The Customer Is the Builder: how AI coding agents change software economics (arxiv.org)

AI coding agents are revolutionizing the economics of software development by enabling organizations to build custom functionality instead of purchasing it. This shift, described in a new analysis, focuses on how domain-savvy professional buyers can leverage these agents to specify and evaluate tools for their work, potentially changing the landscape of software procurement. The study reveals that the capacity of these AI agents has accelerated, with their ability to complete benchmark tasks doubling approximately every 129 days. Notably, it identified 575 potential open-source alternatives to commercial software, highlighting the expanding capabilities of coding agents, albeit with some functional matches requiring further validation. Despite increases in the market share of purchased software, a significant portion—32% of surveyed organizations—indicates they have opted not to buy software due to the potential to develop it internally with AI assistance. This trend could lead to a substantial shift in purchasing behavior, though current self-reports do not quantify the extent of spending displaced by these new tools. The findings emphasize the critical need for both organizations and vendors to adapt to this changing dynamic, as the influence of AI on software development could reshape competitive landscapes and customer decision-making processes in the near future.