Goldman Sachs Warns AI Is Cutting Jobs Beyond Silicon Valley as Investment's Growth Boost Nears Peak

Artificial intelligence is putting pressure on US jobs far beyond Silicon Valley, with hiring weakening in service industries that increasingly use the technology. At the same time, the spending boom that has fuelled AI infrastructure growth is approaching a turning point, Goldman Sachs warns.

The bank's research found that employment in the most AI-exposed non-manufacturing sectors was running about 10,000 jobs below its long-term trend over a three-month period. Industries making greater use of AI in routine business functions also showed weaker employment growth and higher unemployment.

Goldman Sachs chief economist Jan Hatzius expects the economic boost from AI investment to peak in late 2026 or early 2027.

Graduates And Younger Workers Face The Greatest Pressure

Younger, less-experienced workers are facing greater hiring pressure. Goldman Sachs found that the unemployment gap between entry-level and more experienced workers had widened by 0.6 percentage points, while the wage gap had grown by 1.3%, relative to pre-pandemic averages.

Some companies are holding back on recruitment rather than resorting to mass layoffs. This can involve leaving vacancies unfilled, reducing graduate intake and slowing new hiring. Some businesses outside the technology sector have also cited AI-driven efficiencies in restructuring plans. American Express, for instance, has announced plans to cut about 5% of its workforce.

A construction hiring boom for AI data centres and related infrastructure has helped offset weakness in office-based employment. That has kept the technology's overall impact on US payrolls relatively modest.

The cushion could weaken if data centre projects run into power shortages, financing difficulties or supply constraints. A slowdown in construction would leave fewer new infrastructure jobs to offset weaker hiring elsewhere.

Goldman Sachs separates AI's impact into two categories: substitution, where technology replaces human tasks, and augmentation, where it helps workers do their jobs. In research published earlier in 2026, the bank estimates that substitution reduces monthly payroll growth by roughly 25,000 jobs, while augmentation adds about 9,000. The net effect is a drag of around 16,000 jobs a month.

AI Spending Is Set To Lose Momentum

In June, market consensus forecasts suggested hyperscaler capital expenditure growth would slow from 84% in 2026 to 22% in 2027.

Goldman Sachs analysts challenged those estimates at the time, arguing that spending could reach about $1.1 trillion (£825 billion) in 2027, implying growth of approximately 45%. Goldman Sachs estimates that the companies need roughly $300 billion (£225 billion) in annual AI revenue to break even on their infrastructure investments.

Goldman Sachs chief economist Jan Hatzius estimates that the roughly 100-basis-point rise in long-term interest rates since early 2026 could subtract a cumulative 0.6% from US GDP if sustained.

Goldman Sachs estimates that gains in AI-related stocks boosted US consumer spending growth by about 0.4 percentage points over the past year, equivalent to roughly 0.25% of GDP.

Economist Says AI Investment Could Decide Who Loses Jobs

Daron Acemoglu, an economist at the Massachusetts Institute of Technology, argues that job losses are not an inevitable consequence of AI. The outcome depends on what companies choose to build and how they deploy it.

Acemoglu estimates that net job losses from AI will be 'less than, say, 2-4%' over the next five years, with routine white-collar roles particularly exposed. He argues that investment is too focused on replacing workers rather than helping them.

Shifting investment towards tools that complement human skills, he says, could turn productivity gains into stronger labour demand and higher wages.