The Cloud Native Computing Foundation (CNCF) and SlashData announced the findings of a new report on the development of cloud-native computing in China, coinciding with KubeCon + CloudNativeCon, OpenInfra Summit, and PyTorch Conference China events in Shanghai on September 8, 2026. The report estimates that China has approximately 1.75 million cloud-native developers, including nearly 400,000 developers working in AI.
The data indicates that 48% of professional industrial Internet of Things (IIoT) developers in China are classified as cloud-native developers, compared with 42% globally. The proportion of Chinese backend developers classified in this category also rose to 48%, up from 30% two years ago, approaching the global rate of 52%.
From Experimentation to Production Inference
The report focuses on how infrastructure changes as AI teams move from experimentation and training to serving models in production environments and distributed inference. It presents a progression in which tools and practices change according to workload maturity.
During data pipeline operations, Kubernetes and microservices provide the foundational infrastructure layer, followed by event-driven architectures, streaming services, and observability tools. During training and experimentation, feature flag management mechanisms, RPC interfaces, and immutable infrastructure help direct model releases, support distributed training, and maintain reproducible environments.
When models are served in production, service meshes, chaos engineering, and multicluster management become more prominent, particularly as traffic distribution, resilience testing, and distributed inference at greater scale become necessary.
A New Generation Adopts the Cloud-Native Approach
Adoption appears to be strongest among younger developers. The proportion of Chinese backend developers under 25 classified as cloud-native developers reached 58%. The report views this figure as indicating that cloud-native technologies and practices have become an early part of how the new generation builds backend services.
The report also found that 88% of backend developers currently work with some form of infrastructure standardization, up from 80% six months ago. Platform engineering can hide some Kubernetes and container details from developers, but it does not diminish the importance of those layers for operating production workloads.
certi.news Analysis: What Is Changing in Practice?
The most significant value in the findings is not merely the increase in the number of developers, but the connection between cloud-native tools and specific stages of the AI lifecycle. As teams move closer to operating models at production scale, practices such as immutable infrastructure, chaos engineering, service meshes, and multicluster management converge around requirements for consistency, resilience, and distribution.
The report states that the strongest technical association in the dataset was between immutable infrastructure and chaos engineering, with a lift of 2.22, meaning that their combined adoption exceeded the expected adoption rate by more than double. This remains an indicator of an adoption pattern observed by the study, not evidence of a causal relationship. The figures also apply to the report's sample and methodology and, by themselves, do not indicate the scale of spending, the performance of each organization, or the extent to which all sectors of the Chinese market are represented.
Chris Aniszczyk, CTO at CNCF, said that the challenges of scaling inference—such as scheduling, isolation, networking, observability, and operating distributed systems—overlap with the challenges of cloud-native computing. The report presents China as an important market for understanding this intersection, noting that the global cloud-native developer community has reached 19.9 million developers.