Europe's Cloud Problem Isn't the Tech. It's the Mindset.

Tue, 16 Jun 2026 00:00:00 GMT

Everyone agrees Europe finally has the cloud it needs to quit the Americans. OVH, Scaleway, a wave of sovereign providers selling data residency and independence. The pitch is simple: leave AWS, go local, sleep better.

At Mergify, we use Scaleway for small things, never for core infrastructure. That part is deliberate: we already knew they weren't up to the promise. Last Thursday morning, they delivered a new dedicated server. It ran for about an hour, then it died. It hasn't come back since.

It's Monday now. Here is the latest update from their support team:

Read it again. The machine has been down since Thursday. The plan, four days in, is to try replacing the cables. No timeline. No replacement. No credit. Just "thank you for your patience" and "we remain at your disposal," which is a remarkable thing to write to someone whose server you are currently not fixing.

And to be clear, this isn't a free-tier toy. It's a real bare-metal server we ordered and paid for.

It's not the outage. It's the reflex.

Hardware dies. It dies at AWS, it dies at Hetzner, it dies in my garage. Nobody serious judges a provider on whether a disk fails. You judge them on what happens in the next hour.

That's where the gap shows. The default reflex here wasn't "let's get this customer running again." It was "let's investigate, on our schedule, and keep him politely informed." A box that dies within an hour of delivery isn't a support-tier question, it's a defect, and the fix for a defect is a replacement, not a four-day investigation. We told them three times: the machine was freshly installed, there was no data on it, we were perfectly happy to wipe it and start over on a fresh one. It changed nothing. There's no "here's a new machine while we look into the dead one." That option isn't in the playbook. Somebody decided the dead unit would be repaired, so we wait, because that's the procedure. And no, this isn't a margin problem: a spare box for a day costs Scaleway almost nothing. Nobody even offered.

We're in tech. We ship multiple times a day. Our entire job is removing the friction between an idea and production. And here we are, in 2026, sitting on a support ticket like it's 2001, losing days of CI capacity because someone is going to try the cables. (We got lucky: this box runs part of our CI, not our core. If it were core infra, we'd be dead in the water.)

It's tempting to make this about Scaleway, or to make it about Europe versus the US. It's neither. AWS support can be scripted and slow too. One dead server isn't a study.

And it isn't even just about servers. The same week, another European provider changed something on their side, broke our integration, and never told us. When we dug in, their shiny new flow dead-ended in a 404. We reported it, walked them through it, showed them their own process was broken at the last step. No "thanks, we'll fix it." Just "you should have known." End of conversation.

You don't build a company on that reflex. You certainly don't build a startup on it. And yet, here, it's the norm.

A 95-year-old furniture company gets it

Here's what makes me sure it's culture and not geography, or taxes, or scale.

Two years ago I bought a garden table from Lafuma, a French outdoor furniture brand founded in 1930. The table arrived perfect, except it was two centimeters too tall. I wasn't going to return it. I just sent a note saying it felt like a design oversight.

A week later, a QA engineer from Lafuma called me. He listened, then said: "I'll send you custom legs, two centimeters shorter. You'll have them next week." And he did. (I told the whole story here.)

Think about that for a second. A nearly-hundred-year-old company that makes camping chairs and garden tables, with no API, no status page, no SRE on call, has a better service reflex than a cloud company that sells itself as innovative French tech. The furniture maker trimmed custom legs for a customer who didn't even ask. The cloud company can't swap a server that died on day one.

Same country. Opposite instinct. So it's not France, and it's not Europe. It's not the tax code, it's not margins, and it's not scale (Lafuma is not exactly a hyperscaler). It's a choice. It's not that Europe can't do service: a 95-year-old furniture company just proved it can. It's that European tech decided it doesn't have to. One culture optimizes for keeping you whole. The other optimizes for being right about the contract.

What you're actually buying

Here's the part that matters if you're weighing a move to a European cloud for sovereignty in 2026.

The technology is ready. The data centers are real, the hardware is fine, the prices are competitive. That was never the gap. The gap is the same "playing not to lose" instinct I wrote about with French tech: do the minimum, follow the procedure, don't get blamed.

What you're also buying, whether or not it's written in the SLA, is a default reaction for when things break. And the default reaction we keep hitting is "you paid, now wait." That's the thing that doesn't scale and doesn't ship.

The frustrating part is that the playbook isn't a secret. It exists here, at home, in an unglamorous 95-year-old company that ships garden furniture. Geography isn't the excuse. European tech imported the product ambition. It forgot to import the customer obsession that's supposed to come with it.

Build the table. Ship the legs. Or at the very least, replace the server.

europe — jd:/dev/blog

Europe's Cloud Problem Isn't the Tech. It's the Mindset.

It's not the outage. It's the reflex.

A 95-year-old furniture company gets it

What you're actually buying