An updated Goldman forecast raises the displacement estimate to 9% of the US workforce, with some sectors already feeling the pressure.
Goldman Sachs has a new number for the AI labor market debate, and it is not a comfortable one. The bank’s research team projects that artificial intelligence could displace up to 9% of the US workforce over the next decade, a figure that translates to roughly 15 million jobs.
The updated forecast, published on October 22, 2026, and led by economist Joseph Briggs, marks a meaningful revision from the bank’s prior estimate of 6% to 7% displacement.
How Goldman arrived at 15 million
The methodology behind the estimate draws on historical data linking productivity growth to job destruction. According to the research, each 1 percentage point increase in productivity from technological change tends to raise job destruction rates by 0.5 to 0.6 percentage points over a two-year window.
The report also flags that the disruption is not purely theoretical at this point. Sectors vulnerable to AI integration, including technology, management consulting, and graphic design, are already shedding 10,000 to 15,000 jobs per month. That is happening while AI integration across US establishments sits at roughly 23%.
Goldman also estimates a potential 0.6 percentage point increase in the US unemployment rate as a direct consequence of AI-driven displacement. The bank expects this effect to be gradual under normal conditions but more pronounced during economic downturns, when firms historically accelerate automation to cut costs.
The historical parallel Goldman keeps returning to
Goldman’s chief economist Jan Hatzius offered some context that softens the headline number without dismissing it. Approximately 85% of job growth over the past 80 years has come from roles that did not exist at the time, according to Hatzius, meaning entirely new categories of work created by prior waves of technological change.
The report draws a direct comparison to labor market shifts during the late 1990s and early 2000s, when internet-driven automation caused significant workforce disruption before eventually generating new employment categories at scale.
Goldman characterizes the coming displacement as temporary for the same reason: technology-driven specialization and demand growth tend to generate new roles over time.
What this means for workers, investors, and policymakers
For workers in the most exposed sectors, the math is already visible in monthly payroll data. The 10,000 to 15,000 monthly job losses Goldman cites are concentrated in areas where AI can replicate cognitive tasks at scale: drafting, analysis, code review, design iteration.
The fact that this is happening at 23% AI integration suggests the current pace is something closer to a preview than a finale.
The unemployment rate implication is worth watching closely. A 0.6 percentage point rise might sound modest in isolation, but unemployment is one of the Federal Reserve’s two mandated targets. If AI-driven displacement pushes the rate meaningfully higher, particularly during an economic slowdown when the effect is expected to accelerate, it creates pressure on both monetary policy and fiscal stimulus discussions.
The 23% AI integration figure is arguably the most important number in the entire report. At less than a quarter of US establishments using AI in meaningful ways, the displacement curve is still in its early stages. How quickly that number rises, and whether it rises during periods of economic strength or weakness, will largely determine whether Goldman’s 0.6 percentage point unemployment estimate turns out to be the floor or the ceiling.
