I've said it before, and I'll say it again. I *really* don't want European Hyperscalers to be a thing. Hyperscalers are at best a market/regulation failure and at worst a huge risk to create another one of these "too big to fail" scenarios.

I want a sane and healthy landscape of small- to midsize competitors in the hosting landscape.

It makes everything more robust and reliable.

Not sure why this is not a shared understanding anymore.

"but Jan, how should we compete in the AI space without hyperscalers?", you ask.

Maybe by not betting on a technology like LLMs that need "stupid scale" to deliver questionable benefits and instead focus on useful things that can be trained and used on hardware that don't require betting your entire economy on it?

But what do I know.

@halfbyte Yes please — and all built around open standards and interoperability! We can dream, right?

@halfbyte I usually bring up "Your product doesn't need a hyper scaler" in various settings, and then explain how extremely few products exist with such bursty requirements. When you're not one of them, you're paying for potential capabilities you will never ever use.

That works, at the C-level.

@halfbyte

As we see with the Fediverse, one of the cornerstones of decentralization is the ability to move between different service providers not losing your resources (followers list).

We need the similar kind of open protocols and the federation of the IT service providers.

Replacing American vendor lock with European vendor lock is not enough.

@bookwar @halfbyte yeah, this is where I am torn. I see way to many directions and options which could hinder our ability to move freely because there is not enough compatibility between the providers.

On the other hand, I wonder what size would be appreciate for European providers to be able to keep up with demand without having capacity issues.

I think even Azure would need a public capacity meter like the parking lots have, become I ran into not having enough capacity to provision resources already too many times.

If Azure has capacity issues then it's not hard to imagine that European providers would have that too. And it would be bad for everyone when half way into a project the capacity would become a blocker.

If I would have a magic wand, I'd wish for making all datacenter technologies to be open and accessible for the benefit of humanity and not for the benefits of the shareholders.

@halfbyte I'm with you and while I also believe deep in their hearts enterprise IT teams do want "robust and reliable" providers, I also have good reason to believe that C-level incentive to "avoid onboarding another provider with procurement and whatnot" is much stronger.

If you can't provide *everything* you will eventually be replaced with someone who does.

@halfbyte I blame the whole cloud/containerisation thing. And everyone wanting to be Google, therefore starting with infra way too large for them. They cannot even do back-of-the-envelope calculations any more.

@halfbyte Because everyone believes *they* would be the one benefiting from the concentration of capital and wealth, so they don't want diversified and distributed and federated systems.

Everyone believes cancerous growth is the thing to strive for.

Why? Indoctrination.