Skip to content 
No one sounded angry on the call, it was the finance director. The bad news is that was the worrisome part.
He just showed his screen, and said, “Hey, look at this cloud bill, it’s gone up by nearly 70% over the past nine months.The question I have heard more times than I care to count from cloud vendors is, “Hey, look at this cloud bill, it’s gone up by nearly 70% over the past nine months.
What is it that we’re paying for?
Silence.
Three Engineering Leads joined the call. Dashboards were accessed by a DevOps Manager. The procurement team said that it was the fault of architecture. Engineering blamed growth. Finance blamed forecasting.
The fact was that this was not very comfortable.
No one was able to see everything.
It wasn’t an uncommon enterprise. The 2025 Flexera State of the Cloud Report reveals that about a quarter of the organizations’ cloud spending is wasted due to over-provisioned resources, idle infrastructure, and lack of planning. On first sighting I was quite impressed by that number, saying to myself, ‘that seems like a lot of sites, doesn’t it?’ It’s been a few enterprise scenarios, I think it could even be conservative.
In this article, I am going to explain you how we could reduce cloud spend by 40% without impacting on the uptime, engineering teams and the operational chaos.
NO, we did not do it by accidentally shutting down and/or putting teams in a budget crisis.
By fixing habits is how we did it.
The expensive mistake most companies make first
Many teams will attempt to optimize their cloud resources as a crisis management situation. They begin getting rid of services. That usually backfires.
The painful lesson I had to learn was as part of a SaaS infrastructure project where the leadership insisted upon cutting compute costs, the team went too far and came back with a significant reduction in compute. The cost has decreased one month.
After that, there were lots of support tickets.
Latency increased. Customers complained. Due to the number of changes, half of them were reversed by the engineers. The business has ended up spending more money to remedy hasty decisions than it saved.
Cloud optimization is a failure when people think of cloud infrastructure as a spreadsheet problem, rather than a systems problem.
You Can Also Get More Info About Advance Cloud: AI Is Reshaping Cloud Infrastructure Trends Faster Than Expected 2026
Not only do you have to balance performance, reliability, developer productivity and cost, but you also have to do it all at once.
That balance matters.
Many organizations invest more money in the cloud than they need to, due to the fact that visibility and governance are lagging behind the pace of cloud adoption, which is why it has been a recurring theme in a number of Gartner reports. Teams can build up infrastructure quicker than financial accountability.
The outcome is no surprise. Resources multiply quietly. Bills grow loudly.
Before touching anything, we mapped the waste
The first thing we did was so very dull. We audited everything. Not exciting. Not technical. But necessary.
We shipped usage data for production, staging and testing environments, cloud storage, cloud networking, Kubernetes clusters, monitoring tools, databases, and third-party Saas: apps for cloud environments.
Surprisingly, leadership was surprised to see this:
- No one was sure what the workloads were that would be relevant.
- There was one environment for a product experiment that was dropped after 7 months.
- Another cluster was too large as traffic levels were expected to spike that did not happen.
- There was a fully functional database replication environment for some reason or other, but not really business-related.
- The largest cloud savings are typically as a result of awkward finds.
- Not clever engineering.
What we tracked during the audit
We focused on five categories:
- Idle compute instances
- Overprovisioned storage
- Underutilized Kubernetes clusters
- Duplicate monitoring and observability tools
- Data transfer and networking waste
The five areas were responsible for almost 80% of avoidable expenses. This put a stop to that. Cloud costs weren’t just a matter of a general gripe this time, leadership had an opportunity to identify specific waste.
Why rightsizing gave us the fastest wins
It would seem that it’s the renegotiation of contracts that would result in the greatest savings. Typically, it doesn’t. Our world was turned upside down with rightsizing.
This is why we found workloads running on over-sized virtual machines, just for the sake of having some “just in case” performance.
I know what it is all about. No one wants to get the blame for downtime. However, if you’re thinking of building “just in case” infrastructure it becomes costly very fast.
One analytics workload running on AWS was running its compute instances to serve its analysis needs around-the-clock, regardless of the level of demand. The actual utilization was less than 20% on average.
It made some difference to their monthly spending, as a result of that one change they decreased their spending by several thousand dollars per month. Scale it up to an enterprise and savings start to be a serious issue.
In one instance, we reduced compute usage by almost 35% after analysing the actual usage. Nothing broke. That surprised people.
Real World Example: Shopify’s cloud efficiency approach
Shopify has allowed for public discussion of its efforts to be more efficient during times of rampant growth, with the aim of minimizing inefficiencies in its infrastructure.
But what larger organizations know, is easy:
- Too much growth if it’s not price disciplined can be hazardous.
- Cloud spending adds up without a lot of fanfare, or awareness.
Then it’s about the finances.
A counterintuitive truth nobody likes hearing
The more cloud flexibility there is, the more waste there is. With cloud, people think that it will save them a fortune because you only pay for what you use. Technically true. Operationally misleading.
Saving money is only possible when somebody takes action to see to what is being saved. Otherwise, you’ve got a costly convenience machine on cloud.
I have seen many companies transition to the cloud, with the promise of huge efficiencies and then spend more afterwards.
Why?
Removing hardware constraints makes over consumption easier. There is no such thing as just, when talking about data centers.When it comes to data centers, there is no such thing as “just”. Minuteman’s in the cloud, someone can do it in minutes. The benefits of that convenience come at a cost.
Bad technology decisions are more likely to result in cloud waste. It’s a result of decisions made by us which we never notice, which are made each day.
That realization led to the way in which we do infrastructure management. No more “cleaning up cloud optimization. We used it as a sort of financial hygiene.
Reserved instances saved money, but only after we stopped guessing
Initially, the leadership would like to purchase the reserved capacity right away. This was a bad idea!
Reserved instances or savings plans only take advantage of workloads that are stable and predictable. However, if not, you are wasting money on bad spending habits.
We did a lot of analysis on workload consistency in almost 2 months. We made our commitment only when we could!
Reserved pricing significantly lowered infrastructure costs for predictable workloads, particularly for database, and core application services.
In certain locations, savings were from 30 to 50 percent. But timing mattered. It is common for companies to make bad decisions when they commit to buying things too early.Purchasing contracts early puts bad decisions in place for later.
What actually helped us cut cloud costs by 40%
The final reduction came from stacking smaller wins together. No single magic trick existed.
Here is what worked:
| Area | Approximate Savings |
|---|---|
| Rightsizing compute | 15% |
| Removing idle resources | 8% |
| Reserved instances | 7% |
| Storage optimization | 5% |
| Better engineering governance | 5% |
Small improvements compound.
That is the part many executives underestimate.
The week we almost ruined everything
This is important because they don’t typically write articles about errors when talking about cloud optimization.
We made one. A serious one. While cleaning up stored data, someone has set too aggressive rules for data lifecycle. The storage of archived workloads was quicker than anticipated. A reporting system was slow to pick up during the night.
Business teams were NOT pleased.
We bounced back from that but we learned the lesson that:
- Do not optimize costs to the cloud without taking into account the business need.
- Infrastructure that is cheap at first, but then leads to low productivity, is again costly.
- There is a need to make compromises.
Always.
What smart companies do differently
In general, organizations that are good stewards of cloud costs do not focus too much on discounts. They are committed to being held accountable. Teams are aware of owner of costs. Dashboards are visible. Engineering has an awareness of budget considerations. Finance is knowledgeable about technical limitations.
The FinOps Foundation says that when engineering and finance teams work together, companies with mature cloud financial operations (finops) are more likely to outperform other companies in cost predictability.
This Article Might Be Also Very Helpfull For You: Enterprise Cloud Spending Risks That Escalate Operating Expenses
That’s not something that people realize is important. Cost optimization of clouds is not technology of the clouds. A communication problem.
Why visibility matters more than discounts
J.R. Storment, Executive Director of the FinOps Foundation, has consistently argued that cloud financial management works best when organizations create shared accountability between engineering, operations, and finance.
He makes a point I strongly agree with:
- Teams cannot optimize costs they cannot see.
- You cannot fix invisible spending.
That sounds obvious. Yet many enterprises still lack workload-level visibility.
A practical step you can take this week
If you are responsible for the cloud infrastructure, do it before Friday! Go to cloud dashboard. Organize resources for use by utilization.
Then ask the following three questions that are relatively straightforward:
- Which of the following has not been touched for 90 days?
- What processes will always have a utilization of less than 30%?
- Who has the biggest spending categories of the teams?
Waste will likely be more quickly found than you think. Most organizations do. No, enterprise consultants won’t be needed to start. No reasons but curiosity and discipline.
Why How We Cut Cloud Costs by 40% had less to do with technology than people
In retrospect, it’s not been the most difficult technical alterations. Changing behavior was. Engineers are no longer able to “overbuild” to solve problems that might occur. It was time for “finance” to cease and desist from attempting to reduce spending without any technical understanding. It was clear leadership had to embrace a mindset of continuous cloud optimization, not a panic quarterly exercise.
The truth behind How We Cut Cloud Costs by 40% is that. We didn’t find any hidden gadget. We made small smart decisions, challenged assumptions and introduced visibility over and over.
When the cloud bill is just continuing to rise, it’s likely not because of migration discounts, or vendor pressure. It is knowledge as to what it is that you actually use. Being frank and honest about what you don’t.
Author Bio
Talha Qureshi is an enterprise technology analyst and blogger with over a decade of hands-on experience across cybersecurity, cloud infrastructure, B2B SaaS, and enterprise AI. He writes about the gap between how enterprise technology is marketed and how it actually performs in real organizational environments.
