The AI unit economics calculator.

What does fully loaded mean?

Fully loaded means the whole cost of serving an AI product, not just the token bill: model and API invoices plus the cloud infrastructure that exists to serve AI features — GPU instances, vector databases, orchestration, logging, evaluation pipelines. ServiceNow CFO Gina Mastantuono has put AI reasoning at less than 10% of the company’s cost to serve; the rest lives on the cloud bill. Where to draw the boundary is covered in our guide to what counts as AI COGS.

How are the unit economics calculated?

Each result divides the fully loaded cost by a business denominator. The denominators are the point: spend alone says nothing about value, but cost per MAU, cost per interaction, and AI gross margin can be set against pricing, contribution margin, and what a board expects.

For the full formulas, worked examples, and how these metrics behave as products scale, see the guide to AI unit economics.

Why start from actual spend rather than token prices?

Because forecasts and measurements are different instruments. Token-price calculators — which multiply expected usage by published per-token rates — are genuinely useful for estimating a feature before it ships. They are the wrong tool for measurement, because production spend drifts from list-price arithmetic: model mix shifts, retries and caching change consumption, negotiated rates differ from rate cards, and infrastructure never appears on a token bill. Uber exhausted its 2026 budget for AI coding tools by April; Priceline saw a four-to-five-fold cost increase at a single contract renewal. Measured spend divided by measured usage is the arithmetic that survives contact with the invoice.

Cost tools tell you what you spent. AI Value Management tells you what it bought — and unit economics is the exchange rate. This calculator is a one-month snapshot assembled by hand; COGScontrol’s attribution and unit-economics features do the same arithmetic continuously, aggregating AI and cloud invoices into one ledger, attributing them across cost centers and product lines, and reconciling to invoice daily, on a fixed subscription — never a percentage of spend. For the broader discipline, start with measuring the ROI of AI initiatives.