What Firms Expect to Cut Next
McKinsey's 2025 “State of AI” survey asked executives what's already happened in their workforces and what they expect over the next year. Every single business function is net-negative — every one expects more firms to cut headcount than to add it. Service operations leads the list, with 39% of firms planning cuts. And expected cuts outpace cuts that have already landed, in every function surveyed.
Key finding: One-third of all surveyed organizations (32%) anticipate an AI-driven headcount decrease over the next year, rising to 35% at firms with ≥$1B in revenue. The cuts aren't concentrated — they span customer-facing, back-office, and engineering functions. This is the plan, stated by the people doing the hiring.
Actual vs. expected headcount decrease, by function
% of firms reporting decreaseIn every function, expected cuts outpace actual cuts — executives are telegraphing a second wave. The biggest acceleration: supply chain (21% → 33%), manufacturing (12% → 28%), and marketing & sales (18% → 32%).
Why this matters for AI job risk
Our Age Gap insight showed what has already happened to early-career workers in AI-exposed occupations. This insight shows what executives plan to do next. Together they form a pincer: the youngest cohorts are already contracting in exposed roles, and the firms that employ those cohorts are telling McKinsey they intend to cut more over the next year across nearly every function.
The functions most at risk line up almost exactly with the occupations our Capability Coverage Index rates as highest-exposure: service operations (contact centers, CX), supply chain (logistics coordination), software engineering (junior dev work), and marketing / sales (content ops). If you work in one of these functions, your employer is currently more likely to forecast a cut than to forecast growth — regardless of which specific occupation you hold.
The one honest caveat: self-reported survey intentions ≠ actual cuts. The Occupational Churn insight shows that aggregate occupational-mix shifts are still within historical norms. Executives say they plan to cut more than they already have — but the same executives also said similar things in 2023 and 2024. Stated intent is a leading indicator, not a done deal. Watch this space.
Sources
- McKinsey & Company's 2025 State of AI survey, asking executives about actual and anticipated AI-driven workforce changes across 11 business functions.
- Actual decrease = % of respondents reporting a headcount decrease in a given function over the past year attributable to AI. Expected decrease = % of respondents anticipating a decrease over the next year. Net expected change = expected-increase minus expected-decrease; a negative value means more firms expect cuts than expect adds.
- Survey intent is a self-report from executives, not a prediction of actual outcomes. Both the +200% productivity optimists (see Productivity Paradox) and the 50-point public-pessimism group (see The 50-Point Gap) are reading the same underlying uncertainty. What McKinsey's data adds is the direction of executive intent.