Operations | Monitoring | ITSM | DevOps | Cloud

One of the most challenging aspects of running a SaaS business is dealing with highly variable costs. Every service you provide needs resources, and the demand for those resources fluctuates. Just when you think you’ve got this quarter’s budget nailed down, an engineer makes a restructuring decision or a new feature launch exceeds expectations (or falls flat). Suddenly, your bill looks totally different than what you were expecting.

AWS did not intend to build Amazon EKS; it simply had to. Kubernetes adoption beamed light years ahead of AWS’s own managed container orchestration service. This forced AWS to develop a managed service to accommodate customers who wanted to use upstream Kubernetes but did not want to manage it themselves. As soon as AWS got around to it, it knocked the Kubernetes-based container management service out of the park. Not only is Amazon EKS simpler than Kubernetes, but EKS pricing may also be worth it.

When I ask clients to picture cloud cost optimization, they think dashboards, policies, maybe a clever right-sizing purchase. What they don’t picture? Meetings. Misunderstandings. Mistrust. To avoid FinOps failures, we need a new starting line; one that gets to the root of spend misalignment.

In early October, we launched Dimension Studio, a new visual editor for engineers and others that brings point-and-click simplicity to the same powerful, precise allocation engine CloudZero is known for. Before that, when CloudZero users built cloud cost allocations, they got it from our YAML-based CostFormation engine, a code-driven way to describe how cloud and AI costs roll up to products, customers, or teams.

Kubernetes began as a tool to help teams keep thousands of microservices running without falling apart. It gave them a way to schedule workloads, recover from failures, and scale services without constant firefighting. Now, AI has brought back the same chaos, only magnified. Training jobs sprawl across GPUs. Inference traffic spikes without warning. Pipelines stretch across clusters, clouds, and compliance boundaries. Left unchecked, it can break both your workload and cloud budget.

The siren song of building a custom, internal cloud cost management platform is enticing. Many brilliant engineering teams are convinced they can come up with a bespoke solution that perfectly fits their needs. They look at their company’s unique infrastructure and decide they can DIY cost management without having to rely on an external vendor. Believe me, I get the temptation.

A global outage at Amazon Web Services (AWS) did more than knock popular apps offline. It laid bare the cost risks embedded in many cloud architectures. As services fail, the hidden costs of high availability, from redundancy planning to recovery operations, often multiply. For cloud cost leaders, this isn’t an issue of uptime; it’s a visibility and budget-shock issue. It’s a key reminder that architecting for resilience involves difficult trade-offs.

In 2018, attackers gained access to Tesla’s AWS cloud environment through an unprotected Kubernetes console (admin console). Because it lacked proper authentication, the hackers could see and control cluster resources. Once inside, they deployed new pods running cryptocurrency mining software, using Tesla’s compute power for profit. During the breach, the attackers also uncovered credentials stored in the cluster.

So you’ve run into cloud cost challenges — incomplete visibility, overspending, thinning margins — and you’re wondering whether you have what it takes to build a cloud cost optimization platform in-house. It’s a responsible question (one that the FinOps Foundation even advises you to ask). Be warned: Building your own cloud cost optimization platform seems cheaper than adding another SaaS subscription. It’s not.

When CloudZero’s CTO Erik Peterson joined the FinOps Weekly podcast in October 2025, he didn’t hold back. Instead of going on about the usual best practices of AI cost optimization, he posed challenges to how we approach AI spending. From “zombie AI experiments” eating your budget to why you should stop apologizing for using AI, these 10 insights from the podcast are worth considering in how we approach AI FinOps. (Watch the full podcast below and keep reading for more!)