Skip to content

29% of cloud spend is wasted, and AI is pushing it up

Mykhailo PoliarushUpdated 5 min read

In 2026, the cloud is growing faster than most teams can run it well. Spending on cloud infrastructure services reached an estimated $167.7 billion in Q3 2026, up 55% year over year, according to Synergy Research Group, and respondents to Flexera's 2026 State of the Cloud survey estimate that 29% of their cloud spend is wasted. The hard problems in cloud operations are now cost, reliability and the people and processes behind them, not capacity.

How fast is cloud spending growing in 2026?

Fast, and mostly because of AI. Synergy reports that market growth accelerated for the twelfth quarter in a row and that AI-specific cloud services now make up about one-third of the market, with trailing 12-month revenue at $558 billion. Amazon held 26% of the Q3 market, Microsoft 19% and Google 16%, per the same update.

Bills are large even for a single company. More than three-quarters (76%) of large enterprises now spend over $5 million a month on cloud services, according to Flexera, whose survey covered more than 750 cloud decision-makers and users. Flexera sells FinOps software, so read its figures as one vendor's survey.

Why is cloud waste rising again?

AI workloads. Estimated waste rose to 29% in Flexera's 2026 report, the first increase in five years, from 27% in the 2025 report. These are respondents' own estimates, not audited bills. In 2025, cloud budgets were already exceeded by 17%, per Flexera's 2025 release.

AI has become the main FinOps job. In the FinOps Foundation's State of FinOps 2026 (1,192 respondents managing more than $83 billion in annual cloud spend), 98% of practitioners now manage AI spend, up from 31% two years ago, and AI cost management is the number one skill teams want to add. Many teams are asked to fund AI out of optimization savings, which makes optimization more urgent just as the easy savings run out. Workload optimization and waste reduction are still the top current priority for FinOps teams, but the same report notes diminishing returns: practitioners say they have already cleared the big items and now face many small ones that take more effort each. Their scope also keeps growing beyond public cloud, into SaaS, licensing, private cloud and data centers.

Technology categories FinOps teams manage, 2026

AI spend
98%
SaaS (manage or plan to)
90%
Software licensing
64%
Private cloud
57%
Data center
48%
Labor costs (including or planning to)
28%

Source: FinOps Foundation, State of FinOps 2026 (n=1,192)

How reliable is the cloud, and what do outages cost?

Big providers still fail, and the failures spread. On 19 and 20 October 2025, the AWS us-east-1 region was disrupted from 11:48 PM to 2:20 PM PDT, about 14.5 hours, after what AWS describes as a latent defect in DynamoDB's automated DNS management system. Services that depended on DynamoDB, including EC2 launches, Lambda and load balancers, were hit in turn.

Outages are expensive. In Uptime Institute's Annual Outage Analysis 2026, 57% of respondents to Uptime's 2025 annual survey said their most recent major outage cost more than $100,000, and for the second year in a row one in five said it cost more than $1 million. Uptime describes outage costs as continuing to rise slowly (Uptime 2026).

Confidence is slipping. In Uptime's 2026 resiliency survey, 24% said public cloud is not resilient enough for any of their mission-critical workloads, up six percentage points in a year, and only 12% trust it with all of them (Uptime, 2026). Nearly one in five organizations (18%) now insure against outages involving cloud providers.

Is public cloud resilient enough for your mission-critical workloads?

Yes, for all of them
12%
For most of them
28%
For only some of them
36%
Not for any of them
24%

Source: Uptime Institute, Data Center Resiliency Survey 2026, in Annual Outage Analysis 2026 (n=372, Feb to Mar 2026)

What are the main challenges in cloud operations?

  1. Cost control. Managing cloud spend is a top challenge for 85% of organizations, according to Flexera 2026.
  2. Running AI securely. In the same survey, 53% of cloud leaders name security and compliance as their top challenge for cloud-based AI. Large enterprises are responding with structure: 47% are setting up dedicated AI governance teams or leaders, per Flexera.
  3. Process and human error. 92% of respondents say human error contributed at least in part to their most recent impactful outage, and the top cause, at 59%, is staff not following procedures (Uptime 2026).
  4. Skills. AI cost management tops the skills FinOps teams need (FinOps Foundation), and 25% of respondents in Uptime's survey name insufficient staff as a way human error led to data center outages.
  5. Dependency on one provider or region. The AWS event showed how one defect in a shared service can cascade across many services in a region (AWS).

What does good cloud operations look like in 2026?

Most outages can be prevented. In Uptime's 2025 data center survey, 87% of respondents who had an impactful outage said better management, processes or configuration could have avoided it (Uptime, n=98). The practices that make the difference are not exotic:

  • Make cost an engineering metric. Tag every resource, allocate costs to teams and products, and track unit cost (per customer, per order, per model call). FinOps is now mainstream: 63% now have FinOps teams and 71% run a Cloud Center of Excellence (Flexera 2026).
  • Put AI spend under the same controls. Budgets, alerts and owners per model and per workload, reviewed weekly while usage is growing.
  • Define SLOs for what matters. Pick the user journeys that make money, set service level objectives for them and alert on burn rate, not on every CPU spike.
  • Design for the region failing. Map dependencies, run critical paths across availability zones, and decide in advance which services need a second region.
  • Codify and test changes. Infrastructure as code, policy as code and automated checks on every change. Runbooks that are rehearsed, not just written.
  • Learn from every incident. Blameless reviews with actions that are tracked to done.

Many companies get there with outside help: in 2025, 60% of organizations were turning to managed service providers to regain control of cloud spending, per Flexera.

A practical first month: allocate the bill, fix the ten largest sources of waste, set SLOs for your three most important services, and review your last three incidents for process gaps.

Need a team to run your cloud?

SDCLabs is a software engineering company founded in 2011, with offices in Krakow and Dublin. Senior engineers run AI-assisted workflows, and automated tests verify every change. We can work as a dedicated team, in close cooperation with yours, or as managed staff, and engagements start in 1 to 2 weeks. See our cloud operations and managed services or contact us.

Related: our DevOps and TestOps and managed support services.

Sources checked on 2026-11-17.

Frequently asked questions

How much cloud spend is wasted in 2026?

Respondents to Flexera's 2026 State of the Cloud survey estimate that 29% of their cloud spend is wasted, the first increase in five years, which Flexera links to surging AI workloads. These are self-estimates, not audited bills.

How big is the cloud market in 2026?

Spending on cloud infrastructure services reached an estimated $167.7 billion in Q3 2026, up 55% year over year, according to Synergy Research Group. AI-specific services make up about one-third of the market.

How much does a major outage cost?

In Uptime Institute's Annual Outage Analysis 2026, 57% of respondents to its 2025 annual survey said their most recent major outage cost more than $100,000, and one in five said more than $1 million.

Can most cloud and data center outages be prevented?

Uptime's data suggests so: 87% of respondents who had an impactful outage said better management, processes or configuration could have avoided it, and 92% say human error contributed at least in part (Uptime 2026).

Related posts

Get an estimate for your product.