Cloud spending hit $723.4 billion in 2025. Flexera says 27% of it is wasted. That is nearly $200 billion in idle compute, over-provisioned storage, and forgotten services.
I was reading the Flexera 2025 State of the Cloud report a few weeks ago, and the 27% cloud waste number stayed with me (https://www.flexera.com/blog/cloud/cloud-computing-trends-tech-spend-pulse/). Not because the percentage is new, Flexera has been tracking this figure for years, but because of what it means in dollar terms right now. Gartner projects worldwide public cloud end-user spending at $723.4 billion in 2025 (https://www.gartner.com/en/newsroom/press-releases/2026-04-07-gartner-forecasts-worldwide-it-spending-to-grow-9-8-percent-in-2026). Twenty-seven percent of that is roughly $195 billion. That is not rounding error. That is more than the annual GDP of a mid-sized country, sitting in idle compute instances, over-provisioned storage tiers, and services nobody actively uses. The same Flexera report found that 84% of organizations say managing cloud spend is their top challenge, and that organizations are exceeding their cloud budgets by an average of 17%.
I want to try to explain why this situation persists, because from the outside it can look like simple organizational negligence. It is more structurally interesting than that.
The first structural reason waste is hard to eliminate is that cloud provisioning and cloud decommissioning have very different friction profiles. Provisioning a virtual machine, a database, a managed service, takes minutes and often requires nothing more than the right credentials and a command-line call. Decommissioning what you provisioned six months ago requires knowing it exists, knowing whether any active workloads depend on it, getting someone to approve the shutdown, and actually doing it. The friction goes in one direction. Organizations optimize for the thing that is easy, which is provisioning, and they systematically underinvest in the harder side of the lifecycle.
The second structural reason is that cloud pricing is genuinely opaque. The major providers have pricing structures that reward sophisticated buyers who invest in understanding reserved instances, savings plans, committed use discounts, spot pricing, and tiered storage classes. Organizations that do not have the expertise to navigate these structures are effectively paying a premium for the same compute that a more sophisticated buyer gets cheaper. The provider's incentive is not to help you optimize your bill. The provider's incentive is to run your workloads. The bill is your problem to understand, and understanding it is a non-trivial skill that many organizations have not built.
The third structural reason is that cloud spend decisions are distributed while the waste problem is aggregate. A product team spins up a test environment. A data team provisions an analytics cluster for a quarterly project. A developer leaves staging infrastructure running because decommissioning it is technically someone else's job and nobody is enforcing the policy. Each of these individual decisions has a reasonable local justification. The accumulated result, visible only at an organizational level, is a cloud bill with 27% waste embedded in it that no single team owns or feels responsible for.
This is where I find the DeLone and McLean IS Success Model (2003) genuinely useful as an analytical frame. DeLone and McLean argued that IS success depends on system quality, information quality, service quality, and the organizational processes that turn those into use and ultimately into net benefit. The interesting thing about cloud cost governance is that the "system quality" component, the tooling for tracking spend, tagging resources, generating anomaly alerts, recommending rightsizing, is commercially mature. AWS Cost Explorer, Azure Cost Management, and third-party platforms from vendors like CloudHealth and Apptio are real, functional tools. The information quality is available. The service quality, through FinOps consultants and managed service providers, is available.
The failure is in the organizational process layer that DeLone and McLean would call "use" and "organizational impact." The tools exist. The information exists. The question is whether the organizational routines exist to turn that information into action. In most organizations, the answer is partial at best. There are people who can see the waste. There are fewer people whose job it is to systematically eliminate it. And there are even fewer organizational structures that hold business units accountable for the cloud costs they generate.
The Flexera report signals that the market is starting to respond to this. 59% of organizations say they are expanding their FinOps teams in the coming year. 28% expect cloud spend to increase further. Those two things together suggest organizations are investing in governance capacity at the same time as the problem grows. That is a reasonable response, but it is a catch-up game. The waste problem is not new. The 27% figure has been roughly stable for years even as the base it applies to has grown substantially. The dollar value of waste in 2025 is much larger than the dollar value of waste in 2020, not because organizations have gotten worse at governing cloud, but because the cloud market itself has grown faster than governance capabilities have improved.
As an IS researcher, what this points to for me is that cloud waste is fundamentally a question of absorptive capacity (Cohen and Levinthal 1990): the organizational ability to recognize information about cloud spend, assimilate it into decision-making processes, and apply it to change behavior. Building absorptive capacity is not something tooling alone achieves. It requires organizational learning, role clarity, incentive alignment, and management attention. An organization that buys a FinOps platform without investing in the organizational structures that allow it to act on what the platform shows them has not solved the problem. They have made the problem visible, which is a precondition for solving it, but not sufficient.
What genuinely worries me is the trajectory from here. Gartner's April 2026 forecast projects public cloud growing to $1.48 trillion by 2029 (https://www.gartner.com/en/newsroom/press-releases/2026-04-07-gartner-forecasts-worldwide-it-spending-to-grow-9-8-percent-in-2026). If the waste percentage stays near 27%, the absolute dollar value of wasted cloud spend in 2029 will be somewhere around $400 billion. That is a number that makes the current situation look modest. The organizations that build serious cloud governance capabilities now are not just solving today's problem. They are building a structural advantage relative to the ones that are still treating waste as an inevitable cost of doing business in the cloud.
My research question coming out of this is not "why is there cloud waste?" That answer is structurally clear. My question is: what distinguishes the organizations that successfully close the gap between spend and governance, and what does that organizational capability actually look like from an IS perspective? The 16% of digital transformation efforts that McKinsey finds succeed long-term (https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai) share something about how they build and sustain organizational change. I suspect the FinOps success stories have a similar signature.
---
claims_checked:
- "Cloud waste 27% (Flexera 2025)": "https://www.flexera.com/blog/cloud/cloud-computing-trends-tech-spend-pulse/"
- "84% struggle to manage cloud spend (Flexera 2025)": "https://www.flexera.com/blog/cloud/cloud-computing-trends-tech-spend-pulse/"
- "Budgets exceeded by 17% (Flexera 2025)": "https://www.flexera.com/blog/cloud/cloud-computing-trends-tech-spend-pulse/"
- "59% expanding FinOps teams (Flexera 2025)": "https://www.flexera.com/blog/cloud/cloud-computing-trends-tech-spend-pulse/"
- "28% expect cloud spend to increase (Flexera 2025)": "https://www.flexera.com/blog/cloud/cloud-computing-trends-tech-spend-pulse/"
- "Public cloud $723.4B in 2025, $1.48T by 2029 (Gartner April 2026)": "https://www.gartner.com/en/newsroom/press-releases/2026-04-07-gartner-forecasts-worldwide-it-spending-to-grow-9-8-percent-in-2026"
- "McKinsey 70% DX failure, 16% succeed long-term": "https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai"
claims_unverified:
- "Rough $195B waste calculation: derived arithmetic from two verified figures, not a directly cited number"
- "$400B projected 2029 waste: derived arithmetic, directional illustration only"
sources_used:
- "https://www.flexera.com/blog/cloud/cloud-computing-trends-tech-spend-pulse/"
- "https://www.gartner.com/en/newsroom/press-releases/2026-04-07-gartner-forecasts-worldwide-it-spending-to-grow-9-8-percent-in-2026"
- "https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai"
word_count: 1140
About the author
Share
More notes
Related notes