The Cloud Cost Nobody Can Explain — And Why It’s a Decision Problem, Not a Technology One
Article contents
The Cloud Bill Nobody Can Explain
Cloud cost management should be straightforward — yet the March invoice arrives from the cloud provider, gets forwarded to IT, IT explains it, and the following month the same thing happens. Somewhere along the way, someone approved a migration, someone else left a test environment running, a team overestimated volumes during contract negotiations — and no one has yet worked out what that 4.200€ line item on row 23 actually represents. This is not a technical failure. It is not a budgeting failure either. It is a failure of decisions that were never formally made — and of costs that land on the desks of people who never made them.
Cloud platforms are transparent about numbers. They are not transparent about the reasoning behind them. The Flexera 2026 State of the Cloud report estimates that 29% of global IaaS and PaaS cloud spend is wasted each year — a figure that has edged upward for the first time in five years. The finding will not surprise anyone who works in the field. What is surprising is that the same patterns recur year after year, with the same explanations and the same unresolved root causes.
Why Cloud Costs Keep Escaping Control
The underlying mechanism is structural. Procurement negotiates rates, discounts, committed use agreements and contract terms — it optimises what it can see. IT teams and developers consume instances, storage, data transfer and ancillary services — they work against operational requirements, not budget lines. Between the two, there is no one connecting the two worlds in real time. The result is that the end-of-month cloud invoice surprises no one formally — yet it is almost always different from what was expected.
The issue is rarely the absolute level of spend. It is that no one knows what next month will cost. And often, no one can clearly explain what has already been paid for. This gap sits at the core of most cloud cost problems, and it manifests in three recurring patterns.
Peripheral Costs Nobody Estimated
The problem is rarely the headline services — most teams have a reasonable handle on those. The issue lies in ancillary costs: outbound data transfer, test environments running around the clock, forgotten snapshots, resources deployed in non-optimal regions. These items do not appear in contracts because they are variable. They do not appear in budgets because no one estimated them in sufficient detail. They appear on invoices, where no one has the mandate to challenge them.
Assumptions That Were Never Written Down
Every financial evaluation in ICT procurement rests on assumptions: how many users will actually use the platform, at what frequency, at what intensity. These assumptions are rarely made explicit. They live in the heads of whoever built the business case, in presentation slides, in spreadsheets that no one will ever update. When something does not add up — and it almost always does not — there is no reference point to return to. The only document that remains is the signed contract.
Vendor Projections Taken as Neutral Input
Usage estimates provided by cloud vendors are not neutral forecasts. They are produced by parties with a commercial interest in closing the deal, built on aggregated market data that may bear little resemblance to the buyer’s specific context. The vendor arrives at the table with numbers. The buyer arrives with impressions. Treating vendor projections as objective input data — without internal verification — is one of the most consistently underestimated risks in cloud procurement evaluations.
ICT procurement is structured to evaluate at the moment of purchase. It is not structured to learn during execution. This gap — between the decision and the verification — is where the majority of unattributable waste accumulates.
FinOps Is Not About Spending Less. It Is About Not Being Surprised.
This is the most common misunderstanding. FinOps is introduced with the declared goal of reducing cloud costs. Six months later, costs have not fallen significantly — and whoever sponsored the initiative is left justifying it to senior management. The problem was not the method. It was the framing from the outset.
The real value of FinOps is not cost reduction — it is legibility. An organisation that can attribute every line of cloud spend to a specific project, team or decision is an organisation that can make informed choices. The Flexera 2026 data confirms this shift: among more mature organisations, 64% now measure cloud success by value delivered to business units — up 12 percentage points year on year. Cost efficiency and savings metrics fell by 6 points over the same period. The question is no longer “how much have we cut?” but “do we know where we are going?”
Optimising without first building legibility means cutting in the dark. Teams eliminate resources that appear redundant, only to discover they were feeding something undocumented. The cost of recovery frequently exceeds the saving achieved — and internal confidence in the FinOps programme rarely recovers.
Those who build legibility first and optimise second do so with real data. Those who invert the order discover the problem when it is already too late to correct course cleanly.
What Organisations That Manage Cloud Well Actually Do
They do not wait for the renewal to review the numbers. They monitor spend by service, by team and by monthly trend — and they bring that data to the negotiating table before the vendor has already worked out what they want to confirm. Visibility is built continuously, not retrieved at contract review time.
They do not treat vendor projections as a neutral starting point. They build a minimum set of explicit internal assumptions — written down, with an owner and a review date — and compare them against real usage data between six and twelve months after contract start. Not to be infallible, but to have an independent reference point when something does not add up.
They do not keep procurement and IT separate. Even a minimal process that connects the two functions on shared, real-time data — rather than only at invoicing — changes the nature of the conversation entirely. The contract is not the endpoint of the evaluation. It is the moment at which correction becomes impossible. That is precisely why legibility must be built beforehand — not to avoid spending, but to always know why you are spending what you are spending.
The Real Cost of Cloud Is Never Just a Number on an Invoice
Cloud spend is the sum of decisions someone made, assumptions no one wrote down, and conversations between IT and finance that have not happened yet. Organisations that recognise this stop treating cloud cost management as a technical or financial problem — and start treating it as a governance problem. That shift in framing is, in most cases, what separates those who are perpetually surprised by their cloud bill from those who are not.
Further Reading
- Who Pays for the Cloud Spend Nobody Monitors
- Cloud Spend Predictability: The Real Challenge Behind FinOps
- ICT Procurement: When Assumptions Cost More Than Prices
- Cloud: The Real Problem Is Not Cost. It Is Predictability.
- Flexera 2026 State of the Cloud Report
Why does cloud spend almost always exceed the original budget?
In most cases, the negotiated price is not the problem — the starting assumptions are. Usage volumes, frequency of consumption and ancillary costs such as outbound data transfer are routinely underestimated. These assumptions are rarely made explicit and are almost never verified against actual data during contract execution.
What is unmonitored cloud spend and how does it accumulate?
It is the portion of cloud costs that no one has a formal mandate to track in real time. It forms in the gap between those who negotiate contracts — procurement — and those who consume resources — IT teams and developers. Test environments running around the clock, forgotten snapshots, resources deployed in non-optimal regions: items that do not appear in contracts but accumulate steadily on invoices.
Does FinOps actually reduce cloud costs?
Not by the time of renewal — at that point the contract is already running and the available levers are limited. The useful window is between six and twelve months after contract start, when real usage data is available but the vendor has not yet consolidated its renewal position. That is when informed renegotiation becomes genuinely possible.
Share this article