AI Displacement Has Crossed the Attribution Threshold — The 73% Forecast Looks Conservative
textak has held a 73% probability on the first major AI-attributed layoff wave for several months, moved from 72% on the weight of accumulating circumstantial evidence. Today's evidence from SkillSyncer changes the analytical footing: 185,894 workers displaced in H1 2026, with 56% of layoff events — 150 companies affecting roughly 156,270 workers — explicitly citing AI, automation, or machine learning as a driving factor. This isn't companies being nudged to mention AI; it's the majority of documented layoff events using it as a stated cause. The forecast asks whether the attribution threshold would be crossed publicly. It has been.
Our 73% was built on a specific diagnosis: the real variable was never whether AI was displacing workers — that was already happening — but whether companies would publicly attribute it. We weighted toward YES because investor pressure for AI ROI creates an odd incentive: you need to demonstrate AI is doing something, and headcount reduction is a legible, auditable metric. But we also held back from a higher number because we expected most attribution to remain in earnings call language and analyst presentations, not in the HR paperwork that gets counted in layoff trackers.
What the SkillSyncer data changes is the evidentiary category. Prior signals — AI mentioned in earnings calls, attrition not backfilled, junior hiring freezes — were proximate evidence: conditions consistent with our thesis but not proof of it. A layoff tracker showing 56% explicit attribution across 267 events is closer to direct evidence. Oracle's 30,000 cuts, Amazon's 16,000, Meta simultaneously cutting 8,000 while reallocating 7,000 into AI roles — these aren't companies hiding the ball. The combination of infrastructure spend commitment and headcount reduction, publicly disclosed, is the pattern we were forecasting.
The honest complication: 'explicitly cites AI' in a layoff event covers a wide range of disclosure. A press release saying 'we are restructuring to invest in AI capabilities' is meaningfully different from 'this role is being replaced by an automated system.' We don't have granular visibility into how SkillSyncer codes attribution, and the 56% figure may be capturing the softer end of that spectrum. If the resolution criterion requires something closer to direct displacement statements rather than restructuring-toward-AI language, the picture is less clean. That's the gap in our model we're still watching.
What would move us? If a Fortune 100 company publishes a workforce reduction plan that explicitly delineates roles eliminated versus roles automated, with AI named as the operational replacement, that resolves the forecast unambiguously toward YES. The H1 2026 data makes us think that moment is imminent — the attribution behavior has clearly shifted at the sector level. At 73%, we're already pricing in high probability. We're considering whether the H1 data warrants a move toward 80%, pending clarity on how 'explicit attribution' is being operationalized across the 150 companies in the dataset.