Free report·
The State of Enterprise Kubernetes Strategy 2026
Only 38% of organiSations running Kubernetes in production have reached enterprise-grade maturity.
This report examines the current state of Kubernetes strategy at the enterprise level, measuring how large organisations govern Kubernetes in production rather than how mature they consider themselves to be. Respondents answered nine questions about 13 practices across four domains: governance, software supply chain, resilience, and observability. Each respondent's organisation was scored from 0 to 100 and placed into one of three maturity tiers: Enterprise-grade, Established, and Standardising. The analysis begins by comparing Kubernetes standards with the rest of the IT estate and distributing enterprises across the three tiers. It then identifies what separates the most mature estates from the rest, including the organisational arrangements, breadth of adoption, challenges faced, and planned investment over the next 12 months. The report closes with AI and model serving, looking at how much of it runs on Kubernetes and what maturity means for how that traffic is governed. It is based on 445 responses from professionals at organisations with over 500 employees, fielded in August 2026.
Key findings
- Six in ten enterprises that run Kubernetes in production hold it to the same or a higher standard as the rest of their estate.
- One in four otherwise high-scoring organisations miss one of the prerequisites of having a mature strategy, most often observability.
- Where a dedicated platform engineering team runs Kubernetes day to day and a separate central security team sets the policy, enterprise-grade maturity is twice as likely. Yet, only one in five enterprises operate this way.
- Investment plans are concentrated where enterprises are already strongest: security and policy enforcement (42%) and automating cluster lifecycle and upgrades (37%). Enterprise-grade estates lead on both by around 20 percentage points.
- 77% of enterprises serving AI models do so on Kubernetes clusters they operate themselves. Fewer than half inspect the agent tool and MCP traffic moving through them.
Questions this report answers
What share of enterprises reach enterprise-grade Kubernetes maturity?
Only 38% of enterprises running Kubernetes in production have an enterprise-grade strategy. Half (51%) sit in the Established tier and the remaining 11% fall into the Standardising tier.
What organisational arrangement makes enterprise-grade Kubernetes maturity most likely?
Enterprises where a dedicated platform engineering team operates Kubernetes day to day while a separate central security team sets policy are twice as likely to reach enterprise-grade maturity (66% vs 32%), yet only one in five (19%) follow this arrangement.
Which Kubernetes domain most often holds enterprises back from the top tier?
Observability is the weakest domain, with 48% reaching the Enterprise-grade band and the highest share of any domain (20%) in the Standardising band. Governance is the strongest, with 67% reaching Enterprise-grade.
Where will enterprises focus their Kubernetes investment over the next 12 months?
Strengthening security and policy enforcement is the leading priority (42%), followed by automating cluster lifecycle and upgrades (37%), while improving observability and telemetry sits at 28%.
How much AI and model serving runs on enterprise Kubernetes clusters?
Amongst the enterprises serving AI models, 77% do so on Kubernetes clusters they operate themselves, rising to 86% for Enterprise-grade estates against 70% for those in the Established or Standardising bands.
Are enterprises securing their AI inference traffic?
Enterprises serving AI models have extended existing web and API protections to AI traffic (86%), but AI-specific controls lag: fewer than half (43%) apply inspection of agent tool / MCP traffic.
Who it's for
- Platform engineering leads benchmarking how their Kubernetes estate is governed
- Security and infrastructure decision-makers planning Kubernetes investment
- Technology leaders governing AI and model-serving workloads on Kubernetes