A longitudinal study tracking 40 companies from November 2024 to February 2026 reveals AI adoption has boosted developer productivity by just 9.97%, far below the 2-3x gains promised by marketing materials. While AI usage jumped 65% during this period, pull request throughput only saw single-digit percentage improvements.

DX researchers filtered out teams with individual PR targets to avoid gaming effects, making this figure particularly robust. Engineering leaders across organizations report similar modest gains of 8-12%, creating a significant gap between executive expectations and on-the-ground reality. The study suggests that while AI accelerates coding tasks, it barely touches the planning, alignment, and code review work that consumes most developer time.

Developers consistently told researchers that coding was never the true bottleneck in software development. As one senior developer explained, AI makes tedious tasks less annoying and might reduce a four-day task to three days, but this doesn't translate to shipping three times more pull requests. The findings indicate that AI is moving the needle but leaders may need to reset expectations internally about what these tools can deliver.