AI's impact on jobs is real, but it is also nuanced and largely anticipatory. U.S. announced job cuts totalled around 1.2 million in 2025, according to Challenger, Gray & Christmas, with roughly 55,000 explicitly attributed to AI, around 4.5%. In January 2026, AI accounted for around 7% of roughly 108,000 cuts, or about 7,600 roles.
A reasonable way to categorise AI-linked losses is into three buckets. The shares below are rough estimates, based on surveys and market reports, rather than precise accounting.
1. Direct replacement
Direct replacement may account for roughly 10-20% of AI-linked losses. This is where AI automates enough of a role's workload that the role becomes uneconomic.
That is still relatively nascent. A December 2025 HBR survey of 1,006 executives, discussed by Thomas Davenport and Nitin Mittal Srinivasan, found that only around 2% of organisations reported cuts based on actual AI implementation results.
2. Anticipatory reduction and freezing
The larger category, perhaps 60-75%, is anticipatory reduction or hiring restraint. Firms are acting on expected gains before those gains are fully realised.
The HBR survey found that around 60% of executives had reduced headcount or slowed hiring in anticipation of AI-driven productivity improvements, while 29% reported hiring fewer people than usual.
3. Resource reallocation
The third category, perhaps 15-25%, is resource reallocation. Funds shift toward AI initiatives, infrastructure, tooling, and specialist talent, while operating expenditure is trimmed elsewhere.
Morgan Stanley reported in February 2026 that companies using AI were seeing average productivity gains of around 11.5%, alongside net headcount declines of around 4%.
The shape of the risk
Survey data suggests most AI-linked reductions are not yet the clean story of a model doing a job and a person being removed. More often, they are strategic responses to expected efficiency: fewer entry-level hires, paused expansion, rebalanced budgets, and pressure on roles that were previously justified by process volume.
The impacts are likely to concentrate in entry-level white-collar work, where tasks are structured enough to automate or compress, but not senior enough to rely heavily on judgement, accountability, or complex relationship management.
So yes, AI may be coming for your job. But in the near term, it may arrive less like a direct replacement and more like a changed spreadsheet: fewer openings, tighter budgets, altered team shapes, and a shifting definition of what work is worth hiring for.