Goldman Sachs Raises AI Job Displacement Estimate to 9% of U.S. Workers, Says Layoffs Likely Wait for a Recession

Since Goldman Sachs put AI displacement at 6% to 7% of U.S. workers in August 2025, the bank has pushed its estimate up to as high as 9%. The revision came in an October 22 note from economist Joseph Briggs, who co-leads Goldman's global economics team. The number is still driving the debate over when AI job losses actually hit.

What the new forecast says

Goldman's baseline assumes AI lifts productivity 15% in the U.S. and other developed economies. Under that scenario, Briggs wrote, up to 9% of workers could be "displaced during the AI transition." That is about 15 million people.

Slower hiring could weigh on employment by another 3%, according to the note. Fortune summed the combined figure as roughly one in 11 workers.

Goldman says the U.S. is still early in adoption. Less than a quarter of firms have built AI into regular workflows, by the bank's estimate.

Why Goldman says the cuts are waiting

The timing finding is the part that matters most. Briggs wrote that "automation-related labor disruption is limited when the overall economy is healthy but accelerates during cyclical slowdowns."

His evidence is the share of Americans in routine jobs such as clerks, secretaries and machine operators. That share held steady through expansions and dropped in the recessions of the early 1980s, 2001, 2008 and 2020, and never fully recovered.

Chief economist Jan Hatzius framed the baseline as gradual. "AI labor disruption will likely only gradually increase rather than delivering an abrupt labor market shock," he wrote. He added a warning: "A cyclical slowdown would likely accelerate AI-related automation and job losses, thereby amplifying the broader business cycle."

Goldman has made this argument before. In October 2025, economists David Mericle and Pierfrancesco Mei wrote that the full consequences of AI for the labor market "might not become apparent until a recession hits."

What the data shows so far

The spring panic has cooled. Briggs wrote that "concerns of imminent AI-related labor displacement have cooled in recent months after hitting a fever pitch this spring." Layoff announcements ticked lower in September, but AI remains the top cited reason for job cuts in Challenger, Gray & Christmas data as reported by Business Insider, accounting for 21% of all announced layoffs.

Goldman's own numbers show a quieter squeeze. Employment in the industries most exposed to AI has run about 10,000 jobs below trend over three months, the note found. Hiring tied to AI-related construction, such as data centers, has largely covered that shortfall.

Entry-level workers face the sharpest headwinds, according to Goldman research covering more than 800 occupations.

Corporate announcements are following. American Express said on October 23 it will cut about 5% of its global workforce. CEO Stephen Squeri said the company is leaning on AI to become more agile and efficient.

A separate Goldman study led by Mei looked at workers laid off from jobs with high AI exposure starting in 2025. The bank found they searched for new work about a month longer than other laid-off workers. They were about 10 percentage points more likely to still be unemployed a year later. Workers 35 and older who were likely displaced by AI needed about five extra weeks to return to full employment. For those under 35, the gap was about three weeks.

The study covered more than 15,000 workers who lost jobs between 2016 and 2025. The Current Population Survey has no question asking whether AI caused a dismissal. Goldman built its own indicator from occupational AI exposure, industry adoption and whether a position was eliminated, then labeled the top 30% as likely displaced by AI. For layoffs before 2023, it found no significant difference between that group and other laid-off workers.

Displacement or slower hiring?

Not everyone reads the evidence as mass displacement. Researchers at Stanford's Digital Economy Lab say they do not see widespread displacement of workers by AI across the economy. They tie the employment gap among 22-to-25-year-olds in the most AI-vulnerable occupations mainly to reduced hiring.

Goldman's own framing leaves room for that. Hatzius pointed out that about 85% of job growth over the past 80 years has come from roles that did not exist at the time. Goldman characterizes the coming displacement as temporary, on the view that specialization and demand growth create new roles over time.

Hatzius also did not sound an alarm on the broader labor market. He said developed economies are holding up.

The construction cushion

The data-center hiring that Goldman says is masking weakness in AI-exposed industries is tied to the AI spending cycle. Hatzius wrote that the growth lift from AI investment is likely to peak in late 2026 or early 2027. Goldman expects hyperscaler capex growth to slow sharply in 2027 and 2028.

Goldman's note frames the cycle as the key variable: it says displacement tends to stay limited when the economy is healthy and accelerate during slowdowns, so the pace of AI-related job losses will likely track the overall economy.