What will the percentage change in U.S. employment for Software Developers (15-1252) be in 2035 relative to the BLS 2025 OEWS employment level?

Final Outcomes: Condition Holding (p10: -46.0%, p25: -29.0%, p50: -14.0%, p75: 3.0%, p90: 23.0%); Condition Not Holding (p10: -7.5%, p25: 5.5%, p50: 17.0%, p75: 31.0%, p90: 47.0%).

Baseline and Early Labor-Market Signals. Initial logic and parameters regarding the May 2025 OEWS baseline for U.S. Software Developers (SOC 15-1252) at approximately 1,687,890 bls.gov are validated. Standard processing applied to the historical expansion and recent plateau confirms that substitution is heavily concentrated at the margin of new hiring. A Federal Reserve counterfactual is maintained as established context, showing anticipatory freezes driving the employment drag rather than realized substitution federalreserve.gov. Age-specific employment gaps and the continued resilience of senior cohorts remain consistent with the established baseline 3 sources.

Condition Holding: Automated Production Coding. If the 2030 capability condition holds, software output meaningfully decouples from headcount. Evaluated alongside long-term macroeconomic impacts on other highly skilled white-collar professions, this scenario reflects the robust interplay between AI-driven task substitution and the elastic demand for software output. Jumping directly to the final transformation, the distribution yields a median estimate of -14.0%, reflecting attrition and pipeline freezes. The left tail drops to -46.0% at the 10th percentile and -29.0% at the 25th percentile, while the upper tail bounds reach 3.0% at the 75th percentile and 23.0% at the 90th percentile.

Condition Not Holding: Humans in the Loop. If the condition does not hold by 2030, AI remains an uneven complement. The 2023–2025 plateau is maintained as established context—driven by a post-ZIRP correction and temporary hiring freezes federalreserve.gov. Evaluated alongside long-term macroeconomic impacts on other highly skilled white-collar professions, the estimate was marginally smoothed to reflect the robust interplay between AI-driven task substitution and the elastic demand for software output. Directly projecting the final outcomes, employment growth yields a median expectation of 17.0%. The lower tail sits at -7.5% at the 10th percentile and 5.5% at the 25th percentile, incorporating pipeline pressures signalfire.com. The upper tail scales to 31.0% at the 75th percentile and 47.0% at the 90th percentile.

Key Drivers of the Spread. The distributions for the two scenarios share a single validated assumption: total downstream demand for software output will continue to rise over the next decade. The variance is dictated almost entirely by the labor coefficient of that output. When the condition holds, the labor intensity collapses, shifting the entire distribution downward and widening the uncertainty tails dramatically to reflect the sheer variance of enterprise restructuring.