From scattered infrastructure to a single EU cloud

Ask an IT lead what the infrastructure costs per month and you rarely get a single number. You get a rack on the premises, an environment at one hyperscaler, a second at another because a project once asked for it, and a row of SaaS subscriptions taken out department by department. Each piece has its own contract, its own billing date and its own administrator. The sum of it all is written down nowhere.

That is the real problem with scattered infrastructure. Not that one environment is too expensive, but that there is no single place where the cost comes together. Without that overview you cannot steer. You see the invoices arrive separately, you pay them, and the run-rate creeps up year after year without anyone being able to point to the moment it became too much.

Where the money leaks away

Scattering costs in four places at once. Duplicate contracts, because two teams buy the same database or the same monitoring separately. Overcapacity, because each environment is provisioned for peak load on its own and those peaks rarely coincide. Unused licences, because nobody has the full picture. And the cost of leaving somewhere, which kept you in place for a long time.

That first point is no marginal phenomenon. In the Flexera 2025 State of the Cloud, 84% of organisations name managing cloud costs their biggest cloud challenge, and the average budget overrun comes to 17%. Spread that same spend across multiple environments and that figure grows rather than shrinks, because every platform has its own pricing logic and its own blind spots.

The waste also sits in what you already pay for but do not use. Flexera reports in the 2024 State of ITAM that organisations waste 20 to 30% of their IT spend and that more than half of all SaaS licences go unused. In one place that stands out. Spread across a rack, two clouds and a handful of separate subscriptions, it slips through the cracks, because there is no list on which an unused licence sits next to an active one.

Why leaving was expensive for a long time, and no longer is

Scattered infrastructure usually stays put for one reason: the cost of leaving. Pulling data out of a cloud cost money per gigabyte, the so-called egress fees, and that turned every migration into a bill you preferred to postpone. So an environment you had long wanted to leave stayed in place, purely because leaving looked more expensive than staying.

That changed in 2024. Google Cloud dropped the egress fees for those who leave in January, AWS followed on 5 March 2024, both partly under pressure from the European Data Act. That regulation, (EU) 2023/2854, sets out switching obligations, functional equivalence and interoperability, and phases out switching and egress fees. It applies from 12 September 2025. The cost that held you in place for so long is disappearing.

With that, the last argument for leaving everything where it stands falls away. Until recently, consolidating was an expensive exercise with an uncertain payback. Now the bill for leaving is small enough to make the exercise worthwhile. The only question is where you begin, and that is with what you actually spend today. An IT FinOps audit lays the full run-rate bare, contract by contract, and puts the duplicate and unused items on a single list.

A single EU cloud on an open stack

What you put in its place decides whether the gain holds. Bringing five environments together on a sixth you do not control only moves the problem. The way out is a single platform you own, on an open stack, run on an EU cloud. One environment, one invoice, one place where cost and consumption come together.

On an open stack, with Kubernetes, PostgreSQL and the rest in standard components, you are not locked in to any vendor who sets the price unilaterally. The code sits in your own repository, the data stays within the EU, and NIS2-level management is built in rather than bought on top. What was scattered becomes one thing you can read and steer, and move if you have to.

The gain from that exercise is real, but not automatic. McKinsey reported in Cloud's trillion-dollar prize is up for grabs (2021) that by 2030 more than 1 trillion dollars in run-rate EBITDA is up for grabs across the Fortune 500, with an average EBITDA rise of more than 20%. That number only lands for those who clean up first and then build. Anyone who lifts the existing sprawl one to one onto a new cloud moves the mess and keeps the cost.

How you get there without a big bang

Consolidation need not be a weekend in which everything moves at once and you hope it runs on Monday. At YK it happens in waves, with a go or no-go at each step. The existing systems keep running until each wave is stable, a parallel transition with no cutover in a single night. If a moved piece works as it should, the next one follows. If it does not, you stop before the cost mounts.

So put every server, every cloud and every subscription on one sheet, with the cost and the consumption alongside, before you even think about a platform. Once that sheet is there, you see where the duplicate contracts sit, which licences nobody opens any more and which environment runs on a fraction of its capacity. From there, consolidating to a single EU cloud becomes a decision you keep in your own hands, with a run-rate that falls instead of creeping up.