AI ECONOMICS

How AI Cost, Value and Pricing Connect

What AI costs, what it is worth and how it is paid for are changing at the same time, and each one moves the other two.

Every AI decision comes down to three questions: what does it cost to run, what does it deliver, and how is it charged? Most organizations can answer one of the three. This page explains how they connect, and who captures the benefit when one of them moves.

The three forces

Cost:

what does AI actually cost? Model usage (tokens), computing, the software around it, people to run it and ongoing operations. The advertised token price is only part of it. Evidence: in September 2026, OpenAI priced its new lower-cost GPT-6 models at half the API price of the versions they replace, and Anthropic's Claude Opus 5.5 list prices are 20% below Opus 5 (see AI Models). FOCUS, the FinOps Foundation's standard for cost data, released version 1.4 in June 2026; version 1.5, scheduled for ratification on December 3, 2026, adds common labels for AI token billing across providers.

Sources:

Value:

what business benefit does AI create? Productivity, revenue, customer experience, efficiency, and lower risk, measured against a starting baseline and confirmed after the fact. This is the hardest of the three to prove, and the one most often assumed rather than measured.

Pricing:

how is AI paid for? Seats, credits, tokens, agent actions, conversations, resolutions and contracted outcomes. Pricing sets what is counted; value measurement shows whether the customer came out ahead. Evidence: see the AI Pricing battleground for how Salesforce, Zendesk, Microsoft and others are mixing these units.

How they connect

For the supplier: cost per unit of work = total operating cost ÷ completed units of work.
For the customer: return on investment = (benefits realized − costs) ÷ costs, measured over the same period.
For the supplier's business: supplier gross margin = (revenue − cost of revenue) ÷ revenue.

Pricing decides how the gap between the supplier's cost and the customer's benefit is shared. Lower costs do not automatically mean lower prices or more benefit for the customer.

Cost and Pricing: supplier margin
Value and Pricing: what the customer is willing to pay
Cost and Value: benefit relative to cost
Center: who captures the benefit

The question to sit with (hypothetical, not a forecast)

If the cost of running AI fell by 80%, who would keep the savings: the model maker, the software vendor, the platform, or the customer?