CNCF年度云原生调查：人工智能未来的基础设施

The infrastructureof AI’s future Adrienn Lawson, The Linux FoundationJeffrey Sica, Cloud Native Computing Foundation (CNCF)Foreword by Jonathan Bryce, Executive Director, Cloud andInfrastructure,The Linux Foundation January 2026 CNCF Annual Cloud Native Survey: The infrastructure of AI’s future Contents Foreword ............................................................................................................................................4Executive summary ...........................................................................................................................5Introduction .......................................................................................................................................5Kubernetes as the AI platform .........................................................................................................6The maturity of cloud native infrastructure ................................................................................... 9Cloud native maturity profiles ..........................................................................................................13 Foreword If we cut through the hype of chatbots and agents, we can clearlysee that we will need to greatly decrease the difficulty of servingAI workloads while massively increasing the amount of inferencecapacity available across the industry. I believe this is the next This year, we celebrate the 10-year anniversary of the CloudNative Computing Foundation. A decade ago, the concepts wenow consider industry standards were a revolutionary vision. I was honored to step into the role of Executive Director midwaythrough this banner year. This report’s findings confirm thatcloud native has truly arrived. 98% of organizations have adopted This new wave of AI also brings a new responsibility. As the reportwarns, “machine-driven automated usage” puts a strain on theopen source systems we all depend on. Our next decade must be This maturity has revealed our community’s next greatopportunities. The primary challenge in cloud native is no longertechnical complexity; 47% of organizations now cite “culturalchanges with the development team” as their top obstacle. This isthe “why” behind Platform Engineering. Our community is leading Thank you to everyone who contributed code, documentation,testing, stories, and support this year. If you’re new to CNCF,welcome; you’ll find a community that turns hard problems into Onward to the next decade of cloud native! JONATHAN BRYCE, Executive Director, Cloud and Infrastructure, This stable, platform-centric foundation is also enabling the nextmajor shift impacting the entire industry: Artificial Intelligence. This report is titled “The infrastructure of AI’s future,” and its datareveals the emerging substrate: 66% of organizations are alreadyusing Kubernetes to host their generative AI workloads. But thereal story isn’t the one in the headlines. It’s not about training The 2025 CNCF Annual Report reveals a cloud native ecosystemthat has reached a critical inflection point. What began as anexperimental architectural approach has solidified into an essential systems that help build, test, deploy, and distribute Second, the data demonstrates that cloud native maturity followsa predictable progression model. Organizations advance throughfour distinct stages: explorers, adopters, practitioners, and Three key themes define 2025: First, Kubernetes has evolved from container orchestrator toAI infrastructure platform, with 66% of organizations runninggenerative AI workloads on it. The primary obstacles have shifted Third, the report reveals a profound gap between AI ambitionand infrastructure reality. While headlines concentrate onmodel breakthroughs, 47% of organizations deploy AI models Introduction Cloud native technologies have become fundamental to modernenterprise infrastructure. As organizations deploy AI workloads,modernize applications, and pursue operational agility, technology adoption and maturity, revealing not just currentpractices but also trajectory and evolution of the cloud native The Cloud Native Computing Foundation (CNCF) ecosystem nowcomprises 234 projects at various maturity levels, supported byover 270,000 contributors from all over the globe. This report This report presents findings from the 2025 CNCF Annual Survey,which examines organizational perspectives on containers,Kubernetes, and cloud native adoption trends. The survey Kubernetes as the AI platform While headlines tout AI breakthroughs, a quieter revolution is happening in theinfrastructure layer. CNCF survey data reveals that 66% of organizations arebetting on Kubernetes to run their generative AI workloads. However, success proprietary ML platforms, concerns about operational complexity, or the need Moving AI workloads to Kubernetes isn’t simply about containerization.Organizations must address unique requirements such as managing largemodel art