Everyone says EU hosting is good. Few people say what it costs, in money and in friction, before the sales call ends.
We run everything in EU regions by default. Not as a line on a pitch deck. As a filter before we pick a vendor, a CDN edge, or a backup target. If data cannot stay in the EU without a special case, we do not start that special case lightly.
What it actually costs
EU-only hosting usually adds something like ten to fifteen percent on cloud bills compared with default US-East pricing. That number moves with the provider and the workload. Compute is rarely the painful part. Egress, managed databases, and "global by default" products that quietly fan out to other regions are.
The other cost is time. Data processing agreements. Subprocessor lists. Explaining to a US SaaS why Frankfurt is not optional. Procurement for EU clients gets simpler when you already have the answers. Security reviews get shorter when you are not inventing residency after the fact.
What we refuse by default
Analytics that mirror traffic to a US property. Auth that cannot pin sessions to an EU tenant. Object storage with a global endpoint and no clear region story. Those tools can be fine for other products. They are the wrong default for the work we do with public organisations and regulated teams.
Sometimes a client still needs a US edge for users there. Then we say so, document it, and keep the system of record in the EU. The point is not purity theatre. The point is not waking up to a residency problem in year two.
Why we stick with it
EU clients ask earlier every year. Not because a slogan told them to. Because their legal and IT teams got clearer about subprocessors. Building EU-first once is cheaper than refactoring a stack that assumed Virginia was home.
If you are pricing a project and EU hosting is still a checkbox at the end, move it to the start. The architecture changes. So does the vendor shortlist. So does the story you can tell in a DPIA without sweating.