Up to 9% of Workers Could Be Displaced by AI, but the Layoffs May Wait for a Downturn
A Goldman Sachs economist now puts the share of workers who could be displaced during the AI transition at up to 9%, roughly one in eleven. The note's sharper point is about timing: automation-related layoffs have historically stayed limited in healthy economies and accelerated during slowdowns.
The estimate, and its assumptions
Economist Joseph Briggs assumes AI eventually lifts productivity by 15% in the U.S. and other developed economies. On that assumption he finds that up to 9% of workers could be displaced. He has raised the number before: in August 2025 he and a colleague estimated 6% to 7%, within a range of 3% to 14%. By June 2026 he described the figure as about 9%, or roughly 15 million workers over a decade.
The figure counts gross job losses, not unemployment. Some displaced workers find work quickly and others leave the labor force, so the jobless rate hides part of the picture. Briggs also frames the displacement as temporary.
The finding about timing
Looking back across decades of data, the note concludes that companies tend to automate work and shed labor costs when the economy starts to slow. The clearest chart tracks the share of Americans in routine jobs such as clerks, secretaries and machine operators: it stays stable in expansions and falls in steps during recessions, without recovering to its earlier level.
A recession could pull AI layoffs forward
Briggs warns that a cyclical slowdown would likely accelerate AI-related automation and job losses, amplifying the downturn. That is a forecast built on historical patterns, not something observed in 2026 data.
What past technology did to specific jobs
Briggs studied 25 occupations hit hard by earlier technology, including bank tellers, file clerks, proofreaders and film projectionists. On average, employment in those jobs fell 25% in the five years after decline began. Most of the shrinkage came from people leaving or losing jobs, with separation rates up in 19 of the 25 occupations, rather than from employers hiring fewer people. That is why he treats AI-attributed layoffs as a useful real-time gauge.
Why the impact looks small so far
The note offers two reasons. Only about 23% of U.S. establishments have built AI into regular workflows, so most productivity gains are still ahead. And with the economy and labor market resilient in 2026, companies have had less reason to cut labor costs aggressively. Fortune adds that AI-attributed layoffs in Challenger's tracking have pulled back sharply from their spring peak (see our October report analysis).
Goldman also lists reasons the 9% could be too low: displacement ran higher during the late-1990s productivity boom, the period most often compared with today, and Goldman expects AI's productivity boost to beat most past ones.
Primary documentation
We read Fortune's report of the Goldman note; the note itself is not public. Figures and characterizations are Goldman's as reported by Fortune. This is a forecast under stated assumptions.