Should you care about FinOps?

By: Paola Annis
(Author, Hello FinOps)

I started working with public cloud in 2017, I was a Cloud Solution Architect for a hyper- scaler vendor. We pushed for the cloud as a flexible and quick and modern solution to host applications and workloads. As soon as companies started moving their virtual machines and apps, they realized they had a cost issue: a cost no longer a CAPEX but OPEX and a fluid, ever-changing one, with direct financial choices in the hands of computer engineers like never before.

This prompted a new way of handling technical architecture: one that took into account the cost of every single choice in how our applications are deployed, maintained and updated.

On the other hand, while people started using the cloud as this "infinite resource", datacenters were spreading through the globe, with their environmental and social footprint. The AI gold rush exacerbated this, and the datacenters multiplied until some cities started to put regulations to stop datacenters from taking the land.

Do you see a problem here?

Just as back in the 1920 the consumerist movement started (years ago I read it all in the book "Booze, Babe, and the Little Black Dress: How Innovators of the Roaring 20s Created the Consumer Revolution" by Jason Voiovich) and led us to deplete the planet's resources in a few decades, we are now entering a phase of digital consumerism revolution, where our digital activity is quickly devouring the planet, one AI-generated kitten action video at a time.

So, if you are doing cloud and do not care about your own wallet, you should at least care about the planet and how wasting cloud resources leads to building new datacenters and wasting water and electricity...for no kittens at all!

This is my second book on the topic. The first, published in 2022, was called "The Road to Azure Cost Governance" and was the result of a year and a half of struggles building what is now called a FinOps practice for a specific organization. This time I am focusing on multi-cloud operations, for companies using AWS, GCP and Azure (but the suggestions and best practices can apply to other hyper-scalers as well) and willing to put the effort of building with efficiency.

Is it worth doing FinOps? Surely you could just go and negotiate another 10% discount with your cloud vendor. The bottom line is the same, the effect is what we call hiding the dust under the rug. You won't see the problem until it's too late and the effort to start optimizing will be so heavy that you will have to go back and re-negotiate.

“Hallo, FinOps” is organized by sections where you’ll learn the basics of the framework, the operations that you will need to enforce, and how to get proper measurements and KPIs/OKRs to keep the practice in good shape.

My favorite one is the Clean-up chapter: I think I mentioned this recently in a public event, asking the audience: how many photos do you keep in your phone? Are they all relevant and important? And this sparked a conversation on the digital waste of storage, which today is mostly in the cloud. Getting rid of unused resources has a cathartic effect, much as Marie-Kondoing your cloud, keeping only what you are effectively using.

I also added an entire chapter, plus the Annex, to a list of things to check periodically when optimizing, because in my experience there is a certain paralysis when deciding where to start saving and optimizing, and with this list you will be covering the minimum indispensable actions to keep your operations efficient and lean.

In the book there is obviously a chapter dedicated to AI, with the dual angle of using AI to help with cost governance, and cost-governing your AI costs. Using AI in FinOps moves the practice from reactive cost management to proactive optimization by analyzing large volumes of cloud usage and billing data to identify trends, predict future spending, and detect anomalies early. Machine learning models can forecast costs based on historical usage patterns, enabling teams to plan commitments such as reservations or savings plans more accurately and avoid unexpected spikes. AI can also support automated optimization by analyzing telemetry from workloads and recommending actions such as rightsizing resources, adjusting autoscaling policies, or identifying idle infrastructure, up to on-demand scaling of resources. Combining AI insights with normalized multi-cloud cost data, integrating real-time operational metrics from tools such as Kubernetes cost monitors, and validating recommendations through governance processes your engineering teams remain accountable for cost efficiency and have full control over their cloud spend.

So, my suggestion is to start now, put together a team, give them freedom to experiment and the right tools and KPIs to succeed: this book can be your guiding light. You won't regret it and both your wallet, and the planet will be vastly improved by the exercise.