A study tracking over 100,000 GitHub developers finds autocomplete lifts coding activity by 40 percent and autonomous agents by 180 percent. Finished software releases, however, grew by only 30 percent.
A study by Wharton professor Leon Musolff together with MIT researchers Mert Demirer and Liyuan Yang tracked more than 100,000 developers on GitHub between 2022 and 2026. Autocomplete tools raised coding activity by 40 percent, sync agents that edit code alongside developers in real time lifted the cumulative figure to 140 percent, and autonomous async agents pushed it to 180 percent. Yet software projects rose only 50 percent, and finished releases grew only 30 percent.
The authors argue the constraint has shifted away from writing code toward review, integration, and distribution. Across the four largest app marketplaces, including the Apple App Store and Google Play, monthly new releases on the App Store rose from about 30,000 in early 2025 to roughly 100,000 by April 2026, while overall usage stayed flat. Musolff says AI tools that review AI-written code do not solve the problem, since their review quality does not match human judgment.
For companies, the lesson is that investment in AI coding tools only pays off if review, testing, release processes, and user acquisition scale up alongside it. It remains open whether future tool generations will automate these later stages too, and whether more published applications will actually reach more users.
What this means for decision-makers
- Check whether your review and approval processes keep pace with rising coding productivity.
- Invest specifically in integration and testing capacity, not only in coding tools.
- Watch app marketplaces for saturation before committing resources to raw release counts.
This story was produced automatically from the source named above and checked by software before publication. The image is symbolic and shows neither the event nor a real person. How this paper is made