Junior hiring shrinks

Software-developer employment for workers ages 22 to 25 is down nearly 20% from its late-2022 peak, even as older developers’ headcounts kept growing. (hai.stanford.edu) Stanford’s 2026 Artificial Intelligence Index highlighted the drop, and ADP Research said the pattern comes from payroll data covering millions of workers at thousands of private companies through July 2025. ADP said the youngest software developers were 20% below their late-fall 2022 peak, while workers 30 and older in highly AI-exposed jobs grew 6% to 13%. (hai.stanford.edu, adpresearch.com) A separate Northeastern University paper found the public release of ChatGPT in November 2022 was followed by a 16.3% drop in the relative share of junior-versus-senior software-developer vacancies. The authors said the decline was concentrated in larger firms and bigger cities. (aliciasassermodestino.com) Generative artificial intelligence writes, summarizes, and debugs code from text prompts, which makes routine programming tasks faster for people who already know a codebase. Indeed Hiring Lab said software development is among the occupations most exposed to this kind of job transformation, not just outright replacement. (hiringlab.org) Stanford’s broader labor-market paper described the shift as an early-career effect, not a uniform collapse in hiring across all ages. Its authors found early-career workers in AI-exposed occupations had 16% relative employment declines after controlling for firm-level shocks, while experienced workers stayed stable. (digitaleconomy.stanford.edu) That helps explain why the pain is showing up first in entry-level roles: companies can use AI tools to absorb coding, testing, and documentation work that once trained new hires. Mark Russinovich and Scott Hanselman wrote in March 2026 that the new incentive is to “hire seniors, automate juniors,” while warning that cutting early-career hiring would weaken the profession’s talent pipeline. (cacm.acm.org) The same data does not show broad wage cuts or across-the-board layoffs for all software workers. Stanford’s November 2025 paper said the adjustment was happening “primarily via employment rather than compensation,” which fits a market where firms slow junior hiring and rework staffing plans instead of firing senior teams en masse. (digitaleconomy.stanford.edu) The backdrop is unusually fast adoption. Stanford’s 2026 index said generative AI reached nearly 53% population-level adoption within three years, faster than the personal computer or the internet, putting hiring decisions under pressure long before companies settled on a stable training model for new engineers. (hai.stanford.edu) For recent graduates, the market is not disappearing so much as narrowing: fewer classic junior developer openings, more pressure to arrive with AI-tool fluency, and fewer employers willing to treat the first job as paid training. The data so far shows the biggest contraction at the bottom rung. (aliciasassermodestino.com, adpresearch.com)