US Programmer Job Growth Nearly Halved Since ChatGPT Launched, Fed Study Finds

Why It Matters

Slowing programmer hiring signals that generative AI is reshaping the tech labor market, potentially lowering firm costs while altering career pathways for coders. The trend also suggests AI‑driven productivity could offset headcount cuts, influencing broader economic dynamics.

Key Takeaways

• Programmer job growth fell from ~5% to ~2.5% annually
• Roughly 500,000 programmer jobs disappeared since ChatGPT debut
• Contract IT services employ 40% of US programmers; slowdown greatest there
• Programmer wages stayed flat even as hiring fell sharply
• Over 98% of measurement models label programmers as most AI‑impacted

Pulse Analysis

The Federal Reserve’s new occupational analysis cross‑referenced monthly employment surveys with the Department of Labor’s skill taxonomy to isolate programming‑heavy roles, which make up about 3.7 percent of the U.S. workforce. By constructing a counterfactual employment curve that holds the share of programmers constant within each industry, the study shows a persistent three‑percentage‑point annual decline in programmer headcount after accounting for broader tech sector headwinds. This methodological rigor strengthens the claim that generative AI, epitomized by ChatGPT, is a distinct factor reshaping hiring patterns.

For tech firms, the findings translate into a strategic pivot. Contract IT service providers—responsible for roughly 40 percent of all programmer jobs—experienced the sharpest slowdown, suggesting companies are substituting in‑house coding with AI‑assisted or outsourced solutions. Despite the hiring dip, wages have remained stable and job postings have started to climb again in 2024, indicating that demand for programming expertise persists but is being redistributed across roles. The productivity boost from large language models may enable firms to achieve more with fewer developers, reshaping budgeting and talent‑allocation decisions.

The broader labor‑market conversation must grapple with measurement challenges. While half of existing AI‑impact metrics converge on programmers as the most affected group, the lack of a universal occupational classification hampers precise forecasting. Policymakers and business leaders should monitor whether AI‑driven efficiency spurs new demand in adjacent fields or merely compresses existing roles. As AI capabilities mature, the tech sector could see a re‑emergence of hiring in niche areas, offshore shifts, or entirely new categories of software‑related work, making continuous data‑driven analysis essential.