The News
Komodor released its 2025 Enterprise Kubernetes Survey, which highlights rising complexity in cloud-native environments, with most organizations now operating dozens to hundreds of clusters across hybrid and edge deployments. The report underscores persistent challenges in change management, cost control, and skills shortages, even as GitOps, Helm, and platform engineering become standard practice.
Analyst Take
Komodor’s data shows Kubernetes has crossed into mainstream production with ~80% of organizations running Kubernetes in production, with 37% managing more than 100 clusters and 12% managing over 1,000. Multi-cluster and hybrid cloud deployments are the norm, with 48% operating across four or more environments. This aligns with theCUBE Research Day 0 findings, where 76.8% reported GitOps adoption and 54.4% said hybrid was their dominant deployment model.
While this scale unlocks flexibility, it introduces fragility. Nearly 79% of outages originate from recent system changes. Faster releases amplify risk when change management is inconsistent.
Change Management Is Still the Weakest Link
Despite advanced automation, instability remains rooted in change. Komodor found cross-org mean time to detect (MTTD) at 37 minutes and mean time to resolve (MTTR) at 51 minutes. Historically, developers addressed this by over-provisioning (Komodor notes 65% of workloads run at <50% utilization) and using siloed monitoring tools. Our Day 0 data confirms this pattern with over 50% of respondents using multiple observability tools like Datadog, Prometheus, and Elastic, but 41.1% cited lack of expertise as a key security/configuration gap. Tool sprawl has created blind spots instead of resilience.
AI/ML as the Next Kubernetes Frontier
AI/ML workloads are becoming common on Kubernetes, from batch pipelines (11%) to real-time inference (10%). This matches our Day 1 survey, where 74.3% of enterprises listed AI/ML as their top spending priority. Yet operational inefficiencies, especially under-utilized GPUs and fragile scheduling, mirror the earlier CPU over-provisioning wave. Without advanced orchestration, AI could worsen cost and performance pressures.
Platform Engineering and AIOps as Stabilizers
To counter sprawl and skill shortages, 68% of Komodor customers have established platform teams, often building Internal Developer Platforms (IDPs). theCUBE’s Day 2 data similarly found that 64.2% see observability as essential to DevOps strategy, and 71% are already using AIOps to manage scale.
Previously, ops teams leaned on manual reviews or post-mortems to address incidents. Now, Komodor’s data shows organizations using unified telemetry and GitOps automation report 50% less engineering time on disruptions. AIOps adoption is resurging (Komodor found 35% in use, 40% exploring) which could help close the experiment-to-production gap for both ops and AI workloads.
Looking Ahead
The industry is pivoting from container adoption toward operational excellence. Kubernetes is no longer the barrier to entry, cost efficiency, skills, and stability are. As we have noted, the future of cloud-native depends on platform-led models where automation, observability, and business alignment converge.
Komodor’s findings reinforce this trajectory. Expect enterprises to double down on:
- Policy-as-code and GitOps to enforce consistency.
- Rightsizing and autoscaling (event-driven, GPU-aware) to curb overspend.
- Platform engineering to unify developer experience.
- AIOps and AI-native workflows to manage the scale they’ve created.
For developers, success will depend less on whether you run Kubernetes, and more on how you operationalize it at scale with AI in the loop.
Equinix Launches Distributed AI Infrastructure for the Agentic Era
Twilio Doubles Down on Platform Unity to Bridge Innovation and Customer Readiness
Akuity Brings AI-Driven Remediation to Kubernetes Operations
Pure Storage Unites AI, Cyber Resilience, and Sustainability in Platform Overhaul
Infinidat Upgrades Enterprise Storage Efficiency with InfiniBox G4
