Invest1 publisher3 min readPublished
Gartner traces four-fifths of first-half AI layoffs to firms funding their own AI build-out
Of the 21% of first-half 2025 layoffs Gartner could link to AI at all, 17 points were legacy units cut to pay for AI engineers, salespeople and data center staff. Cuts from AI productivity gains stayed near 1%.
The Investor · Invest desk
What happened
- Gartner examined nearly 1.4 million jobs and found 21% of layoffs in the first half of 2025 had any link to AI, a share that fell to 13% in the second half.
- Repositioning, meaning cuts to legacy business units made to free resources for AI, accounted for 17% of first-half layoffs and 7% of second-half ones.
- AI-related hiring freezes made up about 4% of workforce changes in the first half of 2025 and 0.3% in the second half.
Compiled by The InvestorSomething wrong?How this is made
Why it matters
- constraint Pressure to show cash savings against AI spending has to be met without headcount reduction, because Gartner's analyst says productivity gains will not be the force that enables the cuts.
- decision A cut announced as AI-driven is, on these numbers, more often a funding decision for a new product line, so the question for a board is whether the same quarter's AI hiring offsets the reduction.
- contradiction Amodei's one-to-five-year window from May 2025 runs past 2025, so a one-year layoff study cannot falsify the forecast it appears to answer.
Divide the repositioning share into the total and one category carries most of it. In the first half that is 17 of the 21 points, about four-fifths; in the second, 7 of 13, a little over half [1][2]. The rest moved the other way, from 4 points to 6 [3].
What Gartner calls repositioning is a cut to a legacy business unit made to free resources for AI. Nate Suda, the VP analyst who presented the work at Gartner's 2026 Finance Symposium/Xpo, said the pattern sits mostly in tech companies [4][6]. In many of those cases the same firm is hiring AI engineers, AI salespeople and data center personnel [6].
The category everyone was watching is the smallest one. Reductions in force coming directly from AI productivity gains were less than 1% of layoffs in the first half, and 1% for 2025 [10]. Suda said most of the second-half cases were at only two companies [11]. "What we're seeing is that the work is moving within the organization, not that the work is disappearing," he said [12]. He also said AI productivity would not be the material force that allows layoffs to happen [13].
The bases differ. The 1.4 million figure counts jobs studied, the 21% and 13% are shares of layoffs, and the hiring-freeze figures are shares of workforce changes. The three numbers do not sit on one denominator. Gartner did not publish the layoff count behind the percentages [4].
The freeze figure is the weakest part of the case. Suda described those freezes as tentative: "What that means is they're budgeting the headcount. Maybe they're not initiating that hiring, and they're checking in every month," he said, calling it a "live experiment" [9]. A role that is budgeted and never opened does not become a layoff, so a study of layoffs cannot count it. Of the freezes themselves, Suda said, "It looms larger in the mind than it does in the numbers" [8].
Amodei's prediction came in May 2025 with a one-to-five-year window, so a 2025 read covers its first year [1]. CFO Brew's own caveat is that a single year cannot prove what happens four years out [17]. Torsten Slok, chief economist at Apollo Global Management, wrote last week that he saw "zero evidence of job losses because of AI" in the weekly ADP employment data [14]. CFO Brew reported that Amodei, at a recent press briefing, hinted AI might transform the workforce instead of decimating it [15].
On the 2025 record, the defensible sentence for a tech company cutting a legacy unit is that it is moving budget into a new product line. In Gartner's numbers that is 17 points out of 21 in the first half [2][5]. Executives are under pressure to show savings against what they have spent on AI, and Suda's answer was that layoffs are not how to get there [16].
What to watch
- Whether the two companies behind most of the second-half productivity cuts become a larger group in 2026 reporting.
- Whether the budgeted-but-unopened roles Suda called a live experiment convert into cancelled requisitions.
- Whether the AI-linked residual, 4 points in the first half and 6 in the second, keeps growing as repositioning shrinks.