JetBrains Report: Self-Hosted CI/CD Dominates as AI Integration Remains External to Pipelines
A recent report from JetBrains, drawing on multiple surveys including their CI/CD Tools Adoption Survey 2026, reveals that self-hosting of primary CI/CD tools has reached a five-year high, with 72.8% of developers running their CI/CD systems on infrastructure they control. This includes both on-premises installations (30.5%) and self-managed cloud instances (42%). This marks a notable departure from the peak of fully managed cloud CI services in 2022, which has since seen a seven-point decline.
This trend matters significantly to practitioners as it signals a potential re-evaluation of the trade-offs between convenience and control. Organizations may be seeking more granular control over their build environments, data residency, and security postures, which self-hosting often provides. The report also indicates that while nearly half of organizations (48.8%) use AI for at least one CI/CD task, only a small fraction (8%) integrate AI-powered steps directly within their build or test processes. This suggests a cautious, often external, application of AI in CI/CD, such as for code analysis or test generation, rather than directly influencing pipeline execution logic.
This development fits into a broader, well-established trend in cloud and DevOps where organizations continually balance the agility of cloud services with the need for customization, compliance, and cost efficiency. The rise of platform engineering, for instance, often involves building internal platforms that abstract underlying infrastructure, giving developers self-service capabilities while retaining organizational control. The reluctance to embed AI directly into pipelines, despite its widespread use around them, reflects a pragmatic approach to new technologies in critical paths. The industry is still exploring how to best leverage AI for reliability and efficiency without introducing new risks or complexities into core delivery processes.
In practice, this means DevOps teams should be evaluating their current CI/CD hosting strategy. If using fully managed services, it might be time to assess whether the benefits still outweigh the potential for vendor lock-in or lack of customization. For those already self-hosting, the report validates their approach, emphasizing the importance of robust infrastructure management and operational expertise. Furthermore, practitioners should continue to experiment with AI in their CI/CD workflows, focusing on areas where it can augment human capabilities and improve feedback loops, such as intelligent test selection or anomaly detection in logs, rather than immediately handing over control of core pipeline steps to AI agents. The key takeaway is to build a CI/CD strategy that prioritizes control, security, and measurable value, whether that involves self-hosting or carefully integrated cloud services, while strategically adopting AI where it provides clear, auditable benefits.
Read original source