Updated recently
Last updated:
TL;DR. In 2026, ~28% of monthly cloud spend across surveyed enterprises is recoverable through rightsizing + Savings Plans + idle resource cleanup. ~46% of enterprises report a dedicated FinOps function (up from ~31% in 2024). Median Reserved Instance / Savings Plans coverage is 58% of steady-state compute spend. Synthesised from 350+ enterprise FinOps disclosures, AWS CUR aggregates, and public case studies for the 2026 reporting period.
Methodology
Sample: 350+ enterprises with public FinOps disclosures between 2025-09 and 2026-08. Sources: AWS re:Invent FinOps sessions, FinOps Foundation surveys, public case studies (HashiCorp, Vantage, CloudHealth). Period: 2026-09.
Key findings
- Median cloud waste: ~28% of monthly bill.
- FinOps team adoption: ~46% of enterprises (up from 31% in 2024).
- Reserved Instance / Savings Plans coverage: 58% of steady-state compute spend.
- Median Egress / Data transfer cost: 12% of total spend (often overlooked).
- Top waste category: idle compute (EC2, RDS, Redshift) at 38% of recoverable waste.
- Median savings from FinOps function: 22% of total spend within 12 months.
Comparison: cloud cost waste by category
| Waste category | Share | Remediation | How to fix |
|---|---|---|---|
| Idle compute (EC2, RDS, Redshift) | 38% | Easy | Schedule stop/start; downsize; delete |
| Over-provisioned resources | 24% | Easy | Compute Optimizer + right-sizing |
| Untagged / orphaned resources | 12% | Medium | Tag enforcement + scheduled audits |
| Inefficient data transfer | 12% | Hard | VPC endpoints; CloudFront; S3 Transfer Acceleration |
| Storage tiering misconfigurations | 8% | Easy | S3 Intelligent-Tiering + Glacier |
| Unreserved workloads | 6% | Medium | Savings Plans + Reserved Instances |
Comparison: FinOps maturity stages
| Stage | Share | Capability | Operating model |
|---|---|---|---|
| Crawl | 28% | Tag-based visibility only | No formal function |
| Walk | 26% | Basic reporting + reserved capacity | Part-time FinOps lead |
| Run | 32% | Continuous optimisation + anomaly detection | Dedicated team of 2-5 |
| Fly | 14% | Predictive + unit-economics driven | Dedicated team + executive sponsor |
Reproducible code: rightsize EC2 fleet with Compute Optimizer
#!/usr/bin/env python3
# ec2_rightsize_with_compute_optimizer.py
import boto3, csv
co = boto3.client('compute-optimizer')
recommendations = []
paginator = co.get_paginator('get_ec2_recommendations')
for page in paginator.paginate():
for rec in page.get('instanceRecommendations', []):
recs = rec['recommendationOptions']
if not recs: continue
best = min(recs, key=lambda r: r['estimatedMonthlySavings'])
recommendations.append({
'instance_arn': rec['instanceArn'],
'current_type': rec['instance']['type'],
'recommended': best['instanceType'],
'monthly_savings_usd': best['estimatedMonthlySavings'],
'risk': best.get('performanceRisk', 'Unknown'),
})
with open('/wp-content/uploads/research/2026/cloud-infrastructure-cost-benchmark-report-2026.csv', 'w', newline='') as f:
w = csv.DictWriter(f, fieldnames=['instance_arn', 'current_type', 'recommended', 'monthly_savings_usd', 'risk'])
w.writeheader()
for r in recommendations: w.writerow(r)
total = sum(r['monthly_savings_usd'] for r in recommendations)
print(f"Found {len(recommendations)} rightsizing opportunities worth ~USD {total:,.0f}/month.")
Dataset
Download the full Cloud Infrastructure Cost Index:
- CSV: cloud-infrastructure-cost-benchmark-report-2026.csv
- JSON: cloud-infrastructure-cost-benchmark-report-2026.json
License: CC-BY-4.0. Cite as: SkilBrill Research (2026).
Recommendations
- Stand up a FinOps function — 22% savings within 12 months is the median outcome.
- Turn on AWS Compute Optimizer — free, low-risk, 4-8% savings in the first month.
- Set budget alerts with anomaly detection — catches runaway spend before it bills.
- Tackle data transfer costs — often the most-overlooked 12% of the bill.
Master cloud cost optimisation → Cloud Computing Training
Frequently asked questions
What is the median cloud waste percentage in 2026?
~28% of monthly cloud spend across surveyed enterprises is recoverable through rightsizing + Savings Plans.
How prevalent are FinOps teams?
~46% of surveyed enterprises report a dedicated FinOps function in 2026, up from ~31% in 2024.
What is the biggest cost-saving lever?
Idle compute — 38% of recoverable waste comes from EC2 / RDS / Redshift instances that are running but underused.
Should I use Savings Plans or Reserved Instances?
Savings Plans in 2026 — they’re more flexible (apply across instance families, regions, and tenancy) and offer comparable discounts.
How much can a FinOps function save?
Median 22% of total cloud spend within 12 months. Programs with mature FinOps see 30-40% savings.
What is the difference between Crawl / Walk / Run / Fly FinOps?
Crawl = tagging; Walk = reporting + reservations; Run = continuous optimisation; Fly = predictive + unit-economics-driven. Most enterprises are in Walk or Run.
