Goldman Sachs has put its number on AI displacement — and its timing — in the same breath. A research note published Thursday by Joseph Briggs, who leads the bank’s global economics team, finds that under Goldman’s forecast of a 15 percent AI-driven productivity lift, up to 9 percent of US workers could be “displaced during the AI transition”, with slower AI-related hiring adding another 3 percent of drag on employment — roughly one in eleven workers, or about 15 million people over a decade, as Fortune’s analysis of the note reports. The twist: Goldman does not expect the pink slips next quarter. “Automation-related labor disruption is limited when the overall economy is healthy but accelerates during cyclical slowdowns,” Briggs wrote — the layoffs everyone has been bracing for are, in his model, anchored to the next recession.
Briggs does not conjure the pattern; he digs it up. Across 25 occupations hit hard by past technology — photo processors, movie projectionists, proofreaders, file clerks, bank tellers — employment fell 25 percent on average in the five years after decline began. Separation rates rose in 19 of the 25 occupations while hiring rates fell in only 13, meaning people are pushed out faster than they fail to be hired. The share of Americans in routine jobs “has remained stable during expansions but declined sharply during downturns”, dropping in steps through the recessions of the early 1980s, 2001, 2008 and 2020, never fully climbing back — a chart that matches what Goldman’s own economists argued a year earlier when David Mericle and Pierfrancesco Mei called jobless growth “the likely new normal” and warned that “the full consequences of AI for the labor market might not become apparent until a recession hits”.
Why It Is Quiet Right Now
If 9 percent displacement is coming, why has the layoff gauge actually cooled? Briggs’s answer, spelled out in the note as summarised by Startup Fortune: the US economy “has remained resilient in 2026”, so companies have had less incentive to cut labour costs aggressively, and only 23 percent of US establishments have built AI into regular workflows — meaning most of the productivity gain, and most of the displacement, is still ahead. AI-attributed layoffs in Challenger tracking have pulled back sharply from their spring peak. Goldman still expects the aggregate unemployment hit to stay under one percentage point per year, partly because displaced workers get reallocated and “temporary” pain — but temporary and painless are different claims, and the history Briggs assembled suggests the pain will be concentrated, not spread.
The pattern lands on a site that has been arguing both halves of it all month. We covered the rehiring correction — companies quietly undoing AI-motivated cuts — which is exactly the attrition-not-firings behaviour Briggs models. We covered the low-hire, low-fire labour market where layoffs sit at 57-year lows while posting after posting quietly disappears, and Barclays’ chief executive calling AI the icing, not the cake in its own 5,000-job cut. The bank’s new note is the bridge between those stories: a healthy economy absorbs automation through the exit door rather than the layoff letter, and the bill arrives with the next downturn.
What It Means for Workers and for New Zealand
The uncomfortable arithmetic for an individual worker is that Goldman’s reassurance does not apply to any particular job. Across those 25 hard-hit occupations, employment fell 4 percent extra over five years for every point the national unemployment rate climbed, and Briggs warns “more scope for displacement during the AI transition” than history suggests, because AI’s productivity boost may be larger than past ones. Waiting for the recession to reveal whether you work in a routine-job occupation is a poor career strategy. The better read of Goldman’s data is the timing one: routine-heavy roles are safest from cuts while the economy runs hot precisely because mass restructuring is deferred, not because it is cancelled — our September analysis of Goldman’s hiring map found call centres and software already absorbing the slow-hire version of this story.
New Zealand sits a fraction further from the blast radius but not from the mechanism. Our labour market is smaller and more exposed to primary industries where AI displacement is slower; but the same cycle logic applies to any New Zealand firm whose 2026 headcount plans quietly assumed cheap growth without new hires — the “jobless growth” wedge Goldman describes would land here as it does in the US, just later and smaller. The practical note for readers is the same as it is for the bank’s clients: watch separations, not hiring. Goldman found firing, not hiring, is the leading edge of AI-driven job loss — which means the indicator of genuine trouble will not be a freeze on new roles but a sudden jump in departures at exactly the moment the economy starts contracting.
Sources: Fortune, Startup Fortune, Blockgeni, Goldman Sachs Exchanges, Goldman Sachs Insights
