Negotiating with AWS: How Enterprise Teams Cut Cloud Costs Before They Compound

AWS is one of the most under-negotiated enterprise vendors in the market.

That sounds surprising. AWS is often one of the largest technology suppliers in an enterprise portfolio, with annual spend measured in millions, or tens of millions. Yet many organizations approach AWS as if the only available levers are architectural optimization, Savings Plans, and Reserved Instances.

Those tools matter. They are not the whole negotiation.

AWS spend grows organically through consumption rather than seats. New accounts appear. Workloads move into production. Data volumes increase. Development environments multiply. Business units adopt services independently. By the time finance sees the full picture, the organization may already have a substantial AWS run rate, and limited visibility into the commitments supporting it.

That is precisely when a large cloud bill becomes a negotiation opportunity.

The goal is not simply to obtain a larger headline discount. It is to build a commercial structure that aligns with actual consumption, protects flexibility, reduces commitment risk, and preserves leverage for the next renewal.

Why AWS negotiations work differently from SaaS negotiations

Seat-based SaaS contracts usually have a visible commercial unit: users, licenses, modules, or transactions. AWS is different. The bill is a portfolio of variable consumption lines.

AWS pricing can include:

• On-Demand usage
• Savings Plans
• Reserved Instances
• Data transfer and egress
• Storage and requests
• Managed database consumption
• Support fees
• AWS Marketplace purchases
• Promotional credits
• Migration funding
• Enterprise-level private pricing

That creates two common mistakes.

First, teams benchmark the wrong number. They compare the total invoice to public list price without normalizing usage, regions, service mix, pricing model, and credits.

Second, teams negotiate the wrong commitment. They accept a spend target based on an optimistic cloud roadmap rather than a defensible forecast of eligible consumption.

In a seat-based negotiation, unused licenses are visible. In AWS, unused commitment can be buried inside a complicated consumption model and discovered only after the organization has already signed.

The AWS commercial landscape: what can actually be negotiated?

On-Demand pricing

On-Demand is the most flexible AWS pricing model, but generally the least economical for stable workloads. It is also the baseline against which many other discounts are presented.

On-Demand pricing is useful as a reference point, not as a complete benchmark. The real question is:

What effective rate is the enterprise paying for its actual service mix after all discounts, credits, commitments, and exclusions?

Savings Plans

Savings Plans provide discounted rates in exchange for a one- or three-year commitment to a consistent level of eligible compute spend.

Depending on the type, they may apply across services such as:

• Amazon EC2
• AWS Fargate
• AWS Lambda
• Selected database workloads

The tradeoff is flexibility versus discount depth. A Compute Savings Plan can provide broader portability across eligible compute, while an EC2 Instance Savings Plan may offer stronger economics with more constrained coverage.

AWS provides public documentation on how Savings Plans work, but the discount is not automatically a substitute for enterprise negotiation. Savings Plans typically do not address:

• Data transfer out
• AWS Support fees
• Most Marketplace software fees
• Broader enterprise commitment risk
• Contractual protections around future services

Review the official AWS Savings Plans documentation before modeling a proposal.

Reserved Instances

Reserved Instances can reduce the cost of predictable EC2 and database usage, usually in exchange for a defined configuration and term.

They can be attractive when workloads are stable. They can also create avoidable lock-in when a business is changing instance families, regions, operating systems, or deployment patterns.

The right question is not “What is the maximum discount?” It is:

How much of this workload can the business confidently call stable for the full commitment term?

AWS explains the mechanics of Reserved Instances here. Your negotiation should go further by testing whether the commitment is operationally realistic.

EDPs and Private Pricing Agreements

Large AWS customers may negotiate an Enterprise Discount Program, often referred to in current market discussions as a Private Pricing Agreement or PPA.

The basic structure is straightforward:

1. The customer commits to a minimum level of eligible AWS consumption.
2. AWS provides a negotiated discount across defined eligible usage.
3. The agreement runs for a set term, commonly one to five years.
4. The customer must manage its usage carefully to avoid falling short.

AWS does not publish a universal public discount schedule for EDPs or PPAs. Rates depend on factors such as:

• Annual committed spend
• Total term value
• Current run rate
• Forecast growth
• Service mix
• Region and account structure
• Competitive alternatives
• Migration or modernization plans
• AWS Marketplace participation
• Timing and internal AWS priorities

Industry-observed ranges are directional rather than official. As a rough negotiating reference, enterprise proposals may fall into bands such as:

Annual commitment	Directional observed discount range*
$1M–$5M	5%–8%
$5M–$10M	8%–12%
$10M–$25M	10%–15%
$25M–$50M	14%–19%
$50M+	17%–23%

*These are market observations, not AWS-published tiers or guaranteed outcomes. A discount is meaningful only when you understand exactly which services, charges, accounts, and commitment calculations it covers.

Do not negotiate on the headline percentage alone. Ask AWS to define:

• Eligible services
• Excluded services
• Treatment of new AWS services
• Treatment of Marketplace spend
• Treatment of support charges
• Credit application order
• Account and region portability
• Ramp mechanics
• Shortfall consequences
• True-up and true-forward provisions
• Renewal and repricing rights

The growth commitment trap

The most dangerous AWS commitment is not necessarily the one with the lowest discount. It is the one based on a forecast that the business cannot reliably achieve.

AWS teams may present a larger commitment as a way to unlock better economics. The logic can be compelling:

• The business is migrating workloads.
• AI and analytics usage will grow.
• New regions are planned.
• More business units will consolidate onto AWS.
• A modernization program will increase consumption.

Some of those assumptions may be correct. The problem is that a forecast is not the same as committed usage.

If actual consumption falls below the commitment, several outcomes are possible depending on the agreement:

• The customer receives less economic value than expected.
• Unused commitment may expire.
• The customer may need to accelerate migrations or shift workloads.
• Finance may face a true-up or shortfall obligation.
• The customer may have less leverage at renewal because it appears to have underperformed against its promise.

A strong AWS negotiation separates three numbers:

1. Current run rate: what the company is consuming today.
2. Committed base: what the company can defend with high confidence.
3. Strategic upside: what may happen if planned migrations and growth materialize.

Only the first two should form the foundation of a hard commitment. Strategic upside should be rewarded through ramps, expansion rights, reopener clauses, or incremental discounts, not automatically converted into a fixed liability.

Committed spend, true-ups, and unused commitment liability

Before signing, model the agreement month by month.

At a minimum, calculate:

• Expected AWS usage by service
• Eligible versus excluded spend
• Existing Savings Plans and Reserved Instances
• Marketplace spend
• Credits and expiration dates
• Support fees
• Migration-related costs
• Seasonality
• Business-unit allocations
• Downside and upside scenarios

Then test three cases:

Base case

The organization follows its current migration and growth plan.

Downside case

Migration is delayed, a major workload remains on-premises, or optimization reduces consumption.

Upside case

AI, data, or application workloads grow faster than expected.

The agreement should not be economically attractive only in the upside case.

Ask for mechanisms that reduce unused commitment liability, including:

• A lower first-year commitment
• Annual ramps rather than a flat commitment
• Carry-forward rights
• Reallocation across accounts and regions
• Broader eligible-service coverage
• The ability to apply certain Marketplace purchases
• A formal review if consumption falls materially
• No automatic conversion of a shortfall into a new long-term commitment
• Clear treatment of credits when calculating commitment attainment

Use AWS Marketplace, credits, and MAP as negotiation levers

AWS Marketplace can be commercially important in two ways.

First, Marketplace may be part of the broader enterprise cloud commitment. If your organization already purchases significant third-party software through AWS Marketplace, ask whether that spend can contribute to commitment attainment. Do not assume it qualifies. Get the treatment documented.

Second, Marketplace can provide a procurement path for third-party software, including negotiated private offers. AWS explains the mechanics of Marketplace private offers.

Credits are another lever, but they should never be treated as equivalent to a permanent price reduction.

Separate:

• AWS promotional credits
• Migration credits
• Marketplace-related credits
• Partner-funded incentives
• MAP funding
• Service-specific discounts
• Contractual PPA discounts

For migration programs, AWS’s Migration Acceleration Program terms should be reviewed carefully. MAP may support migration and modernization activities through funding or credits, but eligibility, included services, timing, and exclusions matter. For example, credits may not cover all data transfer costs or third-party software charges.

Negotiate the value of credits as part of the total economic package, while keeping the underlying price and commitment structure transparent.

Data egress: negotiate it before you need it

Data egress is often treated as an unavoidable technical cost. In a negotiation, it is also a portability and concentration-risk issue.

Your AWS analysis should distinguish between:

• Routine operational egress
• Cross-region replication
• Internet delivery
• Data movement during migration
• Exit-related data transfer

The EU Data Act is changing the discussion. The European Commission states that switching charges, including switching-related data egress, must be eliminated from 12 January 2027, with a transitional cost-based approach before then. The official European Commission explanation of the Data Act provides the relevant context.

That does not mean all ordinary AWS outbound data transfer becomes free. Day-to-day operational egress remains a separate commercial issue.

For a multinational enterprise, negotiate:

• A clear definition of switching-related egress
• Data export assistance
• Migration support obligations
• Advance notice requirements
• Portability of data and metadata
• Treatment of egress outside the EU
• No contractual language that defeats applicable regulatory rights

Support fees are negotiable too

AWS Enterprise Support is generally calculated as a percentage of monthly AWS charges, subject to minimum fees and tiered rates. At large spend levels, even a modest reduction in the effective support rate can produce meaningful savings.

Ask AWS to clarify:

• The exact fee base
• Whether credits reduce the support calculation
• Whether Marketplace charges are included
• Whether support fees increase automatically as usage grows
• Minimum monthly charges
• Enterprise Support versus other support tiers
• Technical Account Manager coverage
• Service-level commitments
• Response-time remedies

Do not accept support as an automatic percentage of every dollar on the invoice without testing the calculation. Support is a distinct commercial line and should be benchmarked separately. Review AWS’s current Support pricing directly, then negotiate the enterprise-specific application.

How to benchmark AWS spend properly

You cannot benchmark AWS effectively by asking whether you received “20% off list price.”

That number may exclude the services where your spend is concentrated. It may also combine different pricing mechanisms and obscure the cost of commitment risk.

A useful AWS benchmark normalizes:

• Service and SKU
• Region
• Usage volume
• Instance family
• Operating system
• Pricing model
• On-Demand equivalent
• Effective unit rate
• Discount off On-Demand
• Savings Plan or Reserved Instance coverage
• PPA/EDP discount
• Credits
• Support fees
• Data transfer
• Marketplace charges
• Term length and commitment

Track at least three measures:

Effective discount rate

The total discount against the relevant On-Demand baseline.

Effective unit rate

The actual price paid per unit of compute, storage, request, transfer, or database capacity.

Net economic rate

The cost after credits, funding, support, and commitment liabilities are included.

This is where Right Price Benchmarking™ adds value. Benchmarking should reveal not just whether the discount looks attractive, but whether the total structure is commercially fair.

The AWS negotiation playbook

1. Start early

Begin at least six to nine months before a major renewal, PPA, or commitment decision. AWS has more flexibility when the customer has time to create alternatives and does not appear trapped by an expiration date.

2. Build an internal fact base

Bring together:

• CIO or CTO
• CFO or finance leader
• Procurement
• Cloud economics or FinOps
• Enterprise architecture
• Legal
• Business-unit owners

Agree on the target architecture, migration timetable, acceptable commitment, walk-away position, and priority terms before engaging AWS.

3. Build competitive pressure

Azure and Google Cloud do not need to replace AWS entirely to create leverage. A credible alternative may include:

• A workload migration
• A regional deployment
• A data and analytics platform
• A disaster recovery environment
• A new AI workload
• A dual-cloud architecture
• A competitive benchmark

The objective is not to threaten a migration the company cannot execute. It is to demonstrate that AWS is competing for future workload allocation.

4. Use timing deliberately

AWS’s fiscal calendar, quarter-end priorities, renewal windows, migration milestones, and internal account planning cycles can affect responsiveness.

Do not wait until AWS presents its final proposal. Establish your economic requirements early, then create a structured process with multiple proposal rounds.

5. Negotiate the complete package

Ask for more than a discount:

• Better PPA or EDP economics
• A defensible ramp
• Carry-forward rights
• Broader eligible spend
• Marketplace treatment
• Credits and MAP funding
• Reduced support fees
• Egress protections
• New-service coverage
• Renewal caps
• Exit and portability language
• Governance and reporting rights

AWS contract traps to review

Before signing, specifically test for:

• Auto-renewal: Does the agreement renew automatically, and how much notice is required to prevent it?
• Term escalators: Can the commitment or support fee increase during the term?
• Service-level carve-outs: Are critical services excluded from meaningful remedies?
• True-forward language: Can a shortfall or added usage create a higher future commitment?
• Repricing risk: Can AWS reprice new services or move them outside the negotiated discount?
• Credit expiration: Do credits expire before planned workloads go live?
• Account restrictions: Can spend move between accounts, regions, and subsidiaries?
• Marketplace exclusions: Does Marketplace spend count toward the commitment?
• Support calculation: Are support fees calculated on gross or discounted usage?
• Change-in-control language: What happens after a merger, divestiture, or acquisition?
• Exit costs: What data, assistance, and egress obligations apply when workloads move?

Key takeaways for CIOs, CFOs, and procurement teams

• AWS is not simply a usage bill. It is a portfolio of pricing models, commitments, credits, support charges, and contractual risks.
• The largest discount is not always the best deal if it requires an unrealistic commitment.
• EDPs and PPAs should be negotiated around eligible spend, flexibility, shortfall protection, and future-service treatment, not just the headline percentage.
• Savings Plans and Reserved Instances optimize specific workloads. They do not replace enterprise-level negotiation.
• Marketplace, MAP, credits, support fees, and egress can materially change the total economics.
• Benchmark effective unit rates and net cost, not just “discount off list.”
• Competitive pressure from Azure and Google Cloud is useful only when it is credible.
• Contract language determines whether savings survive after the signature.

Frequently asked questions

How much discount can an enterprise get from AWS?

There is no universal public AWS discount table. Directional market ranges may vary from single digits for smaller commitments to the high teens or low twenties for very large, well-structured commitments. The actual result depends on spend, term, service mix, growth, competition, and negotiation timing.

Should we commit to all projected AWS growth?

Usually not. Separate current run rate, defensible committed spend, and strategic upside. Use ramps, expansion rights, and review mechanisms to capture upside without turning an uncertain forecast into a fixed liability.

Do Savings Plans replace an AWS EDP or PPA?

No. Savings Plans address eligible compute consumption. An EDP or PPA may provide broader enterprise pricing, but its scope and exclusions must be negotiated. The two mechanisms can coexist.

Can AWS Marketplace spend count toward an enterprise commitment?

Sometimes, depending on the agreement and eligible products. Never rely on a verbal assurance. Require the treatment of Marketplace spend to be stated clearly in the commercial documents.

Is AWS egress now free?

Not generally. Ordinary operational egress remains a cost category. Switching-related charges are subject to evolving regulatory requirements, including the EU Data Act, and AWS has separate policies and eligibility conditions for migration-related transfers.

When should we begin an AWS negotiation?

For a major renewal or enterprise commitment, begin six to nine months ahead where possible. Earlier preparation creates room for benchmarking, competitive pressure, internal alignment, and contract review.

Negotiate AWS before the commitment becomes the constraint

AWS spend compounds quietly. A small pricing gap across thousands of usage lines becomes a major financial issue when consumption grows every month.

The answer is not to avoid cloud commitments altogether. It is to make commitments that reflect reality, reward growth without overpaying for it, and preserve options when the roadmap changes.

The Negotiator Guru is an impartial negotiation firm. We do not accept vendor referral fees, and we operate discreetly as an extension of your team. We have delivered $1.05 billion in total cost impact globally, with average savings of 37% per deal. If we cannot identify at least 10% savings, we do not take the deal.

For a pending AWS negotiation, we can help with:

• Full Negotiation Support: Build the strategy, manage the negotiation, and support your team through every counter.
• Right Price Benchmarking™: Normalize AWS pricing, usage, discounts, credits, and effective rates.
• Contract Risk Review: Identify commitment traps, repricing risk, auto-renewal exposure, and missing protections before you sign.

AWS may control the pricing model. Your team can still control the negotiation.