KPMG survey: only 26% of companies have a comprehensive view of their AI costs, while 50% have some visibility and 22% have none or only see costs after billing
Only 26% of companies say they have a comprehensive view of their AI costs, according to a new survey
Context & Ripple Effects
Related executive surveys show AI spending plans continuing to rise even as fewer than half of projects are reported to generate returns above their costs. A separate IBM survey similarly found limited rates of expected ROI and enterprise-wide scaling.
This makes cost observability a practical constraint on the move from experimentation to repeatable deployment: companies cannot reliably assess project economics when spending is only partially visible or arrives after billing.
First-order effects
- Most surveyed companies must make AI budget and project decisions with incomplete or delayed cost data, limiting their ability to attribute spend to particular initiatives.
- Teams responsible for AI deployments face immediate pressure to add cost tracking and reporting before they can credibly evaluate ROI or scale usage.
Second-order effects
- Rising AI investment plans will put greater scrutiny on vendors, internal platform teams, and procurement processes to provide clearer usage and billing visibility.
- Projects with uncertain economics may face slower expansion or tighter approval controls, particularly where the reported returns have not exceeded costs.
Third-order effects
- If incomplete cost visibility persists, enterprise AI adoption may increasingly be governed by financial controls and measurable unit economics rather than experimentation alone.
- The pattern points toward AI operations maturing into a discipline that combines deployment oversight, cost allocation, and ROI measurement; whether that improves scaling will depend on organizations translating visibility into decisions.
The trend: Enterprise AI is shifting from an experimentation-and-spending phase toward a governance phase in which cost transparency and demonstrated returns determine which deployments scale.