Cost per Thought (Full Stack)
Definition
Measures the fully-loaded cost of one unit of AI work represented by one end-to-end AI request cost, including all infrastructure overhead, giving a single number for budgeting to Leadership and FinOps Practitioner Personas .
Formula, as published
Cost per Thought = Total AI Spend / Total LLM Observability Traces
- Where:
- A “trace” is a record that captures a complete timeline of an agent’s behavior, including timestamps, errors, inputs/outputs, and contextual metadata.
- Data sources:
- Cloud Cost Billing Data Usage Telemetry (eg. Azure Cognitive Services metrics) LLM Observability/Application Telemetry
Calculate it
Example values to start with; replace them with yours.
How to measure it
Full stack means everything an AI feature costs, not just tokens: observability, evaluations, vector storage, idle provisioned capacity. Divided by traces, it is the cost of one completed piece of work, which is the unit a product owner can price.
From the FinOps Foundation
More Budgeting KPIs
Definition and formula: Cost per Thought (Full Stack), FinOps KPI library, FinOps Foundation, licensed CC BY 4.0, read 16 Oct 2026. Changed: HTML removed, dashes and British spellings changed, formula split from its data sources and examples. The calculator and the notes are Finitizer’s.
Measured once, a KPI is a number. Tracked daily, it is a practice.
Finitizer computes the KPIs your billing data supports every day, by team and product, across AWS and Google Cloud.
See how Finitizer works