Request a free demo

FinOps and Observability: How to Reduce Enterprise Cloud Costs

FinOps and Observability: How to Reduce Enterprise Cloud Costs - FinOps

According to Forrester’s forecast, the global public cloud market may reach $1.03 trillion in 2026. This is nearly double what was spent five years ago.

Previous editions of Flexera’s State of the Cloud Report have consistently put this waste in the 28–32% range, and based on recent trends, a similar share is expected for 2026. Cost forecasting for AI workloads is becoming increasingly complex, driven mainly by the volatility of GPU capacity, autoscaling, and spot pricing.

Most enterprises use cloud services in some form, though many organizations still operate hybrid or on-premise setups as well. What matters is that cloud costs remain transparent. Observability can be a valuable complementary tool here, particularly when paired with FinOps data.

What Makes Up the Wasted 29%? 

The 29% loss doesn’t stem from a single source. It results from a handful of recurring patterns that appear in nearly every organization running cloud infrastructure.

The largest line item is idle, unused capacity: a virtual machine, database, or storage volume that someone spun up for a project and simply forgot to shut down once the project wrapped. The systems keep running and billing while doing effectively no work. This category accounts for roughly a third of the total waste.

Close behind is oversizing: actively used resources provisioned with far more capacity than actually required. This typically happens when a team provisions extra capacity “to be safe” at launch, then never revisits the decision, even when actual load has sat at a fraction of that capacity for months.

Finally, there are orphaned resources: storage volumes or snapshots left behind after the system they belonged to was decommissioned, continuing to accrue charges unnoticed.

It’s worth noting that each of these items originates from a reasonable decision made at a given point in time. The situation simply changed afterward, and no one caught it in time.

Why Does This Waste Accumulate?

The real problem isn’t cost reporting. Most organizations can see exactly how much they spent on cloud in a given month. What’s missing is the practice of someone continuously reviewing that consumption.

It’s not unlike a subscription you never got around to canceling. At the moment it was activated, it was a reasonable decision: necessary, useful, worth the cost. Then the situation changed and the need disappeared, but the monthly charge keeps going through, because there’s never a moment when someone comes back to ask whether it’s still needed.

In an enterprise cloud environment, the exact same thing happens, just at greater scale and in a far less obvious form. A test environment, an oversized server, a reserved capacity block: each is its own “subscription” that someone activated and no one ever canceled.

Why Doesn’t Every Cloud Cost Get Reviewed?

Because it doesn’t hurt. A needlessly running server doesn’t cause an outage, doesn’t trigger an alert, and doesn’t disrupt anyone’s daily work. It bills quietly in the background. On top of that, ownership for decommissioning is rarely clear. Whoever spun it up has long since moved on to the next task, and whoever reviews the invoice has no visibility into what that line item is actually running.

Connecting FinOps and Observability

This is the “silent signal” that connecting FinOps with observability addresses, and the two leading vendors in the market are moving precisely in this direction.

Datadog’s Cloud Cost Management product brings engineering and FinOps perspectives together on a single platform. By linking cost and performance data, engineering teams no longer encounter the bill in a separate spreadsheet; they see it in the same system they already work in. An anomaly, as a result, doesn’t just signal that something has deviated from the norm. It also signals what that deviation is costing.

Dynatrace has moved in a similar direction with its Kubernetes monitoring: energy consumption and carbon footprint are measured from the same data the system already collects on performance. Cost is no longer a separate view – it’s simply another dimension of existing observability.

The underlying principle is the same in both approaches. Review doesn’t happen when someone sets aside dedicated time for it. It happens continuously and automatically, within the same system where all other operational data already lives. This solves exactly the problem described above: silent waste doesn’t go unnoticed because of bad intent, but because no one is continuously watching. Observability makes that continuous watching possible without requiring human intervention.

How Does Observability Reduce Cloud Costs?

The mechanism can essentially be described in three steps, each built on the same principle: making visible what wasn’t visible before.

First: the resource is linked to its owner. The observability platform can map resource usage, provided the organization applies consistent tagging and maintains its metadata. This eliminates the situation where a line item on the bill sits unowned. There’s always context: who provisioned it, what it belongs to, and when it was last actually used.

Second: utilization is continuously visible, not just the fact that something is running. Observability doesn’t just measure whether a server is running – it measures how well-utilized it is. A resource that’s actively running but poorly utilized becomes just as visible as a faulty system – flagged automatically, without anyone needing to spend time hunting for it.

Third: alerting doesn’t wait for someone to look. Rather than review being a task written into a calendar and easily skipped, the system can send alerts for low utilization, provided the organization has configured the relevant rules and thresholds. This fills exactly the accountability gap described earlier. No dedicated person is needed to watch for it, because the system catches it on its own.

Together, these three elements mean cloud cost is no longer discovered after the fact during a quarterly review. It stays visible continuously, in real time, within the same system where performance and reliability data already live.

Closing Thoughts

Cloud cost growth is driven partly by consumption itself, and partly by rising prices for certain services and the spread of new, resource-intensive technologies. It grows because reviewing consumption was never built into day-to-day operations. This isn’t a matter of negligence. There’s simply no one assigned to the task until the system itself makes the waste visible.

This is the point where the CFO and the IT Ops leader finally see the same thing. Instead of piecing together the picture from two separate data sources and two separate vantage points, they work from one shared view: what’s running, why it’s running, and whether it’s still worth running.

At Telvice, we help our clients ensure that their existing observability investment doesn’t just serve system stability. It also drives continuous, automatic review of cloud costs. In many cases, the necessary data already exists, but cost optimization often also requires enabling new modules or FinOps features. In most cases, the necessary data already exists; it simply hasn’t been looked at from this angle yet.

Request an expert consultation.

Sources

Picture of Telvice Zrt.
Telvice Zrt.