Companies Cut 113,000 Jobs for AI in 2026 — Gartner Says the ROI Never Showed Up

Let me show you two numbers from May 2026, and then let me show you why they should worry you if you're making decisions about AI at your company.

Here's the first one: more than 113,000 tech workers lost their jobs across 179 companies in the first five months of 2026 — roughly 825 people every single day since January 1. That's worse than the brutal correction of 2023. And about a quarter of those cuts were blamed directly on AI.

Here's the second one: Gartner asked 350 global executives what actually happened after they deployed autonomous AI. Around 80% had cut staff. But when Gartner looked at who got real returns, the layoffs and the ROI had nothing to do with each other. The companies that cut deeply did no better than the ones that cut lightly. The layoffs didn't create returns. They just created layoffs.

That's the AI ROI paradox of 2026, and if you run a team, a department, or a company, it's the most important thing you can understand right now. The cuts are real. The spending is staggering. And the returns simply aren't where everyone assumed they'd be.

The Gartner finding that should stop you cold

Gartner released this study on May 5, 2026, after surveying 350 business executives in late 2025 about how their AI deployments were really going. The conclusion was blunt enough that Fortune, CIO, and the Financial Times all picked it up within days.

Here's the detail you need to sit with. The companies reporting great ROI and the companies reporting weak or negative ROI were cutting jobs at almost exactly the same rate. Think about what that means. If firing people to "let AI do the work" actually paid off, you'd expect the winners to be cutting hardest. They weren't. The knife came out at the same rate whether the AI was paying for itself or not.

Helen Poitevin, a Distinguished VP Analyst at Gartner, said it plainly: "Many CEOs turn to layoffs to demonstrate quick AI returns; however, this disposition is misplaced. Workforce reductions may create budget room, but they do not create return."

Read that one more time, because it flips the whole story you've been hearing all year. "AI lets us do more with fewer people, so we cut people and bank the savings" — that line sounds obvious, and it falls apart the moment you check the data. Cutting people gives you a one-time drop in your payroll line. It does not give you the compounding productivity gains that actually grow a business.

The numbers that don't add up

That $725 billion line is the one that makes the whole thing surreal. Meta, Amazon, Microsoft, and Alphabet together cranked their AI infrastructure spending up 75% in a single year — building the data centers, buying the chips, and locking down the power infrastructure behind deals like the $67B NextEra-Dominion merger. That is real money, deployed at a scale we've never seen, on the supply side.

Now look at the other side. The companies actually buying and deploying all that AI are cutting staff and not seeing the payoff. Money is pouring into AI infrastructure faster than productivity is coming out the other end. If you feel that tension in your own organization, you're not imagining it — it's the defining business story of the year.

So why do CEOs keep doing it?

Fair question. If the data says layoffs don't create AI returns, why are 80% of companies cutting anyway? Three reasons, roughly in the order leaders are willing to say them out loud.

1. The pressure to look decisive. When every competitor is announcing "AI-driven efficiency," the executive who keeps their team intact looks like they're asleep at the wheel. A layoff is the loud, visible signal to the market that says "we're serious about AI." And in the short term, the stock often rewards that signal — whether or not the productivity gain underneath it is real.

1. The budget-room trap, and this is the one that probably hits closest to home. If you cut 10% of a department, you get clean, measurable savings next quarter. The AI productivity gain that was supposed to justify the cut is fuzzy, hard to measure, and might take years to show up. Faced with a clean number now and a fuzzy number later, most people book the clean number. Gartner nailed it: layoffs "create budget room" but not "return."

1. The genuine automation cases — which are narrower than the headlines make them sound. Some roles really are being automated. Customer support and content moderation are the clearest examples — Freshworks and others have openly said AI chatbots now handle work their human agents used to do. I walked through that in how companies cut support costs 40-60% with AI customer service, and the ROI there is real and measurable. But it's a thin slice of those 113,000 cuts. Most of the broad layoffs are being justified by AI capability the company hasn't even deployed yet.

The finding that should change how you roll out AI

If you take one practical thing from the Gartner study, don't make it the scary headline. Make it this: the companies that actually got returns did something specific, and you can copy it.

The winners didn't try to eliminate the need for people. They amplified the people they had. They invested — hard — in new skills, new roles, and new ways of working that let their teams steer and scale the AI instead of competing with it. They hired AI orchestration specialists. They retrained the staff they already trusted to run fleets of agents. They rebuilt workflows around human-plus-AI, not human-replaced-by-AI.

You can see why this works the moment you look at where the technology actually is. When Claude Opus 4.8 ships with the ability to run hundreds of parallel subagents in a single session, the question stops being "can the AI do the work?" It can. The real question becomes "does anyone on your team know how to direct it, check it, and turn that output into something the business can actually use?" The companies that fired exactly those people are the ones posting the worst numbers. They cut the humans who turn AI capability into AI returns.

It's counterintuitive, but it holds up: AI returns come from amplification, and amplification needs people. Cut the people, and you kill the amplification, and you lose the return. If you remember nothing else, remember that.

Where AI really is taking jobs

I don't want to leave you with the impression that AI job loss is a myth. It isn't. It's just far more concentrated than the scary aggregate numbers suggest. Here's where the genuine, ROI-positive automation is happening in 2026:

• Tier-1 customer support. AI chatbots clear 40-60% of standard tickets without a human ever touching them. This is the most mature, most reliable win.
• Content moderation. High-volume, rules-based sorting that AI handles for pennies on the dollar.
• Entry-level and generalist IT roles. Hiring here is slowing fastest, and the generalist positions are taking the hardest hit.
• Routine document work. Invoices, data entry, first-pass contract review.

I laid out the full map of which roles AI replaces, which it transforms, and which it leaves alone, and the honest answer is that AI is reshaping far more jobs than it's erasing. The social fallout for the middle class is real, and it deserves to be taken seriously. But here's the uncomfortable financial truth: most companies cutting broadly aren't the ones automating well. They're using AI as cover for cost-cutting they already wanted to do.

The opportunity hiding in the paradox

Now for the part that's actually good news, especially if you're thinking about your own career. While companies slash generalist roles, the highest-paying AI jobs are seeing demand and salaries climb. AI orchestration specialists, agent architects, ML platform engineers — the people who turn raw AI capability into deployed business value — are the scarcest and best-paid workers in the economy right now.

The job market is splitting in two, fast. Generalist work that AI can partly automate is getting squeezed. Specialist work that amplifies AI is getting bid up. And the Gartner data tells you which side of that line pays: the companies investing in the second category get the returns, while the ones cutting the first category and stopping there get the budget room and nothing else.

So here's the move, whichever seat you're in. If you're building a career, aim to be the person who orchestrates the AI — not the person whose single task the AI can chip away at. And if you're leading a business, understand that the layoff is the easy lever that doesn't create return, while the retraining-and-orchestration investment is the harder lever that actually does.

The $725 billion question you should be asking

Let's zoom all the way out, because there's a bigger version of this paradox. If the Big Four are spending $725 billion on AI infrastructure, and the companies buying that AI aren't generating ROI from it — where does the return actually come from? Your answer to that question decides whether you think the trillion-dollar AI race is a bubble or a buildout.

The optimistic case: the ROI is coming, just slower than the layoff-happy CEOs assumed. The infrastructure gets built through 2026-2027, the orchestration skills mature through 2027-2028, and the productivity gains start compounding in 2028. Gartner itself expects autonomous business to become a net job creator by 2028-2029, fueled by new kinds of work AI can't absorb. On this reading, today's ROI gap is just the normal lag between spending the money and seeing the payoff — the same lag that's shown up in every big technology shift.

The pessimistic case: a real chunk of that $725 billion is being spent on capability companies can't convert into returns, because they fired the human orchestration layer that would have done the converting. On this reading, the gap isn't a timing problem — it's structural, and a painful correction arrives when the market notices that AI capex isn't turning into AI profit.

The truth is probably somewhere in the middle, and it's specific to each company — maybe yours. The firms that invested in amplification are already getting returns. The firms that just cut headcount have budget room and a tool they can't really use. The aggregate looks like "no correlation" precisely because both groups are blended together in the numbers.

What the smart operators are doing — and what you can copy

Here's the playbook the Gartner data quietly hands you:

1. Stop using AI as cover for layoffs you already wanted. The market sees through it eventually, and the data says it doesn't create return anyway. If you need to cut costs, cut them honestly — don't dress it up as an AI strategy.
2. Invest in the orchestration layer. Hire or retrain the people who turn AI capability into deployed value. That's where the ROI actually lives, and it's the scarcest skill on the market.
3. Start with narrow, measurable wins. Customer support, content moderation, document processing — the categories with proven ROI. Bank those before you try to reinvent the whole org.
4. Measure ROI honestly. Budget room from cuts is not AI ROI. Productivity per person you kept is. Track the second number, and don't let anyone confuse it with the first.

The honest take

The defining business mistake of 2026 is treating AI as a tool for shrinking your headcount when the data keeps showing it's a tool for amplifying your people. The companies cutting 113,000 jobs to "prove AI returns" are mostly proving they don't understand where AI returns come from. Gartner just put hard numbers on what the careful operators already felt in their gut: the layoff buys you budget room, the amplification buys you the return, and mixing the two up destroys value while making you look bold in a press release.

The money is flowing into AI faster than ever — $725 billion this year alone. The returns are real, but they're landing in the companies that kept and amplified their people, not the ones that cut them. By 2028-2029, the productivity finally catches up to the capital and the job market tips net-positive again. Between now and then, the people and companies that bet on human-plus-AI win — and the ones that just swung the axe find out the hard way that budget room and return were never the same thing.

So before you sign off on the next round of cuts in the name of AI, ask yourself the one question this whole story comes down to: are you removing the people who would have made the AI pay off?