Every finance leader I talk to lately opens with the same worry. AI is coming for the team, and they need to decide how many seats to cut. A few have already sketched the new org chart in their heads.

I think they have the pattern backward. When the cost of the work goes down, the amount does not shrink. It grows. A 160-year-old idea called the Jevons paradox explains why, and AI in finance is about to prove it again.

When a professional task gets cheaper, demand for it rarely collapses. It explodes.

The question every finance leader is asking

No. AI is not going to shrink your finance team. I say that as someone who watches finance functions adopt these tools every quarter, not as someone selling you a forecast.

Think about what your board asks for now versus three years ago. Back then, a quarterly forecast and a variance bridge kept everyone happy. Now they want a weekly cash view, a driver-based reforecast, and three scenarios on a pricing change before Monday. The questions got harder while you were busy answering the old ones.

Most people miss why. They treat analysis as a fixed pile of work. Automate half, and half the team looks redundant. But the pile was never fixed. Every time my team got faster, a cleaner data model, a quicker close, a new reporting tool, we never got the hours back. We got new questions. The list of "can finance look at this?" is basically bottomless.

That is the real mechanic. Demand for finance is unusually elastic. The appetite for "help me decide" is close to unbounded. Cheaper analysis does not shrink the function. It expands it.

Tip: If you are modeling AI as a headcount cut, model the opposite first. Ask what your team would do with twice the capacity, then check which version your CEO actually wants to fund.

The Jevons paradox AI is about to prove again

In 1865, William Jevons noticed something strange. More efficient steam engines did not cut England's coal use. They multiplied it. Cheaper energy funded bigger ambitions: railways, factories, whole cities running on coal.

Economists call it the Jevons paradox. Make a resource cheaper, and the world finds more uses for it. Torsten Slok, chief economist at Apollo, recently gave it a modern name for knowledge work: the Jevons employment effect. When professional work gets cheaper, the market for it grows (Apollo, 2026).

Finance is about to live this. The reconciliation that took a day starts taking an hour. Good. Then your CEO wants the fundraising model by tomorrow, three scenarios instead of one. Your head of sales wants the pricing change stress-tested before the board call. The capacity you just freed gets spent fast, on questions you never had time to touch.

The pattern has already played out

This is not a prediction. We have watched it happen, and more than once.

In 2016, one of the founders of modern AI told researchers to stop training radiologists. AI would obviously replace them within a few years. Today, radiologists earn over $500,000, and there are more of them than when he made that call, with the field facing a shortage (Fortune, 2026). AI made reading a scan cheaper, so hospitals scanned more patients and caught more conditions early. The judgment part expanded and got more critical. The task got cheaper; demand for the expertise exploded.

Finance already ran this experiment. When spreadsheets arrived, everyone assumed accountants were finished. Instead, the profession grew. US accounting roles climbed from a few hundred thousand in 1980 to well over a million by 2022, though the job categories were redrawn along the way (Financial Times, 2024). The clerks counting by hand disappeared. The people who could interpret the numbers multiplied. Not despite the spreadsheet, but because of it.

Tip: When someone on your team says AI will do their job, ask them which part. Usually, it is the part they hate, and the part that was never where their value lived.

The honest catch: coal work and steam-engine work

Here is the piece the macro skips. Demand only expands where output is elastic, where the business wants more the moment you can produce more. Not all finance work qualifies.

Split your function into two piles.

The first is transactional work: reconciliations, data entry, and standard report production. Call it coal. The business needs a fixed amount and no more. Nobody wants twice as many month-end decks. Automate this, and it stays flat or shrinks. That is fine.

The second is judgment work: scenario modeling, forward-looking analysis, business-partner conversations, and decision support. Call these the steam-engine applications. The appetite here is close to infinite. Every manager wants another scenario, a sharper read on the pipeline, a faster answer. Make this cheaper, and demand booms.

The teams that get squeezed are the ones stuck in the coal pile. The teams that expand are the ones that push judgment work down to every manager who needs it.

Tip: If a task gets cheaper and nobody asks for more of it, that is coal. Automate it and move on. Do not defend it.

What to do Monday

So the reorg your instinct drafted is the wrong move. Here is the better one.

First, redeploy, do not cut. The capacity AI frees is not savings to bank. It is analyst time you can finally point at the questions you never got to. I see it with our customers on Abacum every quarter: the operational load drops, and the team gets pulled into more conversations, not fewer.

Next, raise the ambition of the analysis. If your FP&A (Financial Planning and Analysis) team produced three scenarios last quarter, ask for ten. If FP&A supported only the exec team, push decision support down to every department head.

Then, change what you measure. Stop counting reports produced. Start counting decisions supported. The first number should fall as AI takes over production. The second is the one that grows your influence.

One honest caveat. If your team was oversized to begin with, that is a different conversation. The Jevons paradox does not rescue bad math.

Tip: Put "decisions supported" on your board deck this quarter, before you have a clean way to measure it. The act of counting will change what your team chases.

The bar just went up

AI is not the layoff you are dreading. It is the promotion your function has been waiting for. The tasks are getting cheaper, the bar for what your team does is getting higher, and that is the best news finance has had in years.

Further Reading: The 10x Finance Leader: Actually Being a "Strategic CFO"