→ Back to Home
CI/CD

JetBrains 2026 CI/CD Report Reveals Shift Towards Self-Hosting and Agent-Driven AI Integration

The JetBrains "State of CI/CD in 2026" report, drawing from extensive developer surveys, reveals several key shifts in the continuous integration and continuous delivery landscape. Notably, self-hosting of primary CI/CD tools has reached a five-year high, with 72.8% of developers running their systems on infrastructure they control, and on-premises installations specifically hitting 30.5%. This marks a significant departure from the trend towards fully managed cloud CI, which has seen a seven-point decline since its 2022 peak. This trend matters to practitioners because it indicates a potential re-evaluation of the trade-offs between convenience and control. While managed services offer ease of use, the increasing complexity of enterprise environments, coupled with growing concerns around data sovereignty, cost optimization, and deep customization, may be driving teams back to self-managed solutions. This allows for greater flexibility in integrating specialized tools, managing sensitive data, and fine-tuning performance, albeit with increased operational overhead. The report also highlights that while nearly half of organizations use AI for CI/CD tasks, only 8% run AI-powered steps *inside* a build or test, suggesting AI is primarily used for auxiliary tasks like code analysis or test generation rather than core pipeline execution. This development fits within a broader, well-established trend in DevOps where organizations seek to balance agility with control. The initial rush to cloud-native and fully managed services is now being tempered by a more nuanced understanding of operational realities. The rise of platform engineering, for instance, aims to provide developers with self-service capabilities on top of carefully curated, often self-managed, infrastructure. Similarly, the cautious integration of AI reflects a pragmatic approach; while AI offers immense potential for efficiency gains, its application in critical CI/CD paths is still being explored, with a preference for human oversight and validation. The report also notes a polarization in tool stacks, with fewer two-tool setups (suggesting completed migrations) and a growth in three-or-more-tool setups, indicating diverse needs or specialized tooling requirements. In practice, this means DevOps teams should critically assess their CI/CD strategy. For those considering a move to self-hosted solutions, it's crucial to factor in the increased responsibility for maintenance, security, and scaling. For teams already self-hosting, the report validates their approach and encourages continued investment in robust internal platforms. Furthermore, practitioners should explore how AI agents and tools can augment their CI/CD workflows *around* the pipeline, focusing on areas like intelligent test selection, anomaly detection in logs, or automated documentation, rather than immediately trying to embed AI directly into build and deploy steps. The emphasis remains on a well-architected, observable pipeline, regardless of its hosting model, with AI serving as an intelligent assistant rather than a fully autonomous operator in the critical path.
#ci/cd#devops#self-hosting#ai#developer survey#platform engineering
Read original source