Contrary to expectations, I believe this new AI-driven era will see a surge in high-quality open-source software.
From a corporate perspective, while it becomes easy to generate a library for a specific purpose, maintaining it is difficult if the original author was an AI. In a world where software creation is commoditized, it makes sense to share the burden of maintaining functional code over time—even with competitors. There is simply no value in keeping it private, but there is a tangible benefit in sharing it.
From the perspective of traditional open-source maintainers, AI makes it feasible to maintain even niche products. Here, “niche” does not mean “useless,” but rather “serving a small community.” For example, with AI-support, maintaining a large open-source library with a legacy in high-performance computing (HPC) suddenly becomes viable, even as an individual developer.
This is demonstrated by the fact that the most common benchmark for AI coding proficiency is the SWE-bench, which tests an AI’s ability to fix bugs in real-world GitHub code. Recent models surpass an 80% success rate, meaning they can resolve the vast majority of bugs in open-source projects.
Therefore, open source is not just staying; it is poised to thrive.
