Operations | Monitoring | ITSM | DevOps | Cloud

In FinOps, rightsizing means adjusting cloud resources (instance types, number of CPUs, amount of memory, storage, databases, containers, and many other configuration parameters) to match actual workload requirements. It’s one of the most powerful levers in the FinOps toolbox, and for good reason. Consider: Average CPU utilization across Kubernetes clusters sits at just 10%, according to Cast AI’s 2025 Kubernetes Cost Benchmark Report.

You wouldn’t buy an industrial dumpster for your kitchen, and you don’t need an EC2 T3 to do an M5d’s job.

AI is reshaping the cost landscape. As a positive person, I’m going to call this change exciting! FinOps teams are integrating AI into cloud platforms and incurring the spend that comes with it. As a FinOps strategist who has helped several companies optimize cloud spend across industries, it became evident that clarity around AI spend unlocks swift, smart decisions. That’s AI … optimized.

AI adoption continues to explode, and so do their costs. By mid-2025, enterprise LLM spend had already hit $8.4 billion, more than double the year before. And in a major shift, Anthropic recently overtook OpenAI as the enterprise leader. Their Claude models are now core tools for companies adding generative AI technology into their products and workflows. CloudZero recently announced we are the first cloud cost platform to integrate with Anthropic.

The popular narrative around AI economics is changing. At one time, Moore’s Law conditioned us to expect that smarter, faster computing would steadily get cheaper. When it comes to AI, that expectation holds true at the unit level. Per-token costs are indeed declining. But the number of tokens consumed per task is growing exponentially, making total costs spike. The tension here is important: on paper, inference is getting cheaper.

Managing containerized applications or microservices can be difficult. It is even more demanding and prone to error if you do it manually. So, what’s the alternative? Container orchestration. Container orchestration is an automation technology that enables engineers to coordinate when containers start and stop, schedule and execute tasks, manage failovers, and perform recovery processes. The technology helps automate these tasks throughout a container’s lifecycle.

Snowflake’s scalable architecture, minimal latency, advanced analytics, simplified data handling, flexible pay-as-you-go model, and always-on security make the data cloud a top choice for many businesses. You can also purchase Snowflake resources on demand or upfront. But if you struggle to control your Snowflake costs, you’re not alone. With the help of this guide, you’ll know how to manage your Snowflake costs better.

Kubernetes, also known as K8s, is an open-source, portable, and scalable container orchestration platform. With K8s, you can reliably manage distributed systems for your applications, enabling declarative configuration and automatic deployment. Yet, K8s can be resource-intensive and costly, with a rather steep learning curve. But in 2019, a lighter, faster, and potentially more cost-effective alternative appeared: K3s. Still, K3s is not a magic wand that works for all Kubernetes deployments.

If you’re struggling to contain cloud costs in this suddenly volatile AI-fixated environment, it might be time to consider FinOps as an exercise in granular allocation and unit economics, with a focus on outcome.

Cloud cost management is a big deal right now. For instance, 58% of organizations say their cloud costs are too high, according to our State of Cloud Cost report. Over the last five years, several other studies have shown that controlling cloud spend is a top cloud computing challenge. There’s more. As AI adoption accelerates, a new challenge has emerged: managing the rapidly growing costs of AI in a scalable and intelligent way.