AWS committed spend discount.
A committed-spend discount is the single largest contractual reduction most enterprises hold with AWS — and the one most often sized wrong. This note sets out the EDP discount tiers by commitment size, how the discount stacks on top of Savings Plans and Reserved Instances, and the shortfall risk that quietly inflates cost when the commitment overshoots real consumption.
AWS committed-spend discounts under an EDP or private pricing agreement typically run 5–23% off on-demand, scaling with commitment size: ~$1M/yr earns 5–8%, ~$10M earns 10–15%, and $50M+ reaches 17–23%, layered multiplicatively on top of Savings Plans and Reserved Instances. The risk lives entirely in the floor — falling short of the commitment does not reduce what you owe. Size to a conservative forecast, reshape the ramp, and the discount sizes itself correctly.
01 Key findings
The commitment is a floor, not a cap. You agree to spend at least the committed amount; overage keeps earning the discount, but a shortfall is pure loss with no discount benefit. All the risk sits in over-committing.
Discount scales with commitment size, term and competitive tension. AWS publishes no EDP rate card. A credible multi-cloud alternative moves the discount more than any single contractual argument.
The layers are multiplicative, not additive. A 50% Savings Plan inside a 13% EDP pays ~0.5 × 0.87 ≈ 43.5% of on-demand. Treating them as additive overstates the saving and tempts over-commitment.
The ramp is where shortfalls originate. A front-loaded ramp commits you to high spend before migration catches up. A back-loaded ramp holds the same total and discount while tracking realistic adoption.
Marketplace spend is an underused lever. Routing third-party software through AWS Marketplace can retire commitment — but not all transactions count at full value, so confirm the treatment in writing.
02 What committed spend means
A committed-spend discount is a contractual agreement to spend a minimum dollar amount on AWS over a defined term — usually one to five years — in return for a flat percentage discount on eligible usage. AWS packages this as the Enterprise Discount Program (EDP) for larger customers. The commitment is a floor, not a cap: you agree to spend at least the committed amount, and if you spend more, the discount continues to apply. The risk lives entirely in the floor, because falling short does not reduce what you owe.
The discount applies broadly across compute, storage, database, networking and most managed services. It does not stack on top of every line; some marketplace purchases and third-party charges are excluded or count toward the commitment at a reduced rate. Understanding what counts toward the commitment, and what receives the discount, is the core of the negotiation.
03 EDP discount tiers by commitment size
These bands reflect negotiated outcomes, not a published rate card — AWS does not publish EDP discounts. The discount is a function of commitment size, term length, growth story, and competitive pressure from Azure or Google Cloud in the account.
| Annual commitment | Typical discount | Common term | Negotiation note |
|---|---|---|---|
| $1M to $5M | 5 to 8 percent | 1 to 3 years | Entry EDP, limited room |
| $5M to $10M | 8 to 12 percent | 3 years | Ramp terms become negotiable |
| $10M to $25M | 10 to 15 percent | 3 to 4 years | Credits and PPA add-ons available |
| $25M to $50M | 14 to 19 percent | 3 to 5 years | Service-specific discounts layer in |
| $50M+ | 17 to 23 percent | 4 to 5 years | Bespoke, executive-sponsored |
04 Discount depth by commitment
Discount deepens with scale, but with diminishing returns: the jump from entry EDP to a $10M commit is far larger than the jump from $25M to $50M+. Midpoint blanket discount at each tier:
05 How it stacks with Savings Plans and RIs
The committed-spend discount and resource-level commitments are different layers, and they combine. Savings Plans and Reserved Instances reduce the rate on compute; the EDP discount then applies a further flat percentage on the already-reduced bill. The order matters for modelling: a workload on a 3-year Savings Plan at 50% off on-demand, inside a $20M EDP at 13%, pays roughly 0.5 × 0.87 — about 43.5% of on-demand. Treating the layers as additive overstates the saving and leads to over-commitment. See our AWS Savings Plans guide and Reserved Instances guide for the compute layer, and the EDP pillar for the full stack.
The most expensive EDP mistake is over-committing on an aggressive growth forecast. If committed annual spend is $15M and actual eligible spend lands at $11M, you still owe the $15M — the $4M shortfall is pure loss with no discount benefit. Negotiate a back-loaded ramp that matches the commitment curve to realistic adoption, a shortfall true-forward that lets unmet commitment roll into a later year, and exclusion clarity so the marketplace and third-party spend you expected to count actually counts.
06 Marketplace and what counts toward commitment
AWS Marketplace purchases can count toward EDP commitment, which is a powerful and underused lever: routing third-party software through Marketplace lets that spend retire the AWS commitment instead of sitting outside it. Not all Marketplace transactions count at full value, and the rules differ by contract, so confirm the treatment in writing before assuming a large software purchase will offset the commitment. Our AWS Marketplace procurement strategy covers how to structure this.
Three mechanics govern what happens when actual spend diverges from the commitment, and each is negotiable. Overage is the good case: the discount continues to apply to the excess, and a well-negotiated EDP confirms that explicitly rather than capping the discount at the commitment. Shortfall is the costly case, and the default treatment is that you owe the difference — a true-forward or rollover clause softens this by letting unmet commitment carry into a later year. The third mechanic is the annual reconciliation, which sets when AWS measures spend against commitment; aligning it to your fiscal calendar avoids surprises at the wrong moment.
07 Multi-year ramp structures
The commitment is rarely a flat annual number. AWS structures most EDPs as a ramp, where committed spend rises year over year to match expected growth. A front-loaded ramp commits you to high spend early, before migration has caught up, and is where shortfalls originate. A back-loaded ramp keeps early commitments below realistic consumption and pushes the larger numbers into years when usage will plausibly support them. Because the discount is usually set against the total commitment, you can often hold the headline discount while reshaping the ramp to reduce shortfall risk.
| Year | Front-loaded commit | Back-loaded commit | Realistic spend |
|---|---|---|---|
| Year 1 | $12M | $7M | $8M |
| Year 2 | $14M | $12M | $13M |
| Year 3 | $16M | $23M | $22M |
| 3-year total | $42M | $42M | $43M |
Both ramps total $42M and earn the same discount, but the front-loaded version overshoots realistic spend by $4M in year one while the back-loaded version tracks it closely. Matching the ramp to the adoption curve is the difference between a clean three-year run and an early shortfall that the buyer eats.
08 Sizing the commitment
The AWS account team builds the proposed commitment from your trailing usage, stated growth plans and visible migration pipeline, then sizes it to capture as much future spend as the discount can justify. The incentive is asymmetric — AWS benefits from a larger commitment because shortfalls accrue to you. Rebuild the number from your own conservative model. A commit-sizing framework holds four factors:
Eligible spend base
Model on eligible, discounted spend — not gross AWS spend. Marketplace, third-party and certain managed services may count at reduced value or not at all, so a gross forecast overshoots the base.
Conservative adoption curve
Forecast on the migration that will actually deliver, not the optimistic curve the cloud team wants. The gap between the two is exactly the shortfall risk you would absorb.
Ramp shape
Reshape early-year commitments to sit below early-year spend. The discount is set against the total, so a back-loaded ramp is close to free risk reduction.
Co-term and inventory
Bring existing RIs, Savings Plans and any prior EDP into one aligned renewal. A single annual renewal concentrates spend and alternatives into one high-leverage conversation.
09 Our recommendation
The discount follows the commitment, so size the commitment to a conservative consumption forecast — not the optimistic one AWS will encourage — and negotiate the ramp so early-year commitments stay below early-year spend.
Stack the EDP discount on the already-reduced Savings Plan and RI bill. Build the model on eligible, discounted, conservatively forecast spend and the commitment sizes itself correctly.
The discount moves on competitive tension more than any other factor, and end-of-quarter or end-of-year deals run deeper. A credible Azure or GCP option is worth several points.
Commit to the right number, not the vendor's
Our Cloud & FinOps practice models your committed-spend tier against real consumption and negotiates the discount curve, ramp and shortfall terms — a median 14% recovery versus the first-quoted commit.
The Licensing Edge
Weekly cloud and licensing intelligence for enterprise IT leaders. 3,000+ subscribers.