Global public cloud spending reached $679.0 billion with $190.0 billion wasted on idle resources (32.0% waste rate), 71.0% of enterprises establishing FinOps teams, and AI workloads driving a +42.0% surge in compute expenditure. While 58% of companies exceed cloud budgets by up to 21% and 36% of virtual machines run at under 10% CPU utilization, FinOps practices slash annual cloud costs by 24%. The figures below come from empirical research published by Gartner, Flexera, the FinOps Foundation, Datadog, a16z, and McKinsey & Company.
TL;DR
- Global public cloud spending reached $679.0 billion annually, growing at +20.4% YoY (Gartner)
- 58.0% of enterprise organizations report exceeding their allocated public cloud budgets (Flexera)
- Cloud budget overruns average +15.0% to +21.0% above initial financial forecasts
- 28.0% to 32.0% of total enterprise cloud spend is completely wasted on idle or orphaned resources
- Over $190.0 billion is wasted annually on unutilized cloud compute and storage globally (Gartner)
- 36.0% of cloud virtual machine instances operate at under 10% average CPU utilization (Datadog)
- 71.0% of enterprise organizations have established dedicated FinOps cloud financial management teams
- Adopting formal FinOps practices delivers an 18.0% to 24.0% reduction in cloud costs in year one (McKinsey)
- Generative AI training and inference workloads drove a +42.0% YoY surge in cloud compute spend (a16z)
- 68.0% of Generative AI startups spend 50% or more of their total funding on cloud GPU compute
- 87.0% of enterprise organizations operate multi-cloud architectures across AWS, Azure, and Google Cloud
- 34.0% of enterprises are evaluating partial cloud repatriation to on-premise or colocation facilities
- Hyperscalers (AWS, Azure, GCP) waived data egress migration fees in compliance with the EU Data Act
1. Global Cloud Spend: $679B Market and 58% Budget Overruns
Enterprise digital transformation and machine learning infrastructure have accelerated cloud hyperscaler revenue expansion. Gartner and IDC record global public cloud spend at $679.0 billion (+20.4% YoY growth).
Cost overruns are chronic: 58.0% of enterprises exceed allocated cloud budgets (Flexera), overshooting financial projections by +15.0% to +21.0% due to unmonitored infrastructure scaling.
| Metric | Value | Source |
|---|---|---|
| Global enterprise public cloud end-user spending across IaaS, PaaS, and SaaS | $679.0 Billion worldwide | Gartner Worldwide IT Spending / IDC |
| Annual growth rate in enterprise public cloud infrastructure expenditures | +20.4% YoY spending growth | Gartner Cloud Forecasts |
| Enterprises reporting that public cloud spending exceeded their original allocated budgets | 58.0% of enterprises over budget | Flexera State of the Cloud Report |
| Average percentage by which enterprise cloud spending exceeds initial budget forecasts | +15.0% to +21.0% over budget | Flexera Cloud Survey |
Data center infrastructure scale connects to our data center statistics. Source: Gartner Worldwide IT Spending Forecast.
2. The $190B Waste Epidemic: 32% Cloud Waste and Zombie VMs
Decentralized developer provisioning without automated decommissioning creates vast operational waste footprints. Flexera and HashiCorp calculate that 28.0% to 32.0% of total cloud spend is wasted.
Capital destruction is massive: $190.0+ billion is wasted annually (Gartner), with Datadog identifying 36.0% of virtual instances running as ‘Zombie VMs’ under 10% CPU and 44.0% of environments retaining orphaned block storage.
| Metric | Value | Source |
|---|---|---|
| Estimated public cloud spend completely wasted on idle, over-provisioned, or orphaned resources | 28.0% to 32.0% of total cloud spend wasted | Flexera State of the Cloud Report / HashiCorp |
| Annual financial value of enterprise public cloud waste globally | $190.0B+ in wasted cloud spend/yr | Gartner / Flexera Analysis |
| Cloud instances and virtual machines operating at under 10% average CPU utilization (‘Zombie VMs’) | 36.0% of active cloud instances | Datadog State of Cloud Costs Report |
| Orphaned unattached block storage volumes (AWS EBS / Azure Disks) continuing to bill monthly | 44.0% of cloud environments | FinOps Foundation Benchmark |
AI computing cost dynamics connect to our ai inference cost statistics. Source: Flexera State of the Cloud Report.
3. FinOps Institutionalization: 71% Adoption and 24% Year-One Savings
Cross-functional Cloud Financial Management has emerged as a mandatory executive discipline. The FinOps Foundation reports that 71.0% of enterprise companies have established formal FinOps teams.
Executive oversight strengthens: 46.0% of FinOps leaders report directly to the CFO, with structured FinOps governance delivering 18.0% to 24.0% first-year cloud cost reductions (McKinsey).
| Metric | Value | Source |
|---|---|---|
| Enterprise organizations that have established a formal FinOps (Cloud Financial Management) team | 71.0% of enterprise companies | FinOps Foundation State of FinOps Report |
| FinOps teams reporting directly to the Chief Financial Officer (CFO) or VP of Finance | 46.0% reporting to Finance/CFO | FinOps Foundation Annual Survey |
| Average annual cloud spend cost reduction achieved in the first year of adopting FinOps practices | 18.0% to 24.0% cost reduction | McKinsey Cloud Economics Study |
Enterprise software budgeting connects to our saas spending statistics. Source: FinOps Foundation State of FinOps.
4. The Generative AI Impact: +42% Compute Surge and Startup Burn
Large Language Model training and real-time token inference have transformed cloud budget allocation. a16z and Stanford HAI track a +42.0% YoY increase in AI cloud workloads.
Venture capital cycles directly to cloud: 68.0% of Gen AI startups allocate 50%+ of total funding to compute (Menlo Ventures), driving 54.0% of AI teams to adopt preemptible Spot instances (Datadog).
| Metric | Value | Source |
|---|---|---|
| Surge in enterprise cloud compute spend driven specifically by Generative AI training and inference | +42.0% YoY increase in AI cloud workloads | a16z AI Infrastructure Report / Stanford HAI |
| Enterprises utilizing Spot instances and preemptible VMs to cut AI batch processing costs | 54.0% of AI engineering teams | Datadog State of Cloud Report |
| AI startups spending 50% or more of total venture funding on cloud compute (AWS, CoreWeave, Azure, GCP) | 68.0% of Gen AI startups | Menlo Ventures The State of GenAI |
Enterprise AI implementations connect to our enterprise ai adoption statistics. Source: a16z AI Infrastructure Report.
5. Pricing Commitments: 62% Reserved Hours vs. Multi-Cloud Friction
Financial engineering through multi-year commitments unlocks massive discount tiers from cloud providers. Flexera data shows 62.0% of cloud compute hours run on Savings Plans or Reserved Instances (RIs).
Commitment risk persists: 38.0% of firms waste unutilized RI commitments (CloudZero), while managing 87.0% multi-cloud architectures across AWS, Azure, and Google Cloud.
| Metric | Value | Source |
|---|---|---|
| Enterprise cloud compute capacity running on discounted Savings Plans or Reserved Instances (RIs) | 62.0% of total cloud compute hours | Flexera State of the Cloud |
| Enterprises that fail to fully utilize purchased Reserved Instances, resulting in unallocated commitment waste | 38.0% of RI buyers | CloudZero SaaS Cost Management |
| Enterprises deploying multi-cloud architectures across two or more hyperscalers (AWS, Azure, Google Cloud) | 87.0% of enterprise organizations | Flexera Cloud Report |
Hardware server shipments connect to our server market statistics. Source: CloudZero SaaS Cost Management.
6. Cloud Repatriation and Egress Fee Reforms: The EU Mandate
Escalating baseline infrastructure bills have prompted enterprises to re-evaluate hybrid on-premise architectures. IDC records that 34.0% of enterprises are evaluating partial cloud repatriation.
Egress friction dismantled: 52.0% cite egress fees as major cost barriers (Gartner), driving AWS, Azure, and GCP to permanently waive migration egress fees under EU Data Act compliance.
| Metric | Value | Source |
|---|---|---|
| Enterprises evaluating or actively executing partial cloud repatriation (moving predictable workloads on-prem/colocation) | 34.0% of enterprises evaluating repatriation | IDC Cloud Infrastructure Survey / 37signals Case Study |
| Companies citing excessive network egress data transfer fees as primary driver of cost friction | 52.0% of enterprise IT leaders | Gartner Infrastructure & Operations |
| Major hyperscalers (AWS, Google Cloud, Azure) waiving data egress fees for customers migrating away | 100% of Big 3 hyperscalers (mandated by EU Data Act) | AWS / Google Cloud / Microsoft Announcements (2024-2026) |
Summary: Cloud Costs by the Numbers
| Metric | Value | Primary Source |
|---|---|---|
| Global public cloud spend | $679.0 Billion | Gartner IT Spending / IDC |
| Annual cloud spend growth rate | +20.4% YoY | Gartner Forecasts |
| Enterprises exceeding cloud budget | 58.0% | Flexera State of the Cloud |
| Average cloud budget overrun | +15.0% - +21.0% | Flexera Cloud Survey |
| Estimated wasted cloud expenditure | 28.0% - 32.0% | Flexera / HashiCorp |
| Annual global cloud waste value | $190.0B+/year | Gartner / Flexera |
| Zombie VMs with <10% CPU usage | 36.0% of instances | Datadog Cloud Report |
| Firms with formal FinOps teams | 71.0% of enterprises | FinOps Foundation |
| Cost reduction in year 1 of FinOps | 18.0% - 24.0% | McKinsey Cloud Study |
| AI compute workload spend surge | +42.0% YoY | a16z / Stanford HAI |
| Gen AI startups spending >50% on compute | 68.0% | Menlo Ventures Report |
| Compute hours on Savings Plans/RIs | 62.0% | Flexera Cloud Report |
| Enterprises on Multi-Cloud setups | 87.0% | Flexera State of Cloud |
| Enterprises exploring cloud repatriation | 34.0% | IDC Infrastructure |
| Hyperscalers waiving egress fees | 100% (AWS, GCP, Azure) | Hyperscaler Disclosures |
Methodology and Sources
The statistics in this report were compiled from international cloud computing market datasets from Gartner and IDC, enterprise cloud cost benchmarks from Flexera and HashiCorp, operational research from the FinOps Foundation, infrastructure telemetry from Datadog and CloudZero, and AI infrastructure studies from a16z and McKinsey.
-
Gartner & IDC: Worldwide Public Cloud Services Spending & Infrastructure Forecasts ($679B market, +20.4% YoY growth, egress friction).
-
Flexera: State of the Cloud Report: Cloud Cost Management & Waste (28-32% waste, 58% over-budget, 87% multi-cloud).
-
FinOps Foundation: State of FinOps Annual Research Report (71% adoption, 46% CFO reporting, orphaned storage).
-
Datadog: State of Cloud Costs: Compute Utilization and Modern Infrastructure Trends (36% Zombie VMs, Spot adoption).
-
a16z & Stanford HAI: The Economic Architecture of AI Infrastructure & Cloud Costs (+42% AI workload surge, 68% startup compute spend).
-
McKinsey & Company & IDC: Cloud Economics, Workload Repatriation, and FinOps ROI (18-24% FinOps savings, 34% repatriation interest).
-
Data watch: Cloud Cost statistics reflect public cloud IaaS, PaaS, and SaaS infrastructure spending across AWS, Microsoft Azure, Google Cloud, and specialized GPU clouds. Private on-premise datacenter capital expenditures are excluded unless explicitly noted.
-
Last updated: August 2026. This roundup is updated quarterly as Gartner spending releases, FinOps Foundation state reports, and cloud provider pricing filings are published.