The 2025 Data on Kubernetes (DoK) Report Is Here
Five years ago, the idea of running data workloads on Kubernetes still raised eyebrows. Today, it’s the norm — and the question has changed from “Can we?” to “How do we do it best?”
The 2025 Data on Kubernetes Report, based on a survey of 182 technology professionals, marks a major inflection point for our community. The results show that DoK has officially crossed the chasm: adoption is complete, and the next chapter is all about optimization, cost management, and AI-driven innovation.
From Adoption to Optimization
Nearly half of organizations now run 50% or more of their data workloads on Kubernetes in production. Among the most advanced, that number exceeds 75%. Even more striking, 62% of respondents say that 11% or more of their company’s revenue is tied directly to DoK deployments.
This isn’t experimental anymore — it’s business-critical infrastructure. The focus has shifted from adopting Kubernetes to mastering it.
AI/ML Is Reshaping the DoK Landscape
AI/ML workloads are now the fastest-growing category running on Kubernetes, reaching 44% adoption in 2025. Behind that growth is an infrastructure revolution: 77% of organizations view vector databases as critical infrastructure, signaling the rise of RAG (Retrieval-Augmented Generation) architectures and a new era of real-time, intelligent systems.
As one respondent noted, “Our biggest challenge isn’t building AI models — it’s optimizing the infrastructure they run on.” That reality is redefining what “data on Kubernetes” really means.
Cost Optimization Takes Center Stage
For the first time, cost optimization has overtaken all other priorities, reflecting both the maturity and the complexity of today’s workloads. Organizations are tackling new cost drivers such as storage growth, GPU utilization, and cross-zone data transfer.
The focus for 2025 is clear: achieving operational excellence while keeping budgets in check.
The Edge + Real-time Shift
The data also shows a decisive move away from centralized architectures.
-
61% of respondents say edge computing is essential to their future data strategy.
-
64% say real-time data processing is critical for their AI initiatives.
This evolution is transforming how data is processed and served — with Kubernetes enabling systems that respond in milliseconds, wherever the data lives.
The Road Ahead: Skills, Performance, and Community
Despite these advances, challenges persist. Storage I/O performance and skills shortages remain top obstacles to optimization. Organizations need practitioners who understand not just Kubernetes operations, but also data workload tuning and AI/ML infrastructure management.
That’s where the Data on Kubernetes Community (DoKC) comes in. Our mission remains the same: to help practitioners share best practices, learn from each other, and push the boundaries of what’s possible with data on Kubernetes.
Get the Full Report
The 2025 report offers deep insights into:
-
Production maturity and workload evolution since 2021
-
How AI/ML workloads are reshaping infrastructure priorities
-
Real-world cost, performance, and operational metrics
-
The next wave of technology trends shaping DoK