The OpenTelemetry (OTel) project officially achieved graduated project status at the Cloud Native Computing Foundation (CNCF) in May 2026, after beginning its graduation path in 2025. This achievement places the project alongside prominent CNCF projects such as Kubernetes and Prometheus, but it does not represent the end of its development; rather, it confirms its transition from a promising project to a mature open architecture targeting broad enterprise use.
Why did OpenTelemetry emerge?
Before OpenTelemetry, observability tools generally relied on vendor-specific formats and libraries. Moving between observability providers required removing and replacing legacy integration libraries, making vendor changes complex and increasing the risks of technical lock-in. The core observability signals—traces, logs, and metrics—were also treated as separate systems, making it difficult to correlate them to obtain a comprehensive view of application performance.
In May 2019, OpenTracing and OpenCensus merged to establish OpenTelemetry. The project provides specifications for traces, metrics, and logs; unified APIs; implementations of those APIs in different languages; and the OpenTelemetry Collector. OpenTracing was archived in January 2022, followed by OpenCensus in July 2023, while OpenTelemetry became the de facto open standard for telemetry, according to a CNCF post.
What qualified the project for graduation?
Graduation status was based on a set of practical criteria, not on the project's size alone. Organizations such as GitHub and Farfetch use OpenTelemetry in production environments, and the project has a documented governance model defining election and resignation roles, decision-making mechanisms, and communication processes.
The review also covered community health, with regular contributors from multiple organizations and a clear process for reviewing pull requests and managing issues. OpenTelemetry underwent an independent security audit, with all critical issues identified during it addressed. Its APIs also met requirements for stability, release management, and backward compatibility, while the documentation covers architectural aspects and guidance for users, operators, and contributors. The graduation application was ultimately submitted to the CNCF Technical Oversight Committee for review.
What changed in practice for users?
The CNCF post says the project recorded more than 12,000 contributions from more than 2,800 companies, with hundreds of maintainers participating in language-specific special interest groups, making it the second-highest CNCF project by pace of activity after Kubernetes. Traces, logs, and metrics reached general availability, while profiling was added as a new signal, and both the demo and the Collector and system components expanded.
In practical terms, graduated status gives organizations an additional indicator of governance, security, stability, and adoption maturity, but it does not eliminate the need to assess the project's suitability for each environment. Graduation also does not mean that all observability challenges have been solved or that adopting OpenTelemetry comes without operational or engineering costs; the source makes clear that the project will continue to evolve.
The next phase
Upcoming work focuses on observing new types of workloads, including agentic workflows, through emerging semantic conventions for generative AI, alongside addressing the observability needs of browsers and mobile devices. The project is also moving toward helping mature teams manage telemetry at scale using Weaver to define and govern telemetry schemas, OpenTelemetry Packaging to provide components in installable packages, and OpenTelemetry Injector to enable instrumentation without modifying code.
Editorial reading from certi.news: The practical change is that OpenTelemetry is no longer merely a project for standardizing observability tools; it has become a project that has passed a stated set of adoption, governance, security, and API-stability tests. This matters to teams that have postponed standardizing their telemetry layer out of concern that the project was not mature enough. Nevertheless, open questions remain around the community's ability to maintain this level as usage expands, and around how the new standards for AI workloads and multiplatform observability will translate into stable implementations.