Cloud waste audits for PE-backed portfolios: a 2026 playbook for operating partners

Why PE Portfolio Cloud Waste Is a Category of Its Own

A single company’s cloud waste problem is tractable. You have one CFO, one platform team, one AWS Organization, and a shared understanding of what the infrastructure is supposed to do. A PE portfolio is different. An operating partner at a mid-market firm typically has visibility across 10 to 80 portfolio companies — each with its own stack, its own FinOps maturity, its own acquired-but-not-integrated business units, and a board that wants EBITDA expansion without disrupting product roadmaps.

The 2026 reality: cloud spend is now one of the top three cost lines for most SaaS portfolio companies, and for data-heavy or AI-centric portfolios it is frequently the single largest variable cost after payroll. The industry-standard figure — roughly 29% of cloud spend is waste — holds across the portfolios we’ve seen, but the distribution is heavy-tailed. One or two companies in an average portfolio will have 45–60% waste, usually because they’ve recently scaled, recently acquired, or recently pivoted. Those outliers are where the operating partner’s audit program earns its keep.

This post is a 2026 playbook for PE operating partners and portfolio-company platform leaders running a cross-company cloud waste audit. It is written for the reality of private equity: constrained headcount, a hold period clock, and boards that will not tolerate audit programs that destabilize product delivery.

The Four Waste Archetypes Across a Portfolio

Every portfolio company’s cloud waste falls into one of four categories, and each requires a different intervention:

Compute over-provisioning. Oversized instances, idle reserved capacity, autoscalers with floor-that-is-too-high settings, and non-production environments running at production size. This is the single largest category and the easiest to address — typically 12–18% of portfolio cloud spend, and most of it is addressable without any product-team involvement beyond approvals.

Orphaned and forgotten resources. EBS volumes from terminated instances, Elastic IPs with no attachment, load balancers serving no traffic, database snapshots from acquisitions three years ago, CloudWatch log groups with no retention policy. Typically 3–6% of spend and worth finding because the remediation requires zero product-team coordination.

Architectural inefficiency. NAT Gateway costs that should be VPC endpoints, cross-AZ traffic that should be same-AZ, S3 without Intelligent-Tiering, RDS without Aurora Serverless where appropriate, Lambda provisioned concurrency where on-demand would suffice. Typically 6–10% of spend and requires a design conversation with the product team — longer to address but the savings compound.

Commitment underutilization. Savings Plans and Reserved Instances purchased too aggressively, expired without renewal, or mis-sized for actual workload patterns. Typically 3–8% of spend. Addressing this usually requires a joint FinOps-Finance conversation.

Across a portfolio, these four categories account for roughly 25–35% of cloud spend combined. For a portfolio with $120M in aggregate cloud spend — common for a mid-market portfolio of a dozen tech-enabled companies — that is $30–42M in addressable annual waste. At a 12x EBITDA multiple, identifying and capturing half of it adds $180–250M to portfolio enterprise value.

The Operating Partner’s Portfolio Playbook

The standard mistake is running the same audit playbook at every portfolio company — a thirty-page checklist, a six-week consulting engagement, a slide deck that nobody reads. That produces a list of problems and zero action. The PE-effective playbook is staged differently:

Stage 1 — Portfolio-Wide Benchmarks (weeks 1–2). Normalize every portfolio company’s cloud spend against a small set of benchmarks: cost per active customer, cost per employee, cost as percentage of ARR, compute-to-storage ratio, and waste percentage (idle or orphaned resources divided by total spend). Do not try to rank by absolute spend — different business models will land differently. Rank by gap from portfolio median, and focus the first intervention on the worst outliers on the waste-percentage dimension.

Stage 2 — Targeted Audits at the Outliers (weeks 3–6). The three to five companies at the top of the waste-percentage ranking get a deep audit. The audit surfaces the specific top-twenty waste items per company, with estimated annual savings and the engineering effort to capture each. The output is a one-page ranked backlog per company — not a report.

Stage 3 — Lightweight Rollout at Typical Companies (weeks 4–12). The companies in the middle of the distribution get a self-service audit — either your tool of choice or a vendor product — plus a 90-minute workshop with the platform team to walk through the top ten findings. Expect to capture 40–60% of identified waste at these companies within the quarter without additional operating-partner involvement.

Stage 4 — Continuous Monitoring (ongoing). Install per-company anomaly detection and quarterly portfolio benchmark refreshes. The goal is to catch regressions — new acquisitions, new product lines, new hires who did not get the FinOps onboarding — before they compound. Most waste returns within twelve months if the monitoring layer is not in place.

The staging matters. Operating partners who try to run Stage 2 at every company at once will burn operating-partner capacity and portfolio-company goodwill. Running Stage 1 first and letting the numbers drive the prioritization keeps the effort proportionate to the opportunity.

See the IAN team run on your cloud. We connect to your AWS account via a scoped read-only role, run the Observe-tier agents, and leave you with a concrete audit report — cost waste, security exposure, compliance gaps, and a labor-offset estimate. You keep the findings regardless of next steps. Get a free infrastructure audit →

The EBITDA Conversation With Boards

Operating partners who frame cloud waste in dollars alone tend to lose the board-meeting conversation. The effective frame is EBITDA margin improvement, multiple expansion potential, and competitive positioning versus portfolio peers.

A useful board slide contains four numbers per company: current cloud spend as a percentage of revenue, peer-group median for the same metric, the gap in absolute dollars, and the portion of that gap that a twelve-month audit can realistically capture. The last number is the one the board will remember. Typically it is 40–55% of the identified gap — the rest is either structural (a different business model genuinely requires more compute) or requires engineering effort that does not fit the hold-period window.

The companion slide is the EBITDA bridge: spend reduction goes 100% to EBITDA in year one, and for a portfolio exiting at 12x EBITDA, a $4M annualized cloud savings captured 18 months before exit contributes roughly $50M to enterprise value. That is a conversation-ending number for most boards.

What Actually Gets Cut

A representative portfolio audit across a dozen companies typically surfaces the following pattern in the waste inventory:

• Oversized EC2 and RDS instances — 40% of findings by dollar value. Most of these are historical: an instance was sized for a launch, usage never materialized, nobody right-sized it. Compute Optimizer recommendations combined with a two-week observation window capture most of this.
• Idle non-production environments — 15% of findings. Dev and staging environments that run 24/7 at production size. Schedule-based scale-down (nights and weekends) is a 65–70% cost reduction on this slice.
• Unattached and orphaned resources — 8% of findings. EBS volumes, EIPs, old snapshots, unused NAT Gateways. High-ROI to cut because there is no product-team impact.
• Unused or mis-sized commitments — 10% of findings. Savings Plans that do not match current workload patterns, expired reservations on on-demand pricing. Requires a joint Finance-Platform conversation.
• S3 lifecycle gaps — 7% of findings. Logs, backups, build artifacts with no lifecycle policies. Typical remediation: move to Intelligent-Tiering or add a 90-day Glacier transition.
• Data transfer — 10% of findings. NAT Gateway costs, cross-AZ traffic, cross-region replication. Requires an architectural conversation and produces the highest variance in savings.
• Everything else — 10%. Lambda provisioned concurrency, API Gateway over-provisioning, DynamoDB provisioned capacity that should be on-demand, CloudWatch log retention.

The specific percentages vary by portfolio, but the shape is stable. An audit program that does not find meaningful waste in the first three categories is not looking hard enough.

The Specific Tools That Are Working in 2026

The 2026 toolchain for portfolio cloud waste audits has converged on a small set of layered tools:

• AWS Cost Explorer and Compute Optimizer for native recommendations. Free, directly actionable for the EC2 and RDS right-sizing findings. Turn it on at the Organization level with delegated admin.
• CUR 2.0 plus Athena or Snowflake for portfolio-wide queries. Enables cross-company benchmarks that neither Cost Explorer nor any single-company FinOps vendor can do without a cross-account data model.
• Vantage, Cloudability, or ProsperOps for commitment management. ProsperOps specifically does automated Savings Plan optimization which removes the quarterly “did we re-cover our reservations” meeting.
• OpenCost or Kubecost for Kubernetes attribution. EKS is the single source of unexplained cloud cost in most portfolio SaaS companies; OpenCost’s split-cost allocation makes it legible.
• IAN as the portfolio-layer aggregator — cross-company benchmarks, anomaly detection, and opening pull requests for the remediation. This is where we spend most of our time in 2026.

Avoid the fallacy of one-size-fits-all tooling. A $100M cloud-spend portfolio company running Kubernetes-heavy ML workloads needs a different toolchain than an $8M-spend transaction-processing SaaS. Match tools to the company profile; run the benchmarks at the portfolio layer.

The Three Common Failure Modes

Programs that run well in quarter one and falter in quarter three typically fail in one of three ways:

Failure mode one — findings without ownership. The audit produces a list; nobody owns the remediation; quarter two ends with the same list. Fix this by naming a specific engineer per company at the time the finding is recorded, with a deadline and an explicit EBITDA-impact estimate that the company’s CFO reviews.

Failure mode two — one-and-done audits. The first audit captures 25% of cloud spend. The portfolio company grows by 40% in the next twelve months. The waste ratio re-normalizes because there is no continuous monitoring. Fix this by installing per-company anomaly detection and a quarterly re-benchmark.

Failure mode three — operating-partner burnout. The operating partner becomes the de-facto FinOps function for five portfolio companies and gets pulled into every dev-environment conversation. Fix this by investing in one cross-portfolio platform (tool or person) who runs the benchmarks and the continuous monitoring, and who escalates to the operating partner only for findings with >$500K annualized impact.

The third failure mode is the most common. The solution is to treat portfolio FinOps as a shared function — either a dedicated platform or a dedicated person (sometimes an EIR, sometimes a fractional operator) whose mandate is cross-portfolio cloud cost.

How IAN Helps

IAN is purpose-built for the cross-portfolio cloud cost use case. Operating partners connect the portfolio companies’ AWS Organizations (read-only, via a cross-account role). IAN runs the portfolio-wide benchmarks on day one, surfaces the per-company top-twenty waste backlog by the end of week one, and opens pull requests against each portfolio company’s infrastructure repository for the remediations — right-sizing, scheduled scale-down, orphaned-resource deletion, lifecycle policies — so the product teams review the fix on their existing workflow instead of receiving a report.

For operating partners, IAN aggregates the portfolio view: cross-company waste percentages, quarterly EBITDA impact tracking, and regression alerts when a portfolio company’s waste ratio starts climbing. The 29% of cloud spend that is wasted across the portfolio becomes a rolling backlog that the platform keeps reducing, with the operating partner reviewing the EBITDA bridge at each board cycle.

Get a free infrastructure audit → | See pricing →