AWS EC2 Cost Optimisation: Reserved Instances, Savings Plans and EDP Strategy for Enterprise

EC2 remains the single largest line item on most enterprise AWS bills — typically 25-45% of total AWS spend — and it remains the area where enterprises consistently overpay the most. Not because AWS EC2 is uniquely expensive, but because the default pricing path — on-demand consumption without a negotiated EDP or systematic commitment strategy — is priced to fund AWS's margins rather than optimise buyer cost.

The good news: AWS offers multiple legitimate mechanisms to reduce EC2 costs by 30-72% compared to on-demand rates. The challenge is that each mechanism involves different tradeoffs in flexibility, upfront commitment, and workload compatibility. Most enterprise FinOps teams understand the mechanisms theoretically — the gap is in having the quantitative framework to build the right combination for their specific workload mix, and the commercial leverage to negotiate EDP terms that unlock additional AWS-wide discounts.

This guide covers the full EC2 cost optimisation toolkit, from Reserved Instances and Savings Plans through to EDP negotiation and Graviton migration — with specific benchmarks and decision frameworks for enterprise buyers.

Why EC2 Remains the #1 AWS Cost Driver

EC2 costs grow because workloads grow. Most organisations migrating to AWS underestimate their eventual compute footprint — initial estimates of cloud compute requirements are routinely 30-60% lower than actual steady-state usage. The result: enterprises that planned to move 200 EC2 instances find themselves running 400-600 instances 18 months later, without the commitment strategies in place to control unit costs.

Compounding this, EC2 pricing complexity actively discourages optimisation. AWS offers hundreds of EC2 instance types across multiple generations, five pricing models (on-demand, Reserved Instances, Savings Plans, Spot, and Dedicated), and different pricing for each availability zone and region. The cognitive load of choosing the right instance type at the right price point for each workload is high enough that many enterprise teams default to on-demand pricing for new workloads and only revisit cost optimisation at year-end — after the overspend has already occurred.

The On-Demand Pricing Trap

AWS on-demand pricing is designed to be simple — it is also designed to generate maximum revenue. Every dollar of EC2 spend at on-demand rates is a dollar that could have been reduced by 30-60% with a commitment strategy. For enterprises spending $1M/month on EC2 at on-demand rates, running a proper commitment analysis typically reveals $3-7M in annual savings opportunity.

The on-demand trap compounds over time. As EC2 spend grows and teams add workloads at on-demand rates, the total commitment optimisation opportunity grows proportionally. A $10M/year EC2 spend running entirely at on-demand rates has a theoretical savings opportunity of $3-7M/year — representing a FinOps maturity gap that most enterprises have but few quantify explicitly.

The secondary on-demand trap is regional pricing. AWS charges different on-demand rates for the same instance type in different regions. A workload running in ap-southeast-1 (Singapore) costs 20-40% more than the same workload in us-east-1 for most instance types. For organisations with multi-region deployments, regional cost distribution is often a significant optimisation opportunity that sits outside the Reserved Instance / Savings Plan framework.

Reserved Instances: When They Work and When They Don't

EC2 Reserved Instances (RIs) provide a discount of up to 72% off on-demand rates in exchange for a 1 or 3-year commitment to a specific instance type, operating system, tenancy, and availability zone (Standard RIs) or to an instance family and region (Convertible RIs).

Standard Reserved Instances

Standard RIs provide the highest discount (40-72% off on-demand, depending on term and payment option) in exchange for the least flexibility. You commit to a specific instance type (e.g., m5.xlarge), operating system (Linux vs Windows), tenancy (default vs dedicated), and optionally availability zone. If your workload changes — different instance size, different OS — your Standard RI cannot be applied to the new configuration.

Standard RIs are appropriate for workloads that are: stable and predictable, running 24/7 or near-continuously, using a known instance type that is unlikely to change, and running on a known operating system. Legacy application servers, databases, and dedicated middleware layers typically qualify.

Convertible Reserved Instances

Convertible RIs provide 31-54% discount off on-demand rates with significantly more flexibility. You can exchange a Convertible RI for a different instance family, size, OS, or tenancy at any time during the term, as long as the new reservation has equal or greater value. This makes Convertible RIs appropriate for workloads that may evolve — for example, applications where you anticipate upgrading from m5 to m6i instances as AWS releases newer generation hardware.

Compute Savings Plans: The More Flexible Alternative

AWS Compute Savings Plans (CSPs) provide discounts of up to 66% off on-demand rates in exchange for a commitment to a specific dollar amount of compute usage per hour — regardless of which EC2 instance type, family, size, OS, or region consumes that compute. This makes Compute Savings Plans the most flexible commitment mechanism AWS offers for EC2.

The key advantages of Compute Savings Plans over Reserved Instances: CSPs automatically apply to EC2 usage regardless of instance type or region changes, they also apply to AWS Fargate and AWS Lambda usage (not just EC2), and they require no "exchange" process when workloads change. The tradeoff is that the maximum discount (66% at the highest commitment level) is lower than Standard RIs (72%).

EC2 Instance Savings Plans are a middle-ground option: they provide higher discounts than Compute Savings Plans (up to 72%) but with less flexibility than CSPs — you commit to a specific instance family and region, but get flexibility within size and OS within that family.

RI vs Savings Plans: The Decision Framework

The right combination of Reserved Instances and Savings Plans depends on your workload characteristics:

Enterprise best practice is a layered commitment strategy: use Standard RIs for the stable "base" of your EC2 footprint (typically 50-60% of total EC2 usage), Compute Savings Plans to cover the more variable mid-tier (20-30% of usage), and on-demand for the unpredictable peak (10-20% of usage). This combination typically achieves 40-55% blended savings versus running the entire estate on-demand.

Spot Instances: Enterprise Use Cases and Risk Management

AWS Spot Instances are EC2 capacity available at discounts of up to 90% off on-demand rates — but with a 2-minute interruption notice when AWS needs the capacity back. This makes Spot inappropriate for latency-sensitive or stateful workloads, but highly effective for a specific class of enterprise workload.

Enterprise Spot use cases that deliver very high ROI: batch data processing jobs, rendering and encoding workloads, CI/CD build pipelines, development and testing environments (where brief interruptions are tolerable), and fault-tolerant microservices that can handle graceful termination. A well-architected Spot strategy for these workloads can reduce the variable component of your EC2 bill by 60-85%.

Enterprise Discount Program: Negotiating Beyond AWS Standard Terms

AWS's Enterprise Discount Program (EDP) provides additional discounts on top of standard Reserved Instance and Savings Plan discounts, in exchange for a multi-year committed spend commitment (typically $1M/year minimum, with better rates at $5M+/year). EDP discounts typically range from 5-25% across all AWS services, applied on top of your standard commitment discounts.

The EDP is one of the most underutilised cost optimisation tools available to large enterprise AWS customers. Many enterprises that qualify for EDP — spending $2-5M+/year on AWS — have never formally negotiated an EDP because AWS does not proactively offer them. AWS account teams will present EDPs to large customers as a favour, but the initial offer is never AWS's best offer.

EDP Negotiation: What's Actually Achievable

From our experience negotiating AWS EDP agreements on behalf of enterprise customers, the following ranges apply to well-prepared buyers:

• $1M–$3M annual AWS commitment: EDP discount typically 5-12% on committed services; minimum annual commitment required; co-invest program credits sometimes included
• $3M–$10M annual AWS commitment: EDP discount typically 10-18% on committed services; co-invest credits typically $100K–$500K; accelerator clauses for spend growth
• $10M+ annual AWS commitment: EDP discount typically 15-25%+; significant co-invest credits; dedicated enterprise support enhancements; custom terms negotiable

AWS EDP negotiations are highly leverage-dependent. The primary levers are: current AWS spend trajectory, whether you have a credible multi-cloud strategy (Azure and/or Google Cloud involvement), the term length you are willing to commit to (longer terms = better rates), and whether you have external negotiation support that understands AWS's internal discount framework.

AWS Graviton Migration: EC2 Cost Savings from Architecture Change

AWS Graviton3 and Graviton4 processors provide 20-40% better price/performance than comparable Intel or AMD-based EC2 instances for most workload types. For enterprises running workloads on x86 instances that can be migrated to Graviton — typically modern application servers, containers, and cloud-native microservices — this represents a significant additional EC2 cost reduction that operates independently of Reserved Instance or Savings Plan strategy.

The Graviton migration decision is straightforward for modern, containerised workloads: the cost reduction is significant, and AWS provides good migration tooling. It is more complex for legacy applications compiled for x86 — particularly those using commercial third-party software where the software vendor may not provide ARM64 binaries. Before any Graviton migration commitment, validate that all application components (including third-party software, ISV applications, and compiled binaries) have ARM64-compatible versions available.

Combined with Reserved Instance or Savings Plan discounts, Graviton migration can produce blended EC2 savings of 55-70% versus standard on-demand x86 pricing — one of the highest-ROI infrastructure changes available in enterprise cloud environments.

Our AWS Cost Negotiation Service

NoSaveNoPay negotiates AWS cost reduction on a 25% gainshare basis. Our AWS specialist team combines FinOps optimisation (Reserved Instance and Savings Plan strategy) with commercial EDP negotiation to produce comprehensive cost reductions across your total AWS footprint. We have negotiated EDPs and commitment strategies for enterprises with AWS spend ranging from $2M to $80M+ per year.

Typical results from our AWS engagements: 25-38% reduction in total annual AWS spend, with the savings split roughly equally between commitment strategy optimisation and EDP discount improvement. For an enterprise spending $5M/year on AWS, this represents $1.25M–$1.9M in annual savings. Our fee is 25% of verified savings — so on a $1.5M saving, the fee is $375K. Zero fee if we save nothing.