EDP commitments ran 15 to 25 percent above realized consumption

The EDP discount runs 5 to 20 percent. The typical overcommitment ran 15 to 25 percent. Put those two numbers on the same page and the default EDP is net negative: the discount is real, and the unconsumed commitment it was traded for is larger. The negotiation is the sizing, not the rate.

Prepared by Redress Compliance · August 14, 2026 · AWS advisory. EDP and Savings Plan reviews across enterprise estates, 2024 to 2026.

• Size the commitment to trailing twelve month consumption, because the growth forecast produced the 15 to 25 percent overshoot and the telemetry did not.
• Trade commitment depth for rate only from an honest baseline, since the discount curve rewards commitments the forecast inflates for free.
• Negotiate the marketplace credit rates and route eligible third party spend through them, so the commitment fills from spend you were making anyway.
• Hold the term at three years unless the rate movement genuinely pays for the extra forecast risk.
• Write the early termination clause at signature: defined notice, pro rated commitment relief, and treat the renewal cycle as the moment all of this reopens.
• Date every proposal and time the ask, because EDP and PPA offers are a mix of credits and discounts whose composition changes quarter to quarter, and the same request lands differently across AWS's own calendar.

Strip the branding and an EDP is a simple trade: AWS buys certainty about your spend, and pays for it with a discount. The question every buyer should ask is whether the price AWS paid, 5 to 20 percent, covers what the certainty cost you. Across our reviews, mostly it did not, and the reason is structural.

The commitment is a firm number and the consumption is an uncertain one, so someone has to absorb the gap between them. The contract assigns that risk entirely to the buyer: consume less than committed and you pay anyway, consume more and AWS simply bills it. There is no scenario where the gap falls on the publisher, which is what makes the sizing conversation the entire negotiation.

And the sizing conversation is where the forecast walks in. Commitments built on growth forecasts ran 15 to 25 percent above realized consumption, and the Savings Plans beneath them repeated the pattern at 14 to 28 percent. The forecast is not a lie; it is the seller's optimism and the buyer's ambition agreeing with each other, and the agreement between them is billed to you. The trailing twelve month file is the only party in the room without a motive.

The marketplace credit is the underused counterweight. At 50 percent general credit and 100 percent for strategic partners, third party software the estate already buys can flow through the marketplace and fill commitment the workloads alone would miss. It converts the commitment from a bet on your infrastructure growth into a routing decision about spend that already exists, which is a far safer thing to promise.

The exit clause completes the frame. An EDP with defined notice and pro rated relief is an agreement you can afford to be wrong about; the default EDP is not, and the difference between those two documents costs nothing at signature and everything after. Ask for the exit while AWS still wants the signature. That is the only moment it is cheap.

The wider cloud commit picture, including the same forecast mechanics at the other hyperscalers, sits in the AWS overspending report and the AWS practice.

Across the EDP and Savings Plan reviews we ran, the pattern held at both layers of the commitment stack:

Three patterns recurred. Commitments sized in the growth planning meeting rather than from the billing data. Marketplace credit left unnegotiated and unrouted, so existing third party spend filled nothing. And exit clauses missing entirely, so the overcommitment had no relief valve when the forecast met reality.

The buyer side move is to let the billing history write the commitment. The wider library sits in the AWS practice.

What discount does an AWS EDP deliver?

Typically 5 percent for smaller commitments rising to about 20 percent at the top commitment thresholds and longer terms. The discount steps at commitment thresholds rather than scaling smoothly, and the overall EDP or PPA offer is a mix of credits and discounts whose composition changes quarter to quarter, which is why the commitment number matters more than any single percentage.

How should an AWS EDP commitment be sized?

To trailing consumption, not to growth forecasts. Typical buyer commitments were sized 15 to 25 percent above realized consumption, and Compute Savings Plans were over committed against trailing twelve month consumption by 14 to 28 percent on most estates. An unconsumed commitment is spend no discount recovers.

Do AWS Marketplace purchases count toward an EDP commitment?

Yes, at a credit rate: typically 50 percent for general marketplace transactions and 100 percent for specific strategic marketplace partners. That makes the marketplace the flexible valve in the commitment, letting third party software spend help fill an EDP that consumption alone would miss.

What is the right AWS EDP term length?

Typically three years, which balances the discount rate against optionality. Longer terms deepen the rate but extend the forecast risk, and the forecast is already the weakest number in the agreement.

Can an AWS EDP be terminated early?

Only on the terms you negotiated at signature: defined notice and pro rated commitment relief, subject to the early termination clauses in the EDP contract. The buyer side discipline is to negotiate that protection as part of the framework, before it is needed.

Is the default AWS EDP structure buyer friendly?

No. The default structure is materially in the publisher's favor: the commitment is firm, the consumption is uncertain, and the discount is the only compensation. The buyer side job is to construct the EDP as a commercial framework that preserves optionality without sacrificing the rate.

When is the best time to negotiate an EDP?

At the renewal cycle, with a trailing consumption file in hand. Renewal is when the commitment resets, the credit rates are negotiable, and the exit clauses can be written in, and the trailing file is what stops the next commitment from repeating the 15 to 25 percent overshoot.