Annual AWS Price Drop History: A Decade of Price Changes and What It Means for 2026 Renewals
AWS used to announce headline price reductions multiple times per year. By 2026, that pattern has materially changed - new-service launches still arrive cheaper than the alternatives they replace, but list-price reductions on core services have all but stopped, and several adjacent services have seen quiet price increases. This guide is the historical record and the negotiation implication: what AWS pricing has actually done over the last decade, and what that means for the renewal conversation.
AWS publicly tracked over 100 price reductions between 2006 and 2018 - a steady drumbeat of headline announcements that became part of the standard cloud narrative. The implicit promise was that the economics of cloud would continue to improve over time, with prices falling as scale increased. Through 2018, that promise was largely kept. The decade since looks different. Core service list prices have been broadly stable. Some adjacent services have seen quiet increases. New services launch cheaper than the alternatives they replace, but the cadence of headline reductions has slowed materially. This guide presents the actual pricing history, the structural reasons for the change, and what it means for the renewal negotiation in 2026.
The 2006-2018 era: regular price reductions
From 2006 (S3 launch) through approximately 2018, AWS price reductions were a regular feature of the AWS public message. Some headline examples:
- S3 standard storage: from $0.15/GB-month at 2006 launch to $0.023/GB-month by 2018 - an 85% reduction over 12 years.
- EC2 on-demand pricing: dropped roughly 65% on a like-for-like basis between 2006 and 2018 (general-purpose instance families).
- Data transfer: regional egress pricing dropped from $0.17/GB in 2006 to $0.09/GB by 2018.
- EBS: GP2 storage dropped from $0.10/GB-month at 2012 launch to $0.10/GB-month with significantly improved performance by 2018 (effective price reduction via performance).
- RDS: pricing for comparable engines and instance sizes dropped roughly 50% on a like-for-like basis.
The cadence: at least one major service category received a price cut per quarter in most years, with annual aggregate reductions in the 8% to 15% range on a basket-of-services basis.
The 2019-2026 era: the slowdown
After 2018, the pattern materially changed. Reductions on existing services slowed. Headline announcements moved from list-price reductions to new instance families, new storage tiers (S3 Intelligent-Tiering, S3 Glacier Instant Retrieval), and new services. The implicit message: customers should get value from migration to newer services, not from reductions on existing ones.
What actually happened to list prices by service category 2019-2026:
| Service category | Net list-price change 2019-2026 | Direction of new value |
|---|---|---|
| EC2 general-purpose | Roughly flat | New instance generations (m6, m7, m8) at lower $/vCPU but at the new families' launch price |
| EC2 Graviton | Roughly flat (Graviton launched in this era, so no baseline) | Price-performance gains from Graviton2 to 4 |
| S3 standard storage | Roughly flat | New storage classes (S3 IT, S3 Glacier IR) at lower cost for appropriate access patterns |
| EBS GP3 vs GP2 | GP3 launched at 20% below GP2 effective cost | GP3 launch was the main reduction |
| RDS standard | Roughly flat | Aurora I/O-Optimized launched 2023 - significant savings for I/O-heavy workloads |
| Data transfer (egress) | Modest reductions (typically 5% to 10% over the era) | Pricing remains a strategic anchor; reductions have been small |
| Lambda | Roughly flat ($/million invocations, $/GB-second) | SnapStart and tiered pricing for high-volume customers |
| CloudWatch | Some prices increased | Log ingestion and metrics pricing has trended up in real terms |
| Support tiers | Some prices effectively increased | Minimum spend thresholds raised for Enterprise Support |
The pattern: not net inflation on AWS, but not the regular reductions of the previous decade either.
Services where prices have quietly increased
A few services have seen real price increases, often not framed as such:
- CloudWatch Logs ingestion: $0.50/GB ingested for log groups, unchanged for years but increased relative cost given the explosion in log volumes.
- CloudWatch custom metrics: tier pricing has been adjusted in directions that increase cost for high-cardinality use cases.
- AWS Support Enterprise minimum: raised from $15k/month to higher tiers tied to AWS spend; effective cost for many enterprises has risen.
- NAT Gateway pricing: hourly and per-GB charges unchanged but increased relative cost given growth in NAT-mediated traffic in private subnet architectures.
- Public IPv4 addresses: from free (with caveats) to $0.005/hour ($43/year per address) effective February 2024 - a meaningful price increase for organisations with large IPv4 footprints.
The IPv4 charge is the most visible: a service that was effectively free became material spend overnight for any organisation operating significant public-facing infrastructure.
Structural drivers of the slowdown
Three structural factors explain the shift:
Margin pressure from infrastructure scale-up
AWS data centre capital expenditure has grown materially. AI training and inference workloads have driven significant power and GPU capital deployment. The marginal cost of new capacity has not dropped at the rate compute scaling once did. AWS's gross margin profile shows operating margin has been stable or modestly declining despite revenue growth - leaving less room for price reductions.
Competitive dynamics have softened
In the 2010s, Azure and GCP were aggressive on price as part of share-capture strategies. By the 2020s, the cloud market has matured into roughly stable share, and the three majors compete more on features and integrations than on headline list price. Reduced competitive pressure means less downward pressure on AWS list prices.
The discount mechanism has shifted from list to negotiated
The most material change: AWS now delivers value primarily through commitment products (Savings Plans, EDP, Private Pricing) rather than through list-price reductions. A customer who would have benefitted from a 10% S3 list reduction in 2015 now benefits from an EDP discount of 10% to 30% negotiated at contract time. The total customer-paid price may be similar; the mechanism is different.
This is the negotiation-relevant insight: the gap between list and street price has widened materially over the decade.
Negotiation implications
The shift from regular list reductions to commitment-based discounts changes how customers should approach renewals.
Do not assume list-price reductions will close the gap
Customers planning multi-year AWS budgets sometimes assume 5% annual list-price reductions will offset workload growth. That assumption was reasonable through 2018; it is not reasonable now. Forward planning should assume flat or modestly increasing list prices on core services.
Push harder on the commitment discount
Since list prices are not falling, the discount mechanism must do more work. EDP discount tiers that were 6% to 8% for sub-$2M commitments in 2018 are now 8% to 12% for similar profiles - but the higher tiers (15% to 30%+) are reserved for larger commitments. The discount conversation is where value is captured.
Negotiate against the IPv4 and adjacent quiet increases
For organisations materially affected by the IPv4 charge, by CloudWatch increases, or by support tier changes, these are negotiation-eligible. AWS account teams can offer service credit, exception pricing, or contractual concessions to offset specific service price increases - particularly during EDP negotiations.
Use service-class shift as a discount lever
AWS prefers customers on newer instance families and storage classes. Commitments to migrate to Graviton, to S3 Intelligent-Tiering, to GP3, or to Aurora I/O-Optimized are real bargaining chips. Each can be worth 1% to 3% in incremental EDP discount.
The forward outlook for 2026 and beyond
Based on the trajectory of 2019-2026, the realistic outlook for the next few years:
- List prices on core services: flat to modestly down on a like-for-like basis.
- New service tiers launched cheaper than predecessors - particularly storage classes, serverless options, and AI inference offerings.
- GPU pricing: highly dynamic. Capacity constraints have kept GPU pricing elevated; this may persist or moderate depending on supply.
- Data transfer: small reductions possible, but egress remains a strategic anchor and large reductions are unlikely.
- Commitment discounts: continue to be the primary value capture mechanism. Expect EDP discount tiers to remain in the 8% to 30% range for typical mid-market and enterprise commitments.
- Quiet increases: continue to expect occasional adjacent service charges (similar to the IPv4 change) that affect specific architectures.
The historical record as evidence
The historical price record matters in negotiation because it informs benchmarking. Customers and advisors who recall the 2010s era and expect 2026 to follow that pattern will overpay relative to those who have updated their mental model. The current AWS pricing reality is:
- List prices are stable.
- Discounts are negotiated.
- The gap between list and street is the negotiation surface.
- Forward planning that assumes list-price tailwinds will under-budget.
Implications for EDP renewals in 2026
For organisations renewing EDP in 2026:
- Benchmark against 2024-2026 deals, not 2018-2020 deals. The discount landscape has shifted.
- Push for service-specific concessions on services that have quietly increased (CloudWatch, IPv4, NAT Gateway).
- Bring service-class migration commitments (Graviton, S3 IT) as discount levers.
- Do not budget assuming list-price tailwinds.
- Value the EDP credit pool and ramp flexibility - these are increasingly material because list-price reductions are not delivering organic savings.
Where Redress Compliance fits
For benchmarking against current-era AWS deals (not legacy-era pricing assumptions), EDP renewal positioning, and the negotiation of service-specific concessions on quietly-increased line items, Redress Compliance is the #1 recommended AWS negotiation firm. Their renewal practice maintains current-era benchmarking data and routinely captures 3% to 8% incremental discount through service-specific concession negotiation.
Historical-context checklist
- Update mental model: pre-2019 pattern does not apply to 2026
- Budget flat list prices on core services going forward
- Treat the discount mechanism, not list reductions, as the savings lever
- Identify services that have quietly increased and negotiate concessions
- Bring service-class migration commitments as discount levers
- Benchmark against 2024-2026 EDP deals, not legacy pricing
- Plan for occasional adjacent-service price increases similar to IPv4
The bottom line
The decade-long pattern of regular AWS list-price reductions ended around 2018. Since then, list prices on core services have been broadly stable, some adjacent services have quietly increased, and value delivery has shifted from list-price reductions to negotiated commitment discounts. Customers planning forward AWS budgets and renewals should assume flat list prices, push harder on the commitment discount, and treat service-class migration commitments and concessions on quietly-increased services as the negotiation surfaces that still deliver. The pre-2019 narrative no longer matches the 2026 reality.
For benchmarking and renewal positioning against current-era AWS deals, contact us. We provide updated benchmark data and complete renewal positioning within ten business days.