Goldman Sachs: AI Is Eliminating 11,000 US Jobs a Month — and Gen Z Is Paying the Price

Goldman Sachs released its latest AI Adoption Tracker this week, and the headline figure lands with more nuance than it might first appear: artificial intelligence is now responsible for the net displacement of approximately 11,000 US jobs per month, down from the firm’s April estimate of 16,000. On the surface, this looks like progress. Look closer, and the story is considerably more complicated.

The Headline Number

Goldman’s AI Adoption Tracker, maintained by chief economist Jan Hatzius and his team, uses a methodology the firm calls “AI displacement attribution.” Every occupation in the Bureau of Labor Statistics Standard Occupational Classification system receives a score from 0 to 100 based on the percentage of core tasks that current AI systems can perform at or above human competency levels. The score is then cross-referenced with employment data and observable automation investment patterns to generate monthly displacement estimates.

The April 2026 figure of 16,000 net monthly job losses has moderated to 11,000 in the June tracker. Goldman attributes the decline primarily to data center construction employment, which has added approximately 212,000 jobs since 2022 and is currently generating roughly 9,000 new positions per month.

The math looks nearly neutral at the aggregate level: 11,000 displaced, 9,000 created — a net loss of 2,000 jobs per month, easily lost in statistical noise. But the aggregate is deeply misleading, for reasons that are becoming the defining anxiety of the current AI boom.

The Temporary Job Problem

The jobs being created and the jobs being destroyed are not the same jobs.

Data center construction is, by definition, temporary. Once a facility is built and operational, it does not require the same density of labor to maintain it. The American Edge Project, a technology advocacy organization, warns that the industry currently projects approximately 4.7 million temporary construction jobs associated with the current data center build-out cycle. But once those facilities are complete, they’re expected to generate only about 697,000 permanent operations roles — just under 15 cents on the dollar.

In other words, the data center construction boom is providing a short-term labor market cushion that will itself disappear as the build-out matures. Goldman’s June report does not assign a timeline to this transition, but engineers and construction economists familiar with the pace of hyperscale deployment suggest the construction employment peak is likely 18 to 30 months away from the current activity level — potentially 2027 to 2028.

When that peak arrives and starts declining, the 9,000 monthly jobs currently offsetting AI displacement will begin shrinking, while the AI-driven displacement of white-collar roles is unlikely to slow at the same rate.

Who Is Actually Losing Their Jobs

The aggregate obscures a sharper demographic reality that the Goldman report documents with unusual directness.

Generation Z — workers under 30 — is facing the worst of the disruption, for a structurally inescapable reason. Entry-level roles across every knowledge-work sector have historically existed precisely because they involve the routine, high-volume, lower-complexity tasks that senior workers don’t have time to handle personally: data entry, legal document review, initial customer service escalations, billing queries, first-pass graphic design, basic code writing and testing.

These are also, not coincidentally, the exact tasks that current AI systems — particularly large language models with code and vision capabilities — can perform at roughly human competency levels at a fraction of the cost. Goldman’s tracker shows “a slight positive correlation between AI adoption rates and unemployment among workers under 30, measured across industries” — meaning that in sectors where companies are deploying AI fastest, youth unemployment is rising fastest.

The numbers become even more striking when you look at explicit attribution. Corporate layoffs attributed directly to AI automation reached a record 21,900 in April 2026, the highest monthly total since Goldman began tracking in 2023. The three-year total is approximately 136,000 AI-attributed job losses — a figure that understates true displacement because it only counts layoffs where companies explicitly cited AI as the cause, a disclosure most prefer to avoid.

The Productivity Paradox

The Goldman report does not omit the other side of the ledger. Generative AI is delivering a 23% average productivity uplift across knowledge work, according to academic studies cited in the tracker. More output per worker, per hour, is genuinely valuable — both for employers and, in aggregate, for the economy.

The catch, documented here and in research from MIT and Berkeley’s Center for Labor Economics, is that the productivity benefits are not distributed evenly across the workforce hierarchy. Senior workers — those with experience, contextual judgment, and established client relationships — are using AI as a lever that amplifies their individual output. Entry-level workers are not being empowered by the same tools; they’re being replaced by them.

The practical result is a compression of the traditional career pipeline. Companies are discovering they can hire fewer junior employees, lean harder on AI-assisted senior staff, and maintain or increase throughput. The entry-level positions that have historically served as the training ground for senior talent are disappearing before the pipeline of trained senior talent can be maintained.

The Sectors Under the Most Pressure

Goldman’s detailed sector breakdown highlights several areas with acute displacement velocity:

Marketing and creative work has seen the fastest adoption curve, with AI-generated content, automated A/B testing, and programmatic creative production eliminating substantial junior-level capacity. Design agencies and content studios have experienced the sharpest employment contractions.

Document processing — from legal discovery to insurance claims to regulatory filings — has been structurally automated. Law firms, financial services companies, and healthcare billing departments have all reduced headcount in proportion to AI adoption.

Software development, and particularly junior testing and documentation roles, has seen significant displacement. The irony of software developers building the tools that are displacing entry-level software developers has not been lost on the industry.

Customer service continues its long automation trajectory, now accelerated by voice AI capable of handling a vastly broader range of interactions than previous-generation chatbots.

What the Data Doesn’t Say

It is worth noting what Goldman’s tracker cannot measure, and the firm is careful to say so.

The model estimates AI-driven net displacement — it cannot independently assess how many of the 136,000 attributed job losses would have occurred anyway due to broader macroeconomic conditions, or how many new roles have been created by AI that simply haven’t been classified as AI-enabled. The AI-native job categories that didn’t exist five years ago — prompt engineers, AI operations specialists, model evaluators, synthetic data curators — are growing rapidly but don’t yet constitute a large enough aggregate to materially offset displacement in traditional categories.

Goldman also does not model the second-order effects of consumer spending changes: if AI-displaced workers reduce consumption, what knock-on effects does that produce in non-automatable sectors?

The Policy Response Vacuum

What makes the June 2026 Goldman report particularly striking is what’s absent from it: a credible policy response to the displacement trend.

The US has produced workforce retraining programs, expanded Pell Grant access, and maintained strong federal investment in community college AI curricula. None of these interventions operates at the scale or speed of the displacement they’re designed to address. Retraining a 24-year-old marketing coordinator to become an AI operations specialist takes 18 months; AI adoption in the next marketing department takes 18 days.

In Washington, the debate has fractured along familiar lines. The administration’s June 2026 AI executive order prioritized innovation competitiveness and explicitly declined to use federal procurement or labor law to slow AI adoption in the private sector. Democrats in the Senate have proposed expanded unemployment insurance duration for AI-displaced workers, but the bill has stalled without bipartisan support.

For the 11,000 Americans losing their net jobs to AI this month, the macro debate is largely academic. The Goldman report is the most granular measurement available of a transition that, for individual workers, is arriving not as a slow trend but as a termination notice.

The race between job destruction and creation remains tight. But for Generation Z entering the labor market in 2026, “tight” is cold comfort.