepaid for cloud you will never burn. Here is how these deals are actually built, where the discount curve really bends, and where the genuine negotiating room sits.
A cloud commitment deal is not really a discount scheme. It is a bet. You agree to spend a minimum amount with one provider over a fixed term, and in return they give you a better rate. The provider is betting that you will commit to more than you would otherwise have spent, and that some of it will go unused. You are betting that you can forecast your own consumption well enough to claim the discount without stranding money. Most of the value, and most of the risk, lives in that single number you commit to. So that is where the work belongs, not in haggling the rate card.
This guide is vendor neutral. The three big vehicles differ in the detail, but the underlying mechanics, and the mistakes buyers make, are remarkably similar across all of them.
How the three big vehicles actually work
The names differ, the shape is the same: commit to spend, unlock a better rate, accept some lock in.
Microsoft Enterprise Agreement and MACC
Microsoft's enterprise customers typically sit under an Enterprise Agreement, increasingly framed around a Microsoft Azure Consumption Commitment. You commit to a total Azure spend over the term, and your discounting, your incentives and a good deal of field attention scale with that commitment. Microsoft also leans heavily on its wider estate here, because your Azure number is rarely negotiated in isolation from your Microsoft 365 and other licensing. That bundling is a lever for them and, handled well, a lever for you.
AWS Enterprise Discount Program
AWS uses the Enterprise Discount Program, an EDP, where you commit to a minimum spend over a multi year term in exchange for a platform wide percentage discount. It sits on top of, not instead of, the service level commitments such as Savings Plans and Reserved Instances, which is a point many buyers miss and which changes how the maths stacks.
Google committed use and committed spend
Google offers committed use discounts at the service level, and larger customers negotiate a committed spend agreement across the platform. The committed use discounts come in resource based and spend based forms, and the broader agreement works much like the others: a multi year total commitment against a better rate and stronger support.
In every case the headline rate is the easy part. The number that decides whether the deal is good or bad is the commitment itself: how large, how long, and how rigidly it is structured. That is the negotiation. The rate is downstream of it.
Where the discount curve really bends
Discounts scale with commitment, but not in a straight line. The curve is steep at first and flattens fast. Moving from no commitment to a meaningful one unlocks the bulk of the benefit. Stretching the commitment higher to chase the next increment of percentage often costs you far more in over commitment risk than it returns in rate. The provider knows exactly where that curve sits for your profile. You usually do not, which is the asymmetry at the heart of the deal.
Term length pulls the same way. A three year commitment prices better than one year, but it locks your floor in place across a period in which your architecture, your workloads and even your provider preference may change. The longer term is not free. You are paying for the better rate with flexibility you may want back.
The over commitment trap, and why they steer you into it
The single most expensive mistake in these deals is committing to a number you cannot consume. If you fall short of your commitment, you generally still owe it. You have prepaid for cloud you never used, and the effective rate on what you did use can end up worse than if you had committed to nothing at all.
This is not an accident of the model, it is the point of it for the provider. The field is incentivised to land the largest defensible commitment, because your commitment is their booked, forecastable revenue. A confident growth story, a generous read of your migration plans, a nudge to round the number up for a slightly better tier, all of it pushes in the same direction. None of it is dishonest. It is simply that their interest is a bigger number and yours is the right number, and those are not the same.
The commitment a provider proposes is almost always anchored to the optimistic end of your own forecast, because that is the number that hits their target. Treat the first proposed commitment the way you would treat any opening position, as the start of the conversation, not the answer.
How the reps are measured, and why it matters to you
Hyperscaler field teams are not measured the way a traditional licensing rep is. Their world is consumption and commitment. They carry targets for committed spend and for actual consumption growth, and their year is shaped by the provider's own fiscal calendar. That has two practical consequences for you.
First, timing is leverage, just as it is anywhere else. A field team that needs to land your commitment before a period close has a reason to move that you can use. Second, and less obvious, they are often as motivated by consumption as by the headline commitment, because growth in what you actually run is what compounds for them. That gives you room to trade: flexibility on how and where you draw down your commitment can be worth more to you than a slightly larger headline discount, and it does not always cost them much to give.
The partner ecosystem matters here too. Marketplace purchases can often be drawn down against your commitment, which means software you were buying anyway can help you reach your number. Used deliberately, that turns a risk, the unused commitment, into a way to consolidate spend you already have.
Where the genuine negotiating room is
The rate card gets the attention. The real movement is usually elsewhere:
- The commitment number. Right sizing this against an honest consumption forecast, not an aspirational one, is the highest value decision in the whole deal.
- The ramp. A commitment that steps up over the term, rather than landing at full height on day one, protects you while your migration or growth actually materialises.
- Draw down flexibility. How freely your commitment can be spent across services, regions and through the marketplace decides how easily you reach your number without distorting your architecture.
- Term and exit. One year versus three, and what happens at renewal, including any true forward mechanics if you exceed your commitment.
- Private pricing and incentives. Above the standard program, large commitments attract bespoke pricing and migration funding that never appear on a public page.
How to avoid locking into a number you cannot burn
The discipline is simple to state and hard to hold under deadline pressure. Build your own consumption forecast before the provider builds it for you, and build it honestly, including the workloads that might not move and the projects that might slip. Commit to the number you are confident you will consume, not the number that unlocks the prettiest percentage. Prefer a ramped commitment and shorter term when your forecast is uncertain, and pay for the better rate only when your consumption is genuinely predictable. And model the downside: work out what the deal costs you if you hit only seventy or eighty percent of your commitment, because that is the scenario the rate card never shows you.
This is the same principle that runs through all of our acquisition work, set out in the pillar guide on how vendors actually build a quote. The first number is an opening position, the value is won by understanding how the deal is constructed, and the buyer who forecasts their own position negotiates from strength. If you are also weighing where your workloads should run in the first place, our guide on cloud repatriation and hybrid covers the consumption side of that decision.
Send us your cloud commitment or renewal
Facing an EA, an EDP or a committed spend agreement? Send us the proposal and your consumption picture and we will give you an independent view: whether the commitment is right sized, where the real room is, and what a sensible number and shape look like. Independent, with no provider to push. We have sat on the other side of these deals.
Prefer email? Reach us directly at hello@c4cgroup.co.uk.
Frequently asked questions
What is the difference between a Microsoft EA, an AWS EDP and a Google committed spend deal?
They are the same idea in different clothing. Each asks you to commit to a minimum spend over a multi year term in exchange for a better rate and stronger support. Microsoft frames it around an Azure consumption commitment, often tied to your wider Microsoft estate. AWS uses the Enterprise Discount Program as a platform wide layer on top of Savings Plans and Reserved Instances. Google uses committed use discounts at the service level and a committed spend agreement across the platform. The mechanics and the pitfalls are very similar across all three.
How big a discount can a cloud commitment really get me?
It depends on the size and term of the commitment and on your profile, so any single number would be misleading. The more useful insight is the shape of the curve: most of the benefit comes from making a meaningful commitment at all, and it flattens quickly after that. Stretching the commitment higher to chase the next slice of percentage usually costs more in over commitment risk than it returns in rate.
What happens if I do not use all of my committed spend?
In most structures you still owe the commitment. You have effectively prepaid for capacity you did not consume, and the effective rate on what you did use can end up worse than committing to less would have been. This is the single most expensive mistake in these deals, and it is exactly the outcome the provider's incentives quietly push you toward, so the commitment number deserves more scrutiny than the rate.
Should I commit for one year or three?
A longer term prices better but locks your floor in place across a period when your architecture, workloads and even provider preference may change. If your consumption is genuinely predictable, the longer term can be worth it. If it is uncertain, a shorter term or a ramped commitment that steps up over time usually protects you for far more than the rate difference costs.
Where is the real negotiating room in a cloud commitment?
Not in the headline rate. It is in right sizing the commitment to an honest forecast, in a ramp that matches when your spend actually arrives, in how freely you can draw the commitment down across services and the marketplace, and in the bespoke private pricing and migration funding that large commitments attract but that never appear on a public page.
Can I negotiate cloud commitments across more than one provider?
Yes, and genuine multi cloud consumption can be a source of leverage, because credible alternative spend is exactly what concentrates a provider's attention. The caution is to make sure the architecture justifies it rather than spreading commitments so thin that you reach none of your numbers. The commitment should follow where the workloads genuinely run.