Carbon-Aware AI Control Plane for DevOps Automation: A Reference Architecture and Next-Generation Sustainability Framework

This paper presents a reference architecture and conceptual framework for integrating carbon awareness into DevOps automation systems. AI workloads are becoming one of the fastest-growing contributors to the global computing demand; however, DevOps systems that deploy and operate them remain completely carbon-blind. They make real-time decisions about scaling, deployment, and rollback using only performance, cost, and reliability signals, without any awareness of the environmental consequences of those actions. We introduced a carbon-aware AI control plane as a new architectural layer that elevates sustainability to a first-class decision input within DevOps automation. We propose a vendor-neutral reference architecture that augments existing CI/CD, autoscaling, and release orchestration systems with predictive carbon intelligence, temporal decision logic, and policy-defined sustainability guardrails. The framework integrates standard container orchestration platforms, continuous integration systems, GitOps tools, and cloud-native scaling engines without disrupting developer workflows. We formalize the control loops and governance model required to operationalize sustainability-aligned automation, and outline a comprehensive methodology for organizations to validate and adopt such systems. We argue that carbon-aware automation represents the next natural evolution of DevOps maturity, following the industry’s earlier shift toward reliability-driven SRE practices.