The previous piece asked a simple question with a complicated answer: are we creating enough value from AI to justify what we are investing? That question does not stay in engineering for long. Sooner or later it lands on the desk of the person who has to sign off on the spend, the CFO.
For finance leaders, AI is turning out to be an unusually awkward cost line. It behaves less like software and more like weather.
Why AI Breaks the Finance Playbook
A SaaS licence gives you a predictable cost per user per month. You can forecast it, cap it, and explain it to the board. AI does none of that. Cost scales with activity: prompts, context size, model choice, agent usage, and the complexity of the workflows being automated. Two engineers with identical licences can generate wildly different bills depending on how they work. The unit price of a token keeps falling, yet the total invoice keeps climbing, because consumption grows faster than prices drop. The result is a genuine paradox for finance: AI is getting cheaper and more expensive at the same time.
The Irish Picture: Adoption Racing Ahead of Proof
Deloitte’s 2026 research ranked Ireland first in EMEA for integrating AI into corporate strategy, with around one in four Irish firms now having a Chief AI Officer, and 81% integrating AI into their corporate vision.
EY Ireland’s 2026 CFO survey, which polled 200 finance leaders, found AI use inside the finance function had nearly quadrupled in a year, from 12% to 47%.
But the spend is running ahead of the returns. PwC Ireland’s 2026 research found only 17% of Irish CEOs had seen additional revenue from AI in the previous year, below the global figure, and most reported neither higher revenue nor lower costs. Widespread use of AI agents sat at roughly 9% of Irish organisations, against 52% in the US. Deloitte Ireland’s spring survey captured the mood well: Irish CFOs are pursuing AI, but through a deliberately governance-first, cost-disciplined lens. Ireland leads on strategy and governance intent but lags on converting AI into financial results, which is exactly the CFO’s problem.
The Art of the Conversation
Technology optimises for capability and uptime, while finance optimises for margin and predictability. When those two functions do not regularly communicate about cost and usage, the invoice becomes a monthly surprise rather than a managed line item. A conversation is needed to ensure finance and technology are not working off different playbooks. An agent left running in a reasoning loop, a handful of enthusiastic engineers driving most of a team’s consumption, a pilot that became production without anyone deciding it should. None of these is visible from an invoice alone, and none of them gets caught if the two functions only meet at budget time or after the reporting period closes.
The first CFO consideration is very basic: build a standing conversation between finance and technology about what AI is being used for and what it costs. Shared visibility, in something close to real time, is worth more than any after-the-fact reporting.
Asking for the Plan
The second consideration is strategic alignment. It is one thing for 81% of Irish firms to say AI is part of their corporate vision, as Deloitte found; it is another for finance and technology to agree on which problems AI is actually meant to solve, and in what order.
This is where a reasonable CFO should feel entitled to ask technology a direct question: is there an organisational plan for how AI is being implemented? Not a list of tools, but a plan; which workflows are in scope, what each is expected to return, how success will be measured, and how spend will scale as adoption spreads. If that plan does not exist, the honest position is that the organisation is experimenting, not investing. Both are legitimate. But finance should know which one it is funding.
Who Owns the Value?
The third, and least comfortable, consideration is ownership. Productivity gains from AI are real and widely reported in Ireland, but, as PwC’s findings show, they have largely not converted into financial metrics. Someone has to own that conversion, and in many organisations nobody clearly does. AI spend reports into technology, whose scorecard is capability, not return, so value measurement falls through the cracks. The CFO does not need to run AI, but finance does need a named seat where value is defined and tracked, whether that sits with a Chief AI Officer, a FinOps function, or the finance team itself.
Takeaway
Cheaper AI will not automatically produce cheaper operations, any more than cheaper cloud produced smaller cloud bills. As we have noted, a bigger AI budget does not always equal bigger AI gains. The winners here will be the organisations that can ensure finance and technology stay aligned on cost, usage, and strategy, and whether someone, clearly, is accountable for proving the value.
Connect with Conor on LinkedIn or at conor.murphy@barden.ie
