Karmada Achieves Graduation from Cloud Native Computing Foundation Recognizing Major Advancements

Karmada Officially Graduates from the Cloud Native Computing Foundation

On September 8, 2026, during the KubeCon + CloudNativeCon + OpenInfra Summit in Shanghai, the Cloud Native Computing Foundation (CNCF) announced the graduation of Karmada. This graduation signifies Karmada's transition into a fully mature open-source project that enables organizations to orchestrate applications across multiple Kubernetes clusters, regions, and clouds without the need to alter their underlying applications.

Karmada's journey began with its first commit in November 2020 and evolved through various phases within the CNCF, officially joining as a Sandbox project in September 2021. After demonstrating substantial growth, organization support, and operational stability, Karmada advanced to Incubation status in December 2023 and achieved its graduation milestone today.

The recent release of Karmada v1.19 brings significant enhancements, particularly multi-component scheduling that optimizes distributed AI training jobs. The upgrade also introduces a priority-based scheduling feature that allows critical workloads to be prioritized, thereby ensuring they receive the necessary resources. These advancements are crucial, especially as global enterprises ramp up their AI infrastructures, demanding reliability and performance in their multi-cluster operations.

Organizations across various industries have turned to Karmada for its capabilities. Notable users include tech giants such as Bloomberg, Wellhub, Alibaba Cloud, Huawei, and Trip.com. They leverage Karmada to maintain hybrid cloud capabilities, ensure resilience across regions, and create a dependable infrastructure tailored for AI applications. Chris Aniszczyk, the CTO of CNCF, stated, "As organizations scale Kubernetes across clusters and GPU-constrained AI environments, having a production-ready method to coordinate that fleet becomes critical for operational success."

Karmada, a term that encapsulates 'Kubernetes Armada', expands upon the standard Kubernetes API to facilitate centralized placement, failover, and autoscaling across multiple clusters. This model supports efficient resource utilization and simplifies the complexities associated with managing Kubernetes at scale—traits that are becoming increasingly vital in today's fast-paced technological landscape.

More than 1,200 contributors from a wide array of organizations have supported the project since its inception, showcasing Karmada’s collaborative efforts within the open-source community. The tool's popularity is evident, as it has garnered over 5,600 stars on GitHub, a testament to its community validation and developer interest.

State-of-the-art features like Prometheus metrics integration, packaged etcd instances, and Helm chart installation support have made Karmada an essential tool for many enterprise users managing vast Kubernetes ecosystems. Its ability to seamlessly work with existing CNCF observability and deployment projects has also fostered integration with numerous tools and processes within organizations.

This graduation marks a newfound commitment to enhancing Karmada's capabilities further and responding to community needs. The project's roadmap for 2026 is robust, aiming to refine resource-aware control planes, implement multi-cluster queuing strategies for AI training and batch jobs, and bolster support for Kubernetes Dynamic Resource Allocation across various hardware accelerators.

Karmada's graduation reflects its readiness for enterprise use, having undergone thorough scrutiny, including a rigorous third-party security audit, establishment of a well-defined governance model, and adherence to CNCF's Code of Conduct. Maintaining a Core Infrastructure Initiative (CII) Best Practices Badge, the project highlights a firm commitment to secure software development.

As the demand for multi-cluster orchestration solutions continues to rise, Karmada stands at the forefront, ready to drive innovation in cloud-native architectures. As shared by project maintainer Hongcai Ren, "Graduation is merely a new starting point. We look forward to collaborating with more adopters and addressing the complexities of AI and infrastructure."

For more comprehensive information about Karmada and its features, visit the official Karmada website.