How ITAM and FinOps Bring Control to AI Spend

The shift is evolutionary, not revolutionary. The question is not whether you need new governance, but whether you are extending the governance you already have. 

ITAM: Re-establishing Commercial Control 

From an ITAM perspective, the first issue is often the lack of visibility into entitlements versus consumption. Many organisations do not have a clear view of which AI-related rights they already own and what they are paying for them. AI functionality now appears as embedded add-ons in existing licences; new SKUs introduced mid-term; usage-based entitlements; and features positioned as free that later convert into chargeable services.  

Without deliberate review, these changes pass quietly into the estate. ITAM can create immediate value by cataloguing AI entitlements, mapping them to live services and establishing commercial guardrails before usage accelerates. That means actively challenging assumptions such as whether AI features are optional or mandatory, whether bundled entitlements can be excluded, capped, or deferred, how usage thresholds translate into incremental costs, and what contractual rights vendors retain to change AI functionality or pricing.  

Once AI features are embedded into core workflows, the ability to step back significantly reduces. Commercial flexibility disappears at the same time dependency increases. This is where ITAM reasserts its role, not as inventory management but as commercial risk control. 

FinOps: Making AI Cost Visible and Debatable 

For FinOps teams, AI exposes structural gaps in cloud and SaaS cost management. AI usage is variable, experimentation-driven, and often disconnected from traditional production workloads. It does not follow predictable consumption curves. It spikes, scales quickly, and is frequently justified as strategic before its value is measured.  

This makes AI uncomfortable for FinOps. The temptation is to delay allocation and treat AI spend as shared overhead until patterns stabilise. That approach creates blind spots. Early allocation, even if imperfect, forces economic conversations. It highlights which teams are driving consumption, which use cases are scaling, and which experiments are failing quietly but expensively. Without this visibility, AI spend becomes culturally protected and politically difficult to challenge.  

FinOps teams should push for allocation by use case rather than just by department, clear separation between experimental and production AI spend, unit metrics that connect consumption to outcomes, even if loosely defined, and anomaly detection tuned to short-lived but high-cost AI workloads. The objective is not accounting precision but economic clarity. AI costs should be debatable, not set in stone. 

Translating AI Into Unit Economics 

In practical terms, this means converting AI consumption into units the business can interrogate. Total cloud or SaaS spend is too abstract to drive behavioural change. Unit economics create accountability. Examples include cost per inference, cost per prompt, cost per training run and cost per user interaction. These metrics do not need to be perfect to be useful. Their value lies in reframing the conversation from aggregate spend to cost per outcome. Without that shift, AI discussions remain conceptual and difficult to challenge. 

The Broader Control Question 

ITAM and FinOps already contain the foundations required to manage AI economics. What changes is the speed, volatility, and opacity of the cost model. Organisations that extend existing disciplines early will prevent experimental AI spend from hardening into embedded run rate cost. Those that wait often find themselves negotiating after dependency is established and leverage is reduced.  

Our eGuide, How AI Is Driving Software and Cloud Spend: Why AI Costs Accelerate Faster Than Value, explores how AI is reshaping commercial exposure across SaaS and cloud, where governance models are failing to adapt, and what practical control points make the biggest difference. If AI is now part of your operating model, its economics need to be part of it too. Download the eGuide to understand where ITAM and FinOps should intervene before AI spend becomes structural.