AI Is Driving a Boom in Modestly Successful Software Projects

Summary

The article examines why the rise of AI coding agents has not yet produced a visible explosion of blockbuster software. Comparing roughly twenty software marketplaces, plugin registries, and modding platforms across 2022-24 and 2025-26, the author finds about twice as many modestly successful projects in several categories, against roughly ten times as many total projects. NVDA screen-reader add-ons with more than 1,000 downloads more than doubled, while higher-download add-ons did not; HACS plugins with more than 100 installs also doubled without a comparable increase above 1,000 installs. New Steam games with more than 100 reviews rose by about 60%, and Obsidian, browser-extension, game-mod, and Blender-extension ecosystems showed similar patterns, although Google Play and WordPress did not. The author argues that installation and usage counts are more informative than AI-disclosure labels, because hobbyist communities often discourage admitting AI use. The evidence therefore suggests that AI is expanding ordinary, useful software made for hundreds or thousands of users, rather than creating many new mass-market winners. The explanation for the gap is speculative: established companies may already have most obvious features and face technical constraints, while genuinely breakout hobbyist ideas may be rare or require years to mature. The analysis itself used AI to help scrape and analyze data, with human guidance.