Apollo Study Links Higher AI Exposure to Weaker US Wage Growth with Limited Employment Effects

US labor data through August 2026 indicate AI deployment first compresses compensation rather than headcount. Slok's occupation-level regressions find a 1.2 percentage point gap in year-over-year wage growth between high- and low-exposure roles, with employment elasticities near zero. This pattern aligns with firm-level decisions tracked by former IBM HR chief Diane Gherson, where automation savings route through hiring pauses and targeted reskilling rather than mass layoffs.

Record business formation cited in the same dataset—reaching 5.5 million applications in the twelve months to June 2026—creates countervailing demand for new roles in AI-adjacent services. These entrants operate at smaller scale and higher productivity, shifting the composition of labor demand without restoring wage pressure in legacy occupations. Primary BLS occupational employment statistics confirm the divergence between establishment births and incumbent wage trajectories.

The incentive structure favors capital owners and new entrants over incumbent workers whose tasks are codifiable. States gain productivity metrics useful for fiscal projections and trade positioning, yet face concentrated wage stagnation that may require later fiscal offsets. Primary records from the Department of Labor show no corresponding acceleration in reemployment rates for displaced cohorts.

Next data releases from BLS and Census Bureau in Q4 2026 will test whether the wage-employment split persists or migrates into measurable net job losses once reskilling cycles complete.