Cutting Heads Is the Cheapest Thing You Can Do With AI.

Ninety-six percent of companies investing in AI report productivity gains. Seventeen percent of them cut headcount.

Read that again. Almost everybody is getting the gain. Fewer than one in five is converting it into a layoff.

That is from EY's AI Pulse survey of 500 US decision makers at senior vice president level and above, across ten industries. The story you are reading every week says AI is a jobs event. The people actually running the programs are mostly doing something else with the capacity.

I think the cutters are wrong. Not morally wrong, though we can have that conversation too. Strategically wrong.

Here is their case, made fairly. You bought AI, a team got faster, and salary is your largest controllable expense. Cutting converts the gain into margin this quarter, and margin this quarter is what your board asked about.

That math is clean. It is also the only move in the playbook that can be run by someone who does not understand the technology, which is exactly why it is so popular.

And it works once. You cannot cut the same head twice.

The Eighty-Three Percent Did Something Else

The companies that gained capacity and kept their people spent it on purpose. From the same EY data: 47 percent expanded existing AI capabilities, 42 percent built new ones, 41 percent hardened cybersecurity, 39 percent funded research and development, 38 percent upskilled their own workforce, 29 percent cut prices to take market share, and 25 percent acquired companies.

Look at that list as a capital allocation decision, because that is what it is.

Freed hours are a funding source. They can buy you a new product line, a security posture that stops losing you enterprise deals, a price cut that takes a competitor's customers, or a severance line item. All four are real choices. Only one of them ends.

The Value Is Going Somewhere, and It Is Not to the Cutters

PwC surveyed 1,217 senior executives across 25 sectors and found 74 percent of AI's economic value is being captured by 20 percent of organizations.

That is a brutal concentration. Four out of five companies are buying the same tools and watching somebody else book the return.

PwC's global chief AI officer, Joe Atkinson, named the separator plainly. The leaders "point AI at growth, not just cost reduction." Those companies are 2.6 times more likely to say AI is changing their business model.

S&P Global backs the same read from the investment side. When companies report what AI spending is actually for, 64 percent name process efficiency and 59 percent employee productivity. Only 24 percent name headcount reduction.

So the stated strategy across the market is capacity. The headlines are cuts. Somebody is lying, and I do not think it is the survey.

What the Cutters Are Trading Away

Forrester's 2026 predictions say it plainly. The firm expects "half of AI-attributed layoffs to be quietly reversed," with the work returning offshore or at lower wages. Not because anyone had a change of heart. Because the job turned out to need a person.

Rehiring a role you cut eighteen months ago costs market rate and buys back none of the institutional knowledge that left with it.

That is the visible cost. The invisible one is worse.

AI value comes from implementation, and implementation is done by people who know how your business actually works. The person who understands why the billing exception exists, which customers will call if the workflow changes, and where the data is dirty. That person is not on the AI team. That person is on the list.

The Conference Board found only 33 percent of workers received any employer AI training in the past six months, and 28 percent get none at all, while 55 percent are using AI weekly or daily anyway. Companies are cutting the people who could implement this and declining to train the ones who stay.

Then they wonder why the pilot never made it to production.

I Am Swimming Against This Current on Purpose

Forrester forecasts AI and automation will take about 6 percent of US jobs by 2030, roughly 10.4 million roles, while augmenting another 20 percent. S&P Global measured a net negative employment effect of five points globally over the past year. The cutting is real, and I am not going to pretend otherwise.

I am saying it is a losing trade, and I will keep saying it while the current runs the other way.

A company that uses AI to do the same work with fewer people has bought itself one good quarter and a smaller company. A company that uses AI to do more work with the same people has bought itself capacity it did not have to hire for. One of those compounds.

How I'd Spend the Hours

Book the capacity like a budget. If a workflow got 30 percent faster, that is not a vibe, it is hours. Count them. Assign them a destination before somebody else assigns them to the expense line.

Fund the backlog you gave up on. Every organization has a list of things worth doing that never got staffed. That list is your first customer for reclaimed hours. Prioritize and execute.

Put people on the AI work. Your best operators know where the value is hiding. Move them onto implementation instead of out the door, and pay for the training. Thirty-three percent trained is not a workforce, it is a rounding error.

Take share instead of banking margin. Twenty-nine percent of the reinvestors cut prices. If your cost to serve dropped and your competitor's did not, that is a weapon. Use it while it is still an advantage.

Name the owner. Reclaimed capacity with no owner evaporates into meetings. Empowerment means somebody has authority to spend it.

The Part That Matters

Cutting is the easiest thing you can do with an AI gain. It requires no strategy, no implementation skill, and no imagination.

Every competitor you have can do it too.

Growth is harder, and that is the entire reason it is worth something.

You can cut your way to a quarter. You cannot cut your way to a company.