Cloud waste has risen for the first time in five years, according to Flexera, which estimates that 29% of infrastructure as a Service (IaaS) and Platform as a Service (PaaS) spend is now wasted as AI workloads and newer cloud services make forecasting harder.
Here, “cloud waste” means the share of IaaS and PaaS spending that survey respondents themselves estimated was wasted, based on their own view of how much cloud spend did not deliver useful value or was avoidable.
The company’s 2026 State of the Cloud report surveyed 753 cloud decision-makers and users, 62% of them in the U.S.
Flexera said wasted cloud spend had been falling for five years, but dynamic AI usage, harder rightsizing decisions and new pricing metrics are now making cost visibility and optimization harder again.
In the same report, 81% of respondents said they use GenAI extensively or sparingly, 45% said they use it extensively and usage of GenAI public cloud services rose to 58% from 50% a year earlier.
AI changes the cost model
Flexera’s challenge data points in the same direction. Security and compliance risks were the top challenge for 53% of respondents scaling AI workloads in the cloud, followed by data quality for model training at 40%. Skills gaps and cost unpredictability each ranked at 30%, while Flexera said dynamic usage makes AI costs hard to forecast and new pricing metrics complicate visibility.
Flexera said the most common SaaS spending band moved to $200,001 to $500,000 a month in 2026 from $50,001 to $100,000 a month in 2025. Separately, Gartner said worldwide software spending is projected to grow 14.7% in 2026, with GenAI model spending still expected to rise 80.8%.
Governance catches up
Flexera said 71% of organizations now have a cloud center of excellence and 63% have a FinOps team, while 64% said they assess cloud progress by value delivered to business units, up 12 percentage points year over year.
Large enterprises are also putting more formal AI oversight in place: 47% now have a dedicated AI governance team or leader, and 85% report a dedicated team or senior leader responsible for AI oversight.
That governance shift lines up with the FinOps Foundation’s latest data. In its State of FinOps 2026 report, the foundation said FinOps for AI is now the top forward-looking priority and AI cost management is the No. 1 skill teams need to develop, with 98% of respondents now managing AI spend.
Its guidance on forecasting AI services says teams should use a weekly or monthly forecasting cadence because AI costs are highly variable, and its practice-operations guidance recommends tracking costs down to the token or GPU level and setting budget thresholds for fast-moving AI projects.
Flexera said nearly half of organizations now use unit economics to track cost per service, but it also said fewer than half use any one commitment discount per cloud provider and more than half still rely on on-demand pricing, a pattern the report places alongside the rise in waste.