Put Tokens in the P&L, Not in a Developer Expense Report
Token costs are production costs, not developer allowances. Put them in the portfolio P&L, measure human plus AI labor against accepted outcomes, and stop mistaking visible inference bills for economic governance.
AI labor is a production input. It should be measured against portfolio value, not audited as individual consumption.
Govern AI cost at the level where value is created.
A visible input cost is not a financial model. Token spend only becomes useful when it is tied to the accepted outcome, the business metric, and the fully loaded cost of the alternative production path.
AI labor and human labor belong in the same financial model for production. Separate accounting for inference usage, engineering time, consultant work, rework, delay, and review creates local savings and portfolio waste.
Resource allocation should follow value, not equal access. Scarce AI capacity is capital, and capital should move toward the initiatives and operators producing the highest measurable return.
Procurement can improve commercial terms, but it cannot own delivery economics. A lower unit price that reduces output quality, increases rework, or slows delivery is cost displacement, not cost control.
Governance belongs at the portfolio level. Budget envelopes, stop-loss rules, value owners, and actual outcome tracking create better controls than making developers ration the input required to do the work.
The first question for any AI spend is not who used it. The first question is whether the portfolio converted it into measurable value.
AI labor is a production input. It should be measured against portfolio value, not audited as individual consumption.
Govern AI cost at the level where value is created.
A visible input cost is not a financial model. Token spend only becomes useful when it is tied to the accepted outcome, the business metric, and the fully loaded cost of the alternative production path.
AI labor and human labor belong in the same financial model for production. Separate accounting for inference usage, engineering time, consultant work, rework, delay, and review creates local savings and portfolio waste.
Resource allocation should follow value, not equal access. Scarce AI capacity is capital, and capital should move toward the initiatives and operators producing the highest measurable return.
Procurement can improve commercial terms, but it cannot own delivery economics. A lower unit price that reduces output quality, increases rework, or slows delivery is cost displacement, not cost control.
Governance belongs at the portfolio level. Budget envelopes, stop-loss rules, value owners, and actual outcome tracking create better controls than making developers ration the input required to do the work.
The first question for any AI spend is not who used it. The first question is whether the portfolio converted it into measurable value.
After 20 years in software development, Norman is both a hands-on leader and defining the new age of AI SDLC for some of the biggest brands in the world — and exploring it with the builders. He writes here about things he is hearing and seeing. All posts are his personal points of view and do not reflect any employer or any customer he has ever had contact with.
The views and opinions expressed in this article are the author’s own and do not represent the positions of any employer, client, or affiliated organization.
