AI is cutting 16,000 U.S. jobs a month — and Gen Z is taking the brunt, Goldman Sachs says

Goldman Sachs estimates AI erased roughly 16,000 net U.S. jobs per month over the past year (about 25,000/month lost to substitution versus ~9,000/month added via augmentation). A one-standard-deviation increase in AI substitution exposure is associated with a ~3.3 percentage-point widening of the entry-level (under-30) to experienced (31–50) wage gap, with Gen Z concentrated in high-risk routine administrative roles. Goldman notes the analysis is regression-based and likely overstates net losses because it undercounts hiring tied to AI infrastructure (data centers, power, construction) and demand created by AI-driven productivity gains.

Analysis

AI-driven substitution is already reshaping demand at the bottom of occupational ladders and creating a predictable bifurcation: capital-intensive infrastructure and high-skill platform owners capture outsized gains, while firms and real-estate tied to routine administrative throughput face persistent secular pressure. Expect this to manifest as a sustained divergence in returns between asset classes — data-center and electric-power exposure compounding over a multi-quarter pipeline, versus office/administrative real estate underperformance as desk-counts and support staffing shrink.

The timing matters: substitution effects show up quickly in payrolls and youth unemployment metrics (near-term, quarters), but the offsetting demand for AI-related construction, power, and specialized services accrues over the medium term (12–36 months) as projects move from planning to build. That lag creates an actionable window where mismatch risk is highest — firms with heavy entry-level hiring see margin relief but also reputational and consumption headwinds if payrolls and wages at the bottom compress.

Policy, regulation, and reskilling execution are the main catalysts that can re-rate this narrative. Rapid, employer-led reskilling or targeted fiscal incentives for AI-hardware builds would accelerate job-creation on the supply side and narrow the gap; conversely, caps on certain AI deployments, slower corporate capex, or a cyclical downturn would amplify displacement and extend the pain. Strategically, the trade is about capturing the infrastructure pivot and skill-upstreaming while hedging exposure to short-cycle macro risk and potential political/regulatory shocks.

Key Decisions for Investors

• Pair trade (6–18 months): Long Digital Realty (DLR) or Equinix (EQIX) + Short major office REITs such as Boston Properties (BXP) or SL Green (SLG). Rationale: data-center demand and power/real-estate premium should outpace legacy office leasing; target a 1:1 notional pair sized to <3% portfolio risk. Risk/Reward: asymmetric — potential 20–40% relative outperformance if AI capex continues; downside if broad commercial real-estate drawdown or recession compresses both.
• Directional infra play (12 months): Buy Jacobs Engineering (J) or Fluor (FLR) — overweight exposure to data-center construction and electrical/infrastructure projects. Positioning: 2–4% portfolio weight via stock or 12-month call spreads to cap downside. Risk/Reward: 30%+ upside on sustained build cycles; downside 20–30% on capex pullbacks or supply-chain inflation overruns.
• Skills/reskilling long (6–12 months): Buy Coursera (COUR) or Chegg (CHGG) via 9–12 month call spreads to play accelerated demand for AI upskilling and credentialing. Rationale: faster reallocation of displaced entry-level workers into higher-value AI-complementary roles. Risk/Reward: double-digit upside if enterprise adoption grows; limited loss via spreads if uptake is slower than expected.
• Tactical hedge/short (3–12 months): Buy puts or enter a modest short on contact-center/BPO operators (e.g., TTEC) to hedge exposure to rapid conversational-AI adoption. Structure: 3–9 month put spread to monetize near-term automation risk. Risk/Reward: profitable if automation reduces outsourcing volumes; loss if companies pivot successfully to higher-value managed services.