Private Equity AWS Cost Optimization: From Diligence to Exit

Most value creation plans cover pricing, procurement, headcount and add-on acquisitions. The AWS bill usually gets one line in the diligence report: “cloud costs look high.” Nobody puts a number on how much of it is recoverable.

That’s a gap. AWS spend is one of the few cost lines you can measure precisely before you sign, reduce within weeks of close, and keep down for the whole hold period. In a software business it often sits in cost of revenue, so every dollar saved improves gross margin as well as EBITDA.

This guide is for operating partners, deal teams and portfolio-company CFOs and CTOs. It covers what an AWS scan shows before the LOI, the questions to ask in diligence, how the savings reach EBITDA, and how to capture them after close without signing commitments the next owner has to inherit.

Why AWS spend belongs in the value creation plan#

Three things set AWS apart from most cost levers.

You can measure it up front. Most of tech diligence is judgment. You can assess an engineering team, but you can’t put a dollar figure on it. The AWS bill is itemized down to the resource, so a scan can show which services are driving waste and what fixing them is worth, account by account.

It doesn’t need a reorganization. Cutting AWS waste means retyping storage, deleting idle resources, resizing overprovisioned instances and paying a lower rate for compute you already run. No layoffs, no supplier renegotiation, no office moves.

The savings are usually there. CloudFix has analyzed $2B+ in AWS spend for 500+ companies. It typically finds 15–60% savings on each AWS service it optimizes, which adds up to 20–35% off the total bill. The average is 23% off the total bill. We’ve never scanned an account with meaningful spend and found nothing.

The catch is timing. Every month waste keeps running after close is money that never reaches EBITDA. So the work starts before you sign.

What a pre-LOI AWS scan shows#

A CloudFix scan reads the target’s Cost and Usage Report, CloudWatch metrics and resource configuration across every connected account. It runs 110+ finders across 30+ AWS services, and results are typically ready within 24 hours.

Here’s what it tends to surface:

• Storage on the wrong type or tier. EBS volumes still on gp2 when gp3 costs less per GB, and S3 data sitting in the wrong storage class.
• Idle resources. Unattached volumes, old snapshots, idle load balancers and NAT gateways: running, billed and serving nothing.
• Overprovisioned compute. EC2 and RDS instances sized for a peak that never came.
• Old software paying a surcharge. RDS and EKS versions past the end of standard support pay AWS Extended Support fees on top of normal usage.
• On-demand rates on steady workloads. EC2 that has run predictably for months with no Reserved Instance or Savings Plan covering it.

What the deal team gets back#

A dollar figure by service and account. Not “cloud costs seem high,” but a specific annualized number, broken down so you can see what’s driving it. That number either supports the thesis or changes the bid. If it’s small, that’s useful too: it tells you the team already runs a tight ship.

A Day 1 plan. The findings are a to-do list. Some map to CloudFix’s 53 automated fixers; others need an engineer’s judgment. Either way, you walk into close with an ordered list instead of a vague mandate to “look at cloud costs.”

The target owns the account, so expect the seller to see the same report. That’s fine. A number both sides can see is easier to agree on than a guess. And if the seller has added projected cloud savings back to adjusted EBITDA, the scan shows whether that opportunity really exists.

What the target has to do#

Access is usually the first objection, so be specific with the seller’s team. They deploy one AWS CloudFormation stack, which creates read-only IAM roles for CloudFix. It takes about 5 minutes. There are no long-term credentials, passwords or access keys to hand over, and nothing changes in their infrastructure without their approval. CloudFix is SOC 2 Type 2 audited, and their security team can review how it connects and what it can read at the CloudFix Trust Center. They can remove the roles at any time.

If a full scan isn’t possible before the LOI, the questions below still work on Cost Explorer exports, and the RightSpend calculator shows what a 20–55% discount, less the 25% fee, would mean on the target’s uncovered on-demand EC2 spend. Then run the full scan in confirmatory diligence, before signing.

AWS due diligence questions to ask#

Use these in management meetings or add them to the data room request list.

• What’s the monthly AWS run rate, and how has it moved against revenue over the last 12 months? Spend growing faster than revenue is the first flag.
• Where does AWS sit in the P&L? If hosting is in cost of revenue, savings improve gross margin, not just EBITDA.
• What share of EC2 runs at on-demand rates? Ask for Reserved Instance and Savings Plan coverage and utilization from Cost Explorer.
• What commitments exist, and when do they end? List every Reserved Instance and Savings Plan with its end date. A 3-year commitment bought for last year’s fleet keeps billing if the fleet shrinks.
• Is there an EDP or private pricing agreement? Check the annual minimum, the end date and the shortfall terms. If optimization takes spend below the commitment, the shortfall payment can eat into the savings. Our piece on EDP negotiations covers the traps.
• Does the target get discounts it won’t keep after close? If commitments or pricing come through a parent company or a reseller, they may not survive a carve-out. AWS terms say Reserved Instances and Savings Plans are for a single end customer’s usage.
• Is anything paying Extended Support? Compare RDS and EKS versions against AWS’s support calendars.
• Is spend tagged by product, environment or customer? Untagged spend is hard to cut and hard to explain to the next buyer.
• Who owns the AWS bill today? A named owner with a monthly review is a good sign. “Finance pays it” is not.

How AWS savings flow to EBITDA#

AWS is an operating cost, billed monthly. A lower bill reaches EBITDA in the month it happens, less whatever you pay to get it. There’s no restructuring charge to explain.

Here’s an illustrative example with round, hypothetical numbers:

• AWS spend: $1M a month, or $12M a year.
• At 20–35% off the total bill: $2.4M to $4.2M a year in savings.
• At the 23% average: about $2.76M a year.
• At a hypothetical 10x EBITDA multiple: each $1M of lasting annual savings adds about $10M of enterprise value at exit.

Two things change the real number.

Fees. CloudFix plans are priced by annual AWS spend and start at $149 a month. RightSpend’s fee is 25% of the net-new savings it generates, so you keep 75%, and savings you already get from your own commitments carry no fee. The RightSpend fee is ongoing and moves with EC2 use, so model the net figure, not the gross.

Timing. In the same example, every month the work waits is about $230K (23% of $1M) that never reaches EBITDA. Savings captured early are also in the trailing twelve months by exit. Actual results are easier to defend in a buyer’s quality of earnings review than a pro forma adjustment.

The first 100 days: fix usage and cut the rate#

The standard advice is to rightsize first and commit second. Commit first, and you lock in a discount on waste. Rightsize first, and you pay on-demand rates while engineers work through the list.

After an acquisition, both paths get worse. The architecture is about to change as you integrate, consolidate accounts or move workloads, so a 3-year commitment sized on Day 30 may not fit Day 300. CloudFix and RightSpend run in parallel, so you don’t have to choose.

Usage: CloudFix#

If the target connected CloudFix during diligence, the findings are already waiting. If not, connect after close: about 5 minutes, with results typically within 24 hours. See how CloudFix works.

From there, the portfolio company’s team approves each fix. Approved fixes run as AWS Systems Manager Automation runbooks in the company’s own account, under a dedicated CloudFix role, and every execution is logged. Start with the simple ones, such as storage retypes and idle resources, then move into rightsizing. The finder and fixer catalog shows what each one changes, and the 30-day plan shows a typical rollout.

Rate: RightSpend#

At the same time, RightSpend covers the EC2 usage that the company’s own Reserved Instances and Savings Plans don’t. It gives Commitment-Free Discounts: 20–55% off on-demand EC2, with no 1- or 3-year commitment to sign. Coverage adjusts every hour as usage moves. RightSpend doesn’t touch instances. It changes the rate you pay, not what you run.

The two products work together. RightSpend doesn’t lock you into instance types or families, and when CloudFix resizes or removes an instance, RightSpend drops that coverage to make room. Cutting usage never strands a commitment. RightSpend is built for companies spending $25K a month or more on EC2, and savings typically show up within 24 hours.

For the integration work itself (moving acquired workloads into your AWS Organization, setting tag policies, deciding which existing commitments to keep), CloudFix founder Rahul Subramaniam’s 18 cost-saving tips for AWS mergers and acquisitions is the practical companion to this guide.

Why commitment-free discounts suit a PE hold#

A 3-year Reserved Instance or Savings Plan is sized to the usage you had on the day you signed. If usage falls, you keep paying for capacity you no longer run until the term ends. On a finite hold, that’s a real problem: integration changes the fleet, add-ons get merged, and at exit the buyer inherits whatever you signed.

RightSpend’s discounts come from a ladder of short-dated commitment rungs. When usage falls, rungs expire and aren’t replaced, so coverage follows usage down. If a company leaves, the remaining rungs expire and it goes back to on-demand, with no stranded commitment. When a global consulting firm left RightSpend, its ladder unwound in 14 days.

For a PE owner, that means:

• No commitment risk during integration. Resize, consolidate or shut down workloads without paying for capacity you’ve stopped using.
• Discounts that match the hold. The discount lasts as long as the usage does.
• A cleaner exit. No multi-year AWS commitment in the data room for the next buyer to price in.

Keeping the savings through the hold period#

AWS waste comes back. Teams ship new services, forget test environments and copy old patterns. AWS keeps releasing cheaper options, which turns yesterday’s sensible choice into today’s overspend. A one-off cleanup decays.

CloudFix keeps scanning after the first pass, so new waste shows up as a new recommendation instead of a surprise in the next board deck. That matters because a recommendation isn’t a saving until someone acts on it. Fixers make acting on it cheap.

CloudFix started as an in-house tool for this exact problem: controlling the AWS bill across a growing portfolio of companies (our story). Across your own portfolio, the same playbook applies:

• Scan every add-on before you sign. Same questions, and a real number in the model before the LOI.
• Buy the same way everywhere. CloudFix and RightSpend are both billed through AWS Marketplace, so each portfolio company buys with its existing AWS account and has no new vendor to set up. Marketplace spend also counts toward its AWS spend commitment (EDP or PPA).
• Prepare for exit early. The next buyer may run the same diligence you did. Better that their scan finds little left to add, and that the savings already show in your actual results.

Start with a scan#

The best time to find AWS waste is before you’ve priced the deal. Connect the target’s accounts, or your portfolio company’s, with read-only roles in about 5 minutes, and see savings by service and account, typically within 24 hours.

Get a free AWS savings assessment