Research, design, and develop computer or network software and specialized utility programs; analyze needs and build software solutions.
AI capability is already advanced across coding, debugging and implementation, but occupation-wide labor effects are arriving unevenly and are clearest in junior hiring and team leverage. EOL expects the positive BLS structural path to weaken materially as software output becomes less dependent on proportional developer headcount.
Growth slowing / plateau risk before 2030
The 2030 range remains close to flat because diffusion, integration and organizational redesign take time. The 2035 downside is much larger because agentic systems have more time to absorb end-to-end production work and allow smaller teams to generate more software.
AI-induced software demand and new applications can offset productivity, but only if they require proportional developer labor.
Agentic workflow substitution; output per developer.
Frontier agents now complete multi-hour software tasks on controlled benchmarks; coding is a leading AI use case.
Directional summary of whether current evidence pushes EOL toward greater or lesser human labor demand relative to the occupation’s existing structural trajectory. It is categorical, not a probability, percentile, automation share, or mathematical adjustment.
AI capability and agentic production are changing the link between software output and developer headcount faster than historical relationships imply.
If software-developer demand weakens, the first adjustment is likely to appear in campus and entry-level hiring rather than broad incumbent displacement. The computing pipeline is already contracting, which could partially cushion a later demand decline. Early-career hiring therefore remains a leading indicator of whether AI-driven productivity is translating into reduced labor demand.
How EOL analyzes labor supply →