What the State of FinOps 2025 actually tells us

Every year the FinOps Foundation publishes the State of FinOps survey. The 2025 edition covered large cloud spenders responsible for more than $69 billion in cloud spend. It is one of the few datasets that reflects how FinOps is practiced at scale — not how vendors wish it were practiced.

Three themes stand out: optimization is still the #1 priority, the scope of FinOps is expanding beyond public cloud, and AI spend moved from niche to mainstream almost overnight.

Workload optimization remains the top priority

About 50% of practitioners still rank workload optimization and waste reduction as their primary focus — the same clear lead as the year before. Full allocation of cloud spend and accurate forecasting follow, but they trail optimization by a wide margin.

That persistence matters. It means most organizations have not “finished” rightsizing, idle cleanup, or commitment coverage. Optimization is not a project with an end date; it is a continuous operating loop.

FinOps is becoming Cloud+

The Framework 2025 update from the FinOps Foundation formally adds Scopes — public cloud, SaaS, AI, licensing, private cloud, and data centers — as a core concept. Teams are no longer asked only to manage AWS/Azure/GCP invoices.

• Roughly 40% of FinOps teams already manage SaaS spend; that share is expected to climb toward ~65% within a year.
• Private cloud / data-center cost management is also expanding (from roughly a quarter of teams toward nearly 40%).
• Practitioners juggle 12+ capabilities on average — which is why investment in tooling and automation is up (~20% year over year).

AI spend tracking doubled

63% of organizations now track AI-related spend, up from about 31% the year prior. Importantly, AI cost is usually additive — it does not replace existing cloud budgets. Most teams are still in the Inform phase for AI: visibility and allocation first, optimization second.

If your FinOps practice only watches EC2/VMs and ignores GPU training jobs, inference endpoints, or third-party model APIs, you are missing the fastest-growing line item on many engineering budgets.

What to do with this data

1. Keep a weekly optimization cadence (rightsizing, idle resources, commitment utilization) — the data says the work is not done.
2. Expand scope intentionally: pick one non-cloud scope (SaaS or AI) and bring it into the same showback rhythm as cloud.
3. Invest in automation where teams are overloaded — anomaly routing, commitment recommendations, and policy-as-code beat more spreadsheets.
4. Treat FOCUS (FinOps Open Cost and Usage Specification) as a normalization goal if you are multi-cloud; adoption intent among practitioners is high even if vendor support is uneven.