Technology is one of the largest lines on the P&L, and the question being asked about it has changed. For years the question was what did we spend. Now it is what we got for it. Which investments are paying off, which are draining margin, which should scale and which should stop.
AI has made that question urgent. Spend is growing faster than any category before it, and the decisions that drive it no longer happen in quarterly cycles. They happen continuously, across hundreds of people and a growing number of agents, faster than any monthly review was built to follow.
This is the work of IT Finance: turning technology cost into business value. Yet most IT Finance practices reach for the quick win. Rightsizing recommendations, idle-resource cleanup, a savings number to report by the end of the quarter. Those tactics are useful, but they optimize the spend without ever answering whether it was worth it.
This is a multi-part series on IT Finance, and it starts where the discipline has to: with the cost model. Every framework reads from it, every stakeholder relies on it, and every decision, human or automated, is only as good as the numbers it provides. FinOps, TBM, EBM, the BI layer on top, all of them sit on the cost model and inherit whatever it gets wrong. When the model is right, the same numbers reach finance, engineering, product, and leadership, and mean the same thing to each of them. Increasingly, to the agents acting on cost alongside them too.
What a cost model has to be
A cost model that can answer what technology is worth has to be four things at once.
Trustworthy and aligned to the business. Nothing material is missing, so a product's true cost includes every source it runs on. Cost is expressed in products, teams, and customers rather than infrastructure terms no one outside IT can read. Shared platforms are split by real consumption. A disputed number stops being used.
Integrated. The model reaches the decision while it is still being made: the planning ticket, the pull request, the coding agent's context. Arriving after the invoice only explains the past.
Evolving and governed. Organizations restructure, products launch, new providers enter the estate every quarter. The model absorbs those changes without a re-architecture, and the controls built on it stay accurate as it does.
Accessible and transparent. Engineering, platform, procurement, finance, leadership: each one traces the same numbers to its own scope, and sees how they were reached, without filing a ticket.
Most cost models are one or two of these. Almost none are all four. This series is about building one, one layer at a time.
Why we're writing this
Our founder created the FOCUS standard; the team built the tooling at Apptio and the global IT Finance practices at Citi and NTT Data. We build an IT Finance platform at StitcherAI for teams aiming at a mature IT Finance practice, whether they are close to one today or a long way from it.
We are not neutral, and we will not pretend to be. Each piece takes a problem every organization with serious technology spend hits, lays out how we think it should be solved, and then shows what we built to solve it. The problem and the thinking come first, and they hold up whether or not you ever talk to us. Where we think we have the right answer, we say so plainly.
For years this capability was Fortune 100 only: eighteen to twenty-four months of custom build at seven figures a year, because nothing existed to buy. The problems are not exclusive to companies that size, and neither is the answer.
"The problems are real regardless of what you solve them with. Get the model wrong and no framework on top of it will save you."
Next: Attribution. Cost doesn't arrive labeled with an owner, so how do you get every dollar to the team, product, tenant, or feature responsible for it? New pieces every few days. Subscribe to be notified.