Back to Blog

AI layoffs doubled this year, but there’s a lot more to the story

TL;DR

Are AI layoffs increasing in 2026?
Yes. Employers cited AI in 120,136 announced U.S. job cuts through September 2026, more than double the 54,836 for all of 2025, according to Challenger, Gray & Christmas.

Did Microsoft lay off 60,000 workers?
No. Microsoft offered voluntary buyouts to about 60,000 eligible employees in October. About 30% of eligible employees took Microsoft’s earlier voluntary retirement offer, and the company expects fewer to take this one.

If AI layoffs are up, why are total layoffs down?
Total announced cuts fell 39% from 2025, mostly because of last year’s federal government reductions. The broader market is stuck in low hiring and low firing: September’s hiring rate was 3.4%, and the layoff rate was 1.0%.

Who is AI hurting most right now?
Early-career workers. Entry-level roles fell from 29% to 10% of salaried postings in the most AI-exposed occupations between 2021 and 2026, and employment for workers ages 22 to 25 in AI-exposed jobs sits 19% below trend.

Is AI lowering salaries?
Not yet. Advertised pay in AI-exposed jobs is up about 46% since 2021, but economists surveyed by Indeed in September now expect more downward pressure on real wages for college-educated workers over the next year.

Why are workers struggling to keep up with AI?
The target keeps moving. Demand for basic AI skills fell 25% after a January 2026 peak, and only 28% of employees trained on AI use it consistently afterward.

What should employers do before cutting roles for AI?
Define the work first. Gartner predicts that by 2029, 30% of employees replaced by AI will be rehired, often at a higher cost.

Employers blamed AI for more job cuts this year than for any other reason. Challenger, Gray & Christmas counted 120,136 announced U.S. cuts tied to AI through September, up from 54,836 in all of 2025. Then, in late October, Microsoft made headlines with a number that sounded even bigger: 60,000.

That number was a buyout offer, and nobody knows yet how many Microsoft employees will accept. The gap between that headline and what actually happened points to the real story.

Layoffs are the visible part of how AI is changing work; most of the change runs through quieter channels like buyouts, backfills that never get approved, junior roles that stop getting posted, and internal moves that stall. Underneath it all is a freeze. Most employers haven’t decided what AI changes about their jobs, and most workers can’t tell what to learn next.

AI layoffs are real, and they peaked in the spring

The AI share of announced layoffs climbed fast. It was 7% in January, 25% in March, and 26% in April. In May, employers cited AI in 38,579 cuts, about 40% of the month’s total and the highest monthly figure since Challenger started tracking the reason in 2023. By September, the monthly AI count had dropped to 3,961.

Technology is the main industry citing AI. The sector announced 165,925 cuts through September, up 54% from a year earlier.

Total announced cuts still fell 39% from 2025, because last year’s federal reductions inflated the comparison. Excluding government, announcements were down 15%. Andy Challenger said in June that AI wasn’t yet the “jobpocalypse” some predicted, and that companies were restructuring as they repositioned for AI. One caution applies to all of these numbers: they reflect the reasons employers give, which may overstate or understate AI’s actual role.

Buyouts: the cuts that don’t look like cuts

Microsoft’s October offer covers roughly a quarter of its workforce. Employees at level 70 and below, which reaches the vice president level, are eligible in the U.S., U.K., Canada and India. The package is comparable to standard severance. Earlier this year, about 30% of employees eligible for Microsoft’s voluntary retirement program accepted it, and Microsoft expects a smaller share this time because the terms are less generous.

A Microsoft spokesperson said AI is not driving the buyouts. KUOW noted the company has spent several years cutting staff while investing billions in AI infrastructure. Both statements can be true at once.

Google has run the same play repeatedly. In February 2026, it offered buyouts to parts of its global business organization, aimed at employees who didn’t like the pace or direction the company set around AI. That followed buyouts in its search, ads, and engineering units in June 2025 and at YouTube during an AI-focused reorganization in October 2025.

Buyouts matter because they shrink a workforce without anyone deciding which jobs AI actually changes. Employees choose themselves, and the ones most likely to leave have options, usually the experienced people a company can least afford to lose. Buyouts also draw far less attention than layoffs. An Indeed career expert told MarketWatch that buyouts often come before layoffs: if too few people accept, involuntary cuts can follow.

The freeze behind the headlines

Employers are acting on AI faster than they’re planning for it. Only 4% of 2,002 technology leaders surveyed by Gartner had plans to redesign roles for people working alongside AI agents. A separate Gartner survey of 297 CHROs found only 19% of organizations have workforce planning that accounts for how AI is changing work.

The federal data shows what that looks like at scale. In September, job openings held at 7.5 million while hires and separations barely moved. The hiring rate was 3.4%, and the layoff rate was 1.0%. Few people are getting fired, and few are getting hired.

The freeze lands hardest on the bottom rung. In Indeed’s salary data, entry-level roles in the most AI-exposed occupations fell from 29% of postings in 2021 to 10% in 2026. Stanford researchers found employment for workers ages 22 to 25 in AI-exposed jobs sits 19% below where it would be had it kept pace with less-exposed peers, up from 13% when they first reported it in August 2025. The gap comes mostly from fewer young hires, and Yale’s Budget Lab still sees no clear AI effect in national occupation data, so the pressure is concentrated rather than economy-wide.

Workers already inside companies are stuck too. Workday found internal moves fell at 57% of employers and promotions stayed flat. Outside, applicants per filled job rose from 58 to 69, yet time to fill held near 60 days because candidates often lacked the right skills.

Workers are chasing a moving target

Most workers think they know what to learn. In Workday’s September survey, 79% said they know the skills they need, but only 66% said their employer helps them build those skills.

The harder problem is that the target moves. Demand for basic AI skills like prompting peaked in January 2026 and then fell 25%, while demand for building AI tools and automating workflows rose 51% between September 2025 and July 2026. Employers are also asking for less of what would help people adjust: mentions of training skills in job postings fell 13% over the same period.

Training isn’t closing the gap. In a September survey of 1,500 U.S. employees, TalentLMS found that only 28% consistently use AI after training, and 24% haven’t been trained at all. Among trained workers who don’t use AI consistently, the top reason was distrust of AI for high-stakes work, followed by time spent checking and reworking its output. Nearly half said they feel rushed to adopt new ways of working.

So people stay put. Workday puts voluntary turnover at roughly 16% a year, with about 70% of employees still at the same company two years later, even as fewer of them would recommend their employer.

Pay: access first, paychecks next

Pay hasn’t fallen. Indeed found advertised pay in the most AI-exposed occupations up about 46% since 2021, compared with 25% in the least-exposed jobs. Most of that premium sits at the senior level, where the gap is about 17 points, against about 2 points at entry level. Stanford’s payroll data agrees that so far, AI is changing who gets hired more than what they’re paid.

Inside companies, compensation is improvised. In Payscale’s August survey of 500 employers, 49% said their salary structures hadn’t kept pace with AI. Fifty-eight percent pay a premium for AI skills or plan to within a year, while 23% have already stopped paying extra and now treat AI fluency as a baseline requirement.

Economists expect the paycheck effect to arrive next, and they’re specific about who feels it. Indeed’s September panel of 123 economists and labor experts shifted toward expecting more downward pressure on real wages for college-educated workers over the next 12 months, the largest quarter-over-quarter change in the survey. The panel saw no meaningful AI wage pressure for workers without degrees.

What the next 12 months look like

The same panel expects the freeze to hold. On average, the panel projects unemployment drifting from about 4.15% this fall to 4.32% by September 2027, with job postings down about 0.89% over that year. Workday’s leaders point the same way: 40% expect AI to help them get more out of current employees, while 28% expect it to cut headcount.

A few things will show whether the quiet channels are widening:

  • December buyout results. How many Microsoft employees accept, and whether involuntary cuts follow where too few do.
  • The monthly AI layoff count. Whether it climbs back toward the May peak or stays near September’s level.
  • Entry-level hiring. Whether junior postings in AI-exposed fields keep shrinking as a share of openings.
  • Degree-holder wages. Whether real pay for college-educated workers starts slipping, as economists now expect.

Gartner warns employers who cut first and plan later. It predicts that by 2029, 30% of employees replaced by AI will be rehired, often at a higher cost.

Breaking the freeze one role at a time

The fix starts with a single position. Before a role is cut, frozen, offered a buyout, or reposted, answer four questions in writing:

  • What does the role deliver? Define the output, then test where AI helps. Count review time and rework alongside speed.
  • Who checks the AI’s work? Put that responsibility in the job description.
  • Where will junior people learn? If AI now handles the task that used to teach a new hire the basics, decide what replaces it.
  • What should it pay? Re-benchmark before posting. A band set in 2024 may underprice scarce skills or overpay for ones that are now baseline.

Gartner found organizations with effective workforce planning are 1.4 times more likely to exceed revenue goals and 1.5 times more likely to exceed cost goals.

Hiring when shifting to being an AI-native team

STEM Search Group works with employers building AI into how they operate, coast to coast. Our searches span industries from automotive, robotics, aerospace and energy to medical devices, biotech, healthcare and deep tech startups, and roles from engineering, AI and data to operations, supply chain, clinical, product, sales and executive leadership. Companies navigating that shift well tend to treat each hire as a design decision: they define what the role delivers with AI in the picture, then look for people who can grow with it. That changes what we screen for. Alongside the technical experience a role requires, we look for evidence that a candidate learns quickly, wants to work with AI, and has adapted when a job changed under them. If you’re building an AI-native team, or deciding what that means for your next hire, we’d like to share what we’re seeing in the market.

Sources

Recruiting redefined; built for high-tech,
high-growth teams

Hire Talent
Job Board