Does adopting AI mean fewer people, or different work for the same people?

The question behind the question

When a leadership team asks whether AI reduces headcount, it is usually asking one of two different things: can we grow without hiring, or will the board expect us to cut. The evidence answers the first more clearly than the second.

What the exposure data shows

PwC’s 2026 barometer, built from more than a billion job advertisements across 27 countries plus company financials, compared companies by how exposed their work is to AI. From 2018 to 2025, the most exposed companies grew labor productivity 34% against 24% for the least exposed. They also grew headcount faster: 52% against 36%.

That is not what “AI replaces workers” predicts. It is what “AI makes the company grow, and growth needs people” predicts. The top fifth of AI-exposed companies recorded 163% productivity growth over the same period, nearly five times the group average, which says the spread between adopting and adopting well is far larger than the spread between adopting and not.

The same report puts the wage premium for workers with AI skills at 62% in 2026, up from 57% a year earlier, and finds jobs requiring AI skills growing 69% against 9% for the job market overall. The demand is for people who can work with the tools, not for fewer people.

What happens inside a team

The field evidence is smaller but more precise. A study of about 5,000 customer-support agents at a Fortune 500 software company, published through NBER and the Quarterly Journal of Economics, tracked a generative AI assistant rolled out in stages. Productivity rose 13.8% on average and 35% for the least experienced agents. Attrition among agents with access to the tool was 8.6% lower than among comparable agents without it, and the difference came from newer workers staying.

The mechanism the authors describe is worth repeating: the assistant spread the habits of the best agents to the newest ones. The tool did not replace the experienced people; it shortened the time it took to become one.

Both things are true

Gallup’s February 2026 survey of 23,717 U.S. employees shows the other side. In organizations that have adopted AI, 34% of employees report hiring expansion against 28% elsewhere, and 23% report workforce reductions against 16% elsewhere. Adopting organizations do more of both. Employees there are also more likely to think their own job could be eliminated within five years: 23% against 18% overall.

So the honest summary is that AI adoption makes organizations move, in both directions, and the direction is a choice the organization makes rather than a property of the technology.

For MSPs specifically

Kaseya’s 2026 survey of more than 1,000 managed service providers found the share reporting difficulty hiring skilled technicians nearly doubled in a year, from 9% to 16%. For most MSPs the question is not whether AI will let them cut technicians; it is whether AI will let them serve more clients with the technicians they cannot replace.

What to do with this

Decide what the new work is before the tools arrive. The organizations in these studies that kept and grew their people had an answer to “what does this person do with the four hours a week the tool gives back.” The ones that did not had a reduction in force eighteen months later and called it an AI strategy.

That decision is the first thing an assessment produces, and it is the reason we insist on one before any implementation.