AWS Enterprise Agreement and EDP: the 2026 buyer reference.
AWS EDP discount tiers run 5 to 25 percent on net AWS spend, with ramping annual commits, 3-year and 5-year terms, and exit penalties that can exceed 75 percent of the remaining unused commit. Private Pricing Agreements add another 5 to 15 percent on specific services; Marketplace purchases burn down commit at full value with no discount. This note documents the full EDP commercial surface.
An EDP converts AWS's variable public price list into a contracted private price in exchange for a multi-year commit. Discount scales with commit and term — 5 to 25 percent, peaking at $50M+ over five years. The value is real, but the risk is in the fine print: ramping commits, Marketplace burn-down at 0 percent discount, and exit penalties above 75 percent of unused commit. Size the commit to your own forecast at 80–90 percent, negotiate a step-down clause, and lock PPA layers in the original deal.
01 Key findings
EDP is the AWS "Enterprise Agreement" in all but name. AWS does not officially use the term, but functionally the EDP plays Microsoft's EA role: a multi-year commit to a defined annual spend in exchange for a percentage discount on most services, with documented exclusions.
Discount correlates with commit and term, not negotiation flair. Bands run 5–8 percent at $1–3M up to 22–27 percent at $50M+ over five years. AWS does not publish the table; above-tier rates exist only for competitive-win or strategic-service concessions.
The ramp is where money is forfeited. Aggressive 50 percent year-on-year ramps that outrun actual adoption leave commit stranded, and AWS will not reduce a committed amount mid-term without major concession.
Marketplace burns down commit at full value with zero EDP discount. A $1M Marketplace purchase costs the full $1M and forfeits the ~$200K a direct AWS spend would have saved at a 20 percent EDP rate.
Exit penalties are the most punitive, most overlooked clause. Early termination can require paying the full unused remaining commit — $33M on a Year-2 exit from a 5-year $10M ramping EDP. A step-down clause is the only realistic mitigation.
02 EDP discount tiers
An EDP is a multi-year commitment to spend a defined annual amount on AWS in exchange for a percentage discount, signed separately from the standard AWS Customer Agreement. Discount percentages correlate with annual commit and term length. AWS does not publish the table; the bands below reflect typical negotiated outcomes in advisor-led deals during 2024 to 2026.
| Annual commit | 3-year EDP discount | 5-year EDP discount | Add-on PPA potential |
|---|---|---|---|
| $1M to $3M | 5 to 8 percent | 7 to 10 percent | 0 to 5 percent |
| $3M to $10M | 8 to 12 percent | 10 to 15 percent | 3 to 8 percent |
| $10M to $25M | 12 to 17 percent | 15 to 20 percent | 5 to 10 percent |
| $25M to $50M | 15 to 20 percent | 18 to 23 percent | 8 to 13 percent |
| $50M+ | 18 to 22 percent | 22 to 27 percent | 10 to 15 percent |
Tiers are not strict thresholds. AWS will negotiate above-tier discounts for strategic customers — a competitive win against Azure or GCP, an unusually high public profile, or willingness to concede on AWS-favoured services (Bedrock, SageMaker, AWS-published Marketplace software). Sub-$1M buyers can secure a 1-year EDP, but at a minimal 3 to 5 percent.
03 Discount depth by commit
The relationship between committed spend and peak achievable discount is the single most important input to the commit-sizing decision. Bars show the top of the 5-year EDP band at each commit tier — the maximum discount a well-run negotiation reaches before PPA layers.
EDP applies to net AWS spend after Reserved Instances and Savings Plans. A 20 percent EDP layered on a 45 percent Compute Savings Plan yields roughly 56 percent on covered compute (20 percent applied to the 55 percent net-of-SP rate), not 65 percent. PPA layers compound the same way — a 20 percent EDP plus a 10 percent PPA is 28 percent, not 30. Model every stacked discount as a product, never a sum.
04 Ramping commits and multi-year terms
Most enterprise EDPs use a ramping commit that grows year over year. A 5-year EDP at a $5M Year 1 commit might ramp to $7M, $9M, $11M and $13M, total commitment $45M. The ramp accommodates expected growth and lets AWS lock the multi-year commitment without forcing Year 5 dollars in Year 1. EDP terms are 3-year or 5-year by default; 3-year is the typical enterprise sweet spot, balancing discount realisation against commitment risk.
Ramps create two risks. The ramp acceleration trap: AWS sometimes proposes 50 percent year-on-year growth that actual adoption does not support, forfeiting commit in later years. The ramp negotiation lag: a customer who reaches Year 4 and realises Year 5 is unreachable has limited remediation, because AWS will not cut the committed amount mid-term without major concession.
Moving from a 3-year to a 5-year EDP typically adds 4 to 6 percent discount on annual commit — worth $2.0M to $3.0M cumulative over five years at $10M annual commit. The cost is two extra years of exposure to AWS pricing changes, service deprecations and business change. Stable AWS trajectories favour 5-year; uncertain futures or active multi-cloud strategies favour 3-year to preserve optionality at a known cost.
05 The overcommit and exit trap
Exit penalties are the EDP's most punitive clause and the most commonly overlooked at signing. The standard clause requires paying the unused remaining commit at full value on early termination. A customer in Year 2 of a 5-year $10M ramping EDP who terminates owes the cumulative unfunded commit of $33M, plus any contractual penalties — a figure that can exceed 75 percent of the remaining commitment.
Mitigations exist but require explicit language. AWS will sometimes agree to a step-down clause allowing a 25 to 40 percent reduction in annual commit under specific conditions (M&A divestiture, business-unit closure, demonstrated AWS service issues). A true exit clause is rare and expensive. Buyers should negotiate the step-down as the primary mitigation and treat the EDP as a binding multi-year obligation, sizing the commit accordingly.
The costliest EDPs are signed on Year 1 forecasts built around AWS sales adoption goals rather than the customer's own model. Commit at 80 to 90 percent of an internally built forecast, not 100 percent of AWS's. Without a step-down clause, an over-optimistic ramp becomes stranded commit with no remediation short of paying it out.
06 PPA and Marketplace burn-down
Private Pricing Agreements (PPAs) are service-specific discount layers added on top of an EDP where a customer has unusually high consumption or AWS wants to drive strategic-service adoption. Common PPAs cover EC2 compute (5 to 12 percent additional), S3 (3 to 8 percent), DynamoDB (5 to 10 percent) and Bedrock (8 to 15 percent in 2026 as AWS pushes AI). PPAs stack multiplicatively: EDP discount applies to the public price first, then PPA to the EDP-discounted price.
AWS Marketplace purchases burn down commit at full value but receive no EDP or PPA discount. A $10M commit customer buying $3M in Marketplace software consumes $3M of commit yet pays full price — an apparent benefit that hides real cost. The table shows how $1M is treated by spend type at a 20 percent EDP rate.
| $1M spend type | EDP discount applied | Net cost | Commit burndown |
|---|---|---|---|
| Direct AWS service (EC2, S3, RDS) | 20 percent | $800,000 | $800,000 (post-discount) |
| AWS Marketplace software | 0 percent | $1,000,000 | $1,000,000 |
| AWS-published service (Bedrock, SageMaker) | 20 to 28 percent (with PPA) | $720,000 to $800,000 | Variable |
Discipline: consolidate software on direct AWS services where possible, and negotiate Marketplace-specific discount lanes (5 to 15 percent off for named vendors) where Marketplace consumption is unavoidable. Note too that Enterprise Support — mandatory for EDP customers — is charged on top of EDP discount, budgeted at 3 to 5 percent of total AWS spend at enterprise scale.
07 Commit-sizing framework
Four inputs govern how much to commit and on what terms. Weight them to your situation before signing — most EDPs that fail in operation were mis-sized at signature.
Forecast independence
Build an internal AWS consumption forecast independent of AWS sales input, then commit at 80 to 90 percent of it. AWS projections are built around adoption goals that may not materialise.
Ramp realism
Match the ramp to actual adoption, not to a 50 percent year-on-year curve. An over-steep ramp strands commit in later years with no mid-term remediation.
Term length
Five years adds 4 to 6 percent discount — $2.0M to $3.0M over five years at $10M — against two more years of exposure. Stable trajectories take it; uncertain ones keep 3-year optionality.
Exit protection
Negotiate a 25 to 40 percent step-down clause and the PPA layers in the original deal. Both are far cheaper to secure with commitment leverage than after signature.
08 Our recommendation
Active multi-cloud strategy or an unclear AWS trajectory. Preserve optionality at a known cost — the 4 to 6 percent forgone discount — and revisit at renewal with a fresh baseline.
A predictable, growing AWS estate. Capture the 22 to 27 percent top band, worth $2.0M to $3.0M cumulative at $10M commit — but only with a step-down clause protecting the back years.
Compute Savings Plans (54 to 66 percent) compound with EDP and flex across families, regions and OS. Worth the small discount gap versus 3-year all-upfront Reserved Instances for anything but steady-state workloads.
09 Negotiation sequencing
The buyer-side process runs four phases: baseline (build the independent forecast and identify PPA-worthy services), competitive engagement (get written Azure and GCP proposals for the same workloads), AWS negotiation (open 20 to 30 percent below AWS's expected discount ask), and contract close (verify final language on exit clauses, ramp formulas and PPA mechanics before signing). Within the AWS negotiation, one process choice drives outcomes more than any other:
Parallel Recommended
Negotiate discount, ramp, PPA layers, Marketplace treatment, step-down clause, support cost and renewal terms together. Every lever stays live, and concessions can be traded across the whole deal while commitment leverage is highest.
Sequential Weaker
Settle the headline discount first, then raise Marketplace, PPA and exit terms afterwards. Each is negotiated with the commitment position already surrendered — the pattern that produces the worst realised rates.
For the broader framework see AWS EDP negotiation, AWS Savings Plans, AWS vs Azure vs GCP pricing, the cloud contracts guide, cloud renewal strategy, multi-cloud strategy and the AWS vendor hub. For an active engagement, see cloud contract negotiation or software licensing advisory.
Negotiate your AWS EDP with buyer-side discipline
Our Cloud & FinOps practice sizes the commit, structures the ramp, and closes the exit, Marketplace and PPA terms buyer-side teams routinely miss.
The Licensing Edge
Weekly cloud and licensing intelligence for enterprise IT leaders. 3,000+ subscribers.