Komodor has launched Capacity Intelligence and Predictive Placement for its AI SRE platform, aimed at cutting Kubernetes cloud infrastructure costs by addressing wasted cluster capacity.
The new features target a problem many engineering teams face after using workload rightsizing tools and node autoscalers to make initial savings. Komodor argues those methods are largely reactive and can leave deeper inefficiencies in place when workloads, scheduler decisions and reliability rules prevent further consolidation.
At the centre of the launch is what Komodor describes as stranded capacity inside Kubernetes clusters. This includes resources tied up by Pod Disruption Budgets, anti-affinity rules, unevictable workloads and nodes that cannot be terminated, all of which can prevent clusters from shrinking even when spare capacity exists.
Capacity Intelligence is designed to detect these issues across Kubernetes environments and identify the configuration problems blocking node consolidation. The feature provides root cause analysis, financial impact estimates and remediation options, with checks intended to protect production systems.
Predictive Placement is designed to act before waste appears in the cluster. Komodor says the feature uses AI-based simulations to assess drain scenarios, identify consolidation candidates and steer workloads away from nodes likely to be drained or removed.
It can also place unevictable workloads on designated nodes so autoscalers have more flexibility elsewhere in the cluster. The aim is to reduce node growth caused by scheduling decisions that leave workloads on infrastructure that should otherwise be drained.
Reactive limits
The announcement reflects a broader effort across cloud operations teams to move beyond basic cost-control tools and address the structural causes of waste in container environments. Kubernetes clusters often become more expensive over time as teams add rules and safeguards to maintain reliability, but those same controls can limit the ability of autoscaling systems to consolidate infrastructure.
Komodor says more than 30% of cluster capacity is typically stranded by optimisation blockers, misconfigurations and autoscaler limitations. It is positioning the new functions as a continuous loop that detects inefficiencies, diagnoses the cause, applies fixes and helps prevent the same waste from returning.
The new tools are integrated into Komodor's wider AI SRE platform, which it markets to site reliability engineering teams managing cloud-native applications. Komodor says its Klaudia agentic AI technology evaluates optimisation recommendations against reliability concerns so cost reductions do not introduce instability or performance issues.
Komodor operates where reliability engineering and financial control increasingly overlap. Companies using Kubernetes are under pressure to reduce cloud spending without weakening application resilience, creating demand for tools that can assess infrastructure efficiency and operational risk at the same time.
The new cost-optimisation features are available immediately within the platform. Komodor has also raised USD $90 million in venture funding and counts Fortune 500 businesses in sectors including financial services, healthcare and retail among its users.
"Traditional cloud infrastructure cost optimization is reactive, causing it to miss significant savings opportunities," said Itiel Shwartz, co-founder and CTO of Komodor.
Shwartz said Komodor's broader view of cluster operations helps address the problem. "Because Komodor's AI SRE has complete awareness of both workload behavior and cluster state, it can prevent structural inefficiencies before they occur and continuously optimize pod placement to maximize cluster utilization. This context-aware approach finally allows teams to eliminate structural waste without risking reliability," he said.