EDP NegotiationSavings Plans OptimizationReserved Instances StrategyEC2 Right-SizingS3 Cost ReductionEgress NegotiationMigration CreditsSupport Tier AdvisoryMulti-Cloud LeverageBedrock AI PricingEDP NegotiationSavings Plans OptimizationReserved Instances StrategyEC2 Right-SizingS3 Cost ReductionEgress NegotiationMigration CreditsSupport Tier AdvisoryMulti-Cloud LeverageBedrock AI Pricing
Service · Compute Spend

AWS compute spend negotiation, top to bottom.

EC2 right-sizing, Graviton migration cost modeling, Spot Fleet strategy, auto-scaling cost impact analysis, and container orchestration optimization across EKS, ECS, and Fargate. We shrink the floor before negotiating the rate.

$2.4B+
AWS spend reviewed
500+
Engagements
38%
Avg reduction
$340M+
Documented savings
Overview

Right-sizing comes before
commitment, not after.

The most expensive mistake in AWS commitment planning is buying Savings Plans against an inflated baseline. If your fleet is over-provisioned by 25%, a three-year Compute Savings Plan locks in that over-provisioning for 36 months. Right-sizing must come first — and it has to be more than a CloudWatch screenshot.

We work the entire compute stack: instance-family selection, generation upgrades, Graviton migration paths, Spot capacity strategy, container density, and the auto-scaling policies that determine how your fleet behaves under load. Each lever changes the floor that commitments need to cover, and each has second-order effects on architecture and operations.

Done in the right order — right-size, then re-architect, then commit — compute spend typically falls 30–45% versus its uncorrected baseline before any negotiation begins. The commitment work then layers on additional discount against a cleaner baseline.

What we negotiate

Every lever in the compute stack.

01EC2 right-sizing and generation refresh+
90-day CloudWatch and Compute Optimizer analysis. We identify over-provisioned instances, recommend exact target sizes, and stage refreshes from M5/C5/R5 to current-generation M7i, C7i, and R7i where price-performance gains warrant the cutover.
02Graviton (ARM) migration modeling+
Graviton-based M7g, C7g, R7g instances are priced approximately 15–20% below x86 equivalents and frequently deliver 20–40% better price-performance. We model the migration workload by workload, identify portability risk, and sequence the cutover for cost-impact-per-week.
03Spot Fleet and capacity-optimized allocation+
Spot capacity remains the deepest compute discount in AWS — 60–90% off on-demand for interruption-tolerant workloads. We design Spot Fleet configurations with capacity-optimized allocation, instance diversification, and fallback strategy for batch, CI/CD, ML training, and stateless serving.
04EKS, ECS, and Fargate cost engineering+
Container density, node pool sizing, Karpenter configuration, Fargate vs EC2 launch type, and bin-packing efficiency. We audit container platforms end to end and rebuild node pool strategy to maximize utilization without sacrificing reliability.
05Auto-scaling cost impact analysis+
Aggressive scale-up thresholds and conservative scale-down delays are the two most common cost leaks in auto-scaling. We tune ASG policies for cost-aware behavior without compromising user-facing latency SLOs.
06License-aware compute planning+
SQL Server, Oracle, RHEL, and Windows licensing add hidden compute cost. We model license-included vs BYOL options, identify dedicated host opportunities, and integrate license cost into the right-sizing decision.
Process

Right-size, re-architect, then commit.

1.

Compute audit

90-day usage analysis across every account. Output: right-sizing recommendations, Graviton portability scoring, Spot eligibility map, and container utilization report.

2.

Architecture co-design

Work with your engineering teams to sequence the highest-impact changes: generation refresh, Graviton waves, Spot adoption, and container density.

3.

Commitment alignment

Re-baseline the compute floor and align Savings Plans and EDP commitments to the post-optimization baseline rather than the inflated starting point.

Results

What clients actually save.

Related services

Often combined with compute work.

Your compute baseline
is inflated.

38% average reduction across 500+ engagements. We audit your fleet in two weeks.

How we deliver

Four phases. One outcome.

01

Diagnostic (week 1)

Cost and Usage Report ingestion, contract review, EDP scorecard. You get a benchmark against 500+ comparable deals.

02

Strategy (weeks 2-3)

Negotiation positions, BATNAs, target outcomes by line item. We build the playbook and the supporting models.

03

Execution (weeks 4-9)

We sit in your seat opposite AWS. You stay in control of the relationship; we shape the deal.

04

Hand-off (week 10+)

Signed terms, internal playbook, monitoring framework. So you can defend the deal at the next renewal yourself.

Questions

Frequently asked. Directly answered.

How much can I save by migrating to Graviton?+

Workloads that successfully run on Graviton typically see 15-25% price-performance gain over comparable x86 instances. The real cost is migration effort: ARM compatibility for compiled binaries, container base images, and third-party software. We model both the gross savings and the migration burden.

Is Spot worth using outside of batch and stateless workloads?+

Spot is most valuable for any workload that can tolerate interruption with checkpoint/restart logic. Modern Spot Fleets with capacity-optimized allocation strategies have made interruption rates much lower (often <5% for diversified fleets), so the addressable surface area is larger than most teams realize.

What's the cost impact of EC2 right-sizing?+

We typically find 20-35% of EC2 spend is on over-provisioned instances based on 14-day p95 utilization. Right-sizing alone usually unlocks 8-12% of total cloud spend before any negotiation.

Can I negotiate a custom EC2 price for my workload?+

Custom EC2 rates are negotiated through EDP-bound PPAs, not as standalone deals. The leverage comes from concentration: if a single instance family is >40% of your EC2 spend, you have a case for family-specific PPA pricing.

Do auto-scaling groups affect Reserved Instance and Savings Plan utilization?+

Yes — aggressive scale-down can leave RIs and SPs uncovered, eroding your savings rate. Coverage strategy must account for daily and weekly troughs in your scaling pattern, not just average usage.