CircleCI’s Data Shows AI Is Widening the Software Delivery Gap

On Feb 18, 2026, from San Francisco, Circle Internet Services, Inc., known to the market as CircleCI, released its 7th annual State of Software Delivery report, and the signal is not subtle. Built on nearly 28M workflows executed across more than 22,000 organizations in 149 countries during the first 28 days of Sept 2025, the dataset shows a 59% YoY surge in average daily workflow throughput. That is the largest increase since 2019. In the world of SaaS, that kind of acceleration is not cosmetic. It is operational pressure measured in commits, tests, and deployments.

The top 5% of teams on CircleCI nearly doubled throughput, jumping from 6.8 to 13.4 daily workflow runs. The top 10% gained 47%. The top quartile rose 25%. Then gravity returns. The median team increased just 4%. The bottom 25% showed no measurable movement. Same AI tailwind. Same market conditions. Radically different execution. This is not a tooling gap. It is a systems gap, and in SaaS, systems are destiny.

Rob Zuber, CTO at CircleCI, is clear about what separates winners from spectators. “The industry is far past thinking about how to use AI to accelerate software delivery,” Rob Zuber said. “The teams that are realizing AI’s benefits in software development are using autonomous validation to run laps around the competition that can’t validate AI-generated code at scale.” Translation for operators: AI can generate code faster than ever, but without validation at scale, you are manufacturing risk at scale.

CircleCI positions itself as infrastructure for teams to ship confidently at AI speed, serving more than 2M developers at companies including Okta, Hinge, and Hugging Face. Through pipeline automation, the company claims to reduce time from build to delivery to 3x the industry average. Sponsored this year by Thoughtworks, the report frames a widening performance divide: elite teams are converting AI-driven coding velocity into deployable change, while others are flooding pipelines that were never designed for this volume.

This is where SaaS economics get real. Throughput without reliability is noise. Code without feedback loops is liability. The elite 5% have tightened validation, compressed recovery cycles, and aligned automation with scale. The bottom quartile has more code and the same constraints. AI did not create the gap. It revealed it.

For leaders operating inside the modern SaaS stack, the takeaway is direct. The competitive edge is no longer who can write more code. It is who can validate, integrate, and recover faster than the code arrives. CircleCI has quantified the divergence. The market will decide who closes it.