AI is getting the blame for 2026's layoffs, while total layoffs are falling

At the end of June, Oracle disclosed that it had shed tens of thousands of roles over the past year, and the savings are going into AI data centers. It is the latest in a long run of layoffs where the company names AI as the reason. The numbers behind those announcements say something more complicated than “the machines are taking the jobs”, and a business planning its own headcount needs the complicated version.

AI has become the default explanation for a layoff

Since March, artificial intelligence has been the most-cited reason for US job cuts in the outplacement data Challenger, Gray & Christmas collects, a streak with no precedent in its records. Through June, AI was named in 101,743 US layoff announcements, roughly a quarter of all cuts. That is close to double the figure for all of 2025. The tech sector announced 139,156 cuts in the first half of the year, up 83% on 2025.

Oracle’s June 22 filing put its cuts at around 21,000 roles over twelve months. Earlier in the year Block cut close to 4,000 people, nearly half its workforce, and Jack Dorsey told shareholders that AI tools had changed what it means to run a company. Meta, Amazon, PayPal and Salesforce have also cut staff while pointing, directly or loosely, at AI.

Total layoffs went down

The share of cuts blamed on AI has grown fast. The total has shrunk. Announced US cuts in the first half of 2026 were around 40% below the same period in 2025, and without last year’s one-off federal reductions, 2026 looks much like 2024. What changed is the label on the cuts.

Executives say as much when asked directly. In an NBER survey of roughly 6,000 executives, about 90% said AI had essentially no effect on their hiring over the past three years. Sam Altman, who sells the technology, has said firms announce AI-driven cuts “whether or not it really is” about AI. Wharton’s Peter Cappelli points out that companies usually say they expect AI to cover the work in future, which is a forecast. And a Forrester report early in the year found many firms announcing AI-related cuts had no mature, tested system ready to do the jobs they were eliminating.

My reading is that most of this is reallocation. The companies cutting staff are spending close to $700 billion on AI infrastructure this year, and many are posting record profits. Cutting payroll helps pay for the data centers, and “because of AI” plays better with investors than “restructuring”, which is what the same decision would have been called three years ago.

Denmark has its own version in the government platform

Denmark has not seen US-style layoff waves blamed on AI, but the idea is now government policy. The platform the new government published in June sets a target of freeing at least 30,000 full-time positions in the public sector by 2035 through AI, with a transition fund for reskilling.

That target should get the same scrutiny as the American announcements. Work-years are only freed if AI systems really take over the tasks, and so far the US record shows that promising the savings is much easier than delivering them. Danish companies promising their boards AI savings deserve the same question: which system, doing which work, measured how?

Where AI is already changing jobs

AI is having an effect on work. It is just narrower than the headlines, and it shows up first at the bottom of the ladder.

Stanford’s Digital Economy Lab, using payroll records covering millions of workers, found that employment for software developers aged 22 to 25 fell close to 20% from its late-2022 peak, while employment for developers over 26 kept growing. The most exposed roles are those where AI now does the task instead of helping with it: customer support, content moderation, data entry, QA and junior analysis. If your staffing plan depends on hiring juniors to do exactly that work, it is the plan to revisit first. Senior staff are safer than the coverage suggests.

For your own workforce planning, keep reallocation and replacement apart in the numbers. Cutting people on the assumption that AI will cover their work is a bet, and a Gartner study found the firms cutting hardest showed no improvement in financial returns. Put the system into production first, then decide on headcount.

Think about the pipeline as well. An industry that stops hiring juniors because AI handles entry-level tasks will have too few experienced people in five years.

And watch what you signal to the people who stay. “Job hugging”, where anxious employees cling to roles they have mentally left, is a documented drag on productivity. Announce AI cuts you cannot back up and you get the morale damage with none of the efficiency.

“We cut jobs because of AI” is carrying a lot in 2026 press releases, and much of it is not true yet. Before you plan your own cuts around it, find out which of your tasks AI can do reliably today.