The European Cloud/Computing Situation

A brief addition to the 50000 words I wrote earlier on the cloud: what is the European situation?

Software

Initially, companies and governments would buy licenses to software. You’d typically have a piece of software in your office, on one of your computers, to calculate payroll with. Most other computers would have copies of WordPerfect installed. This software would function for years without updates or maintenance. If WordPerfect-the-company would disappear, you would not even notice.


Here, have some free nostalgia

Software As a Service

This “software you own” model had several problems. For one, software companies did not make that much money. Why buy a new license if you already have WordPerfect 5.1 installed? People also did not really want to rent software on a subscription basis.

However, for quite some software, it also became possible to use that remotely. So you’d no longer run a payroll calculation program yourself, you’d log in to some website, where you fill out details, and they do the calculation for you.

For the provider this is pretty nice, you have to pay for this stuff on a monthly or yearly basis. And you also can’t stop paying, since you can no longer do payroll then.

This software-as-a-service (“SaaS”) model also made it possible to do away with a lot of local IT knowledge. And in fact, many company departments would sidestep their own computer people completely, and do business directly with SaaS providers, a practice known as “shadow IT”.

Such SaaS providers are sometimes European, but often aren’t.

Software As a Service on a hyperscaler

Over time, most SaaS-providers started to base their software not on their own servers, but on services rented almost exclusively from the big three US cloud providers: Amazon, Microsoft/Azure, Google. This is not just “renting servers”, in these situations, the hyperscaler cloud provider runs advanced services for you. This hyperscaler provider then becomes an integral and very hard to displace part of your operations.

Europe does have providers of servers, compute capacity, storage, networking, and these are in fact very good and affordable. But no one here in Europe offers the advanced services out of the box.

As a practical Dutch example, municipalities first managed the appointments with citizens themselves, then outsourced that to a provider, often Swedish QMatic, and these then outsourced their operations to Amazon Web Services. And the upshot is that most Dutch people who want to renew their passport now have to file their details on a US controlled server first. Which is what we want to get away from.

Hyperscaler saturation

Note that above there was an intermediate situation where people did run software on their own, non-hyperscaler, servers. This window has now mostly closed. It is the norm now to develop software that can only run on a specific US hyperscaler cloud, and not on individual servers, nor on the capable European hardware providers.

Doing anything else is considered odd, and you spend a lot of your time trying to explain people why you are being so difficult. This is now ingrained so deeply that governments seriously ponder putting their whole population register or their elections on non-European clouds “because what else can we do”. Which is odd, since there are still a number of very successful places who CAN do different things.

However, this means that even software that is not provided “as a service” is now almost always locked into a US cloud. Deviating from this norm is considered very weird, and in practical terms may no longer even be considered feasible. It is hard to go against the grain.

This is compounded by education systems only teaching Amazon, Microsoft or Google cloud technologies. These vendors also hand out large vouchers to students so they can launch their own startups based on exclusively US cloud platforms.

It is very impressive. A whole generation has forgotten that you can run software on servers instead of on clouds. Even though this can be an order of magnitude cheaper.

Change?

From the above, it is clear that if we want to change anything, there are two ways. We could try to get “as a service” providers that have relearned the skill of hosting on non-US clouds. With upcoming sovereignty rules, this might prove to be a viable business model.

Or we get European cloud providers that look sufficiently like US cloud providers that we could plausibly force operators to migrate to these. Currently, no single European compute provider looks to have the scale to make this happen individually. But by cooperating, we might be able to get to European Cloud Modules, which could enable even smaller players to uplift themselves into credible providers of advanced cloud services.

Once these exist, we could shame governments and public utilities into using those European providers. We can’t yet do that since they can plausibly claim the gap is too big. But we should be able to convince people that you want to run your own government communications, population registry and tax systems on computers you control.

Some things are however incredibly hard to change, like organizations deeply plugged into MS-365 technologies. These require years and years of work to move their systems to technologies that as yet do not cover every feature that Microsoft managed to cram into their systems over the past 30 years.

Summarising

Most organizations are now locked in to Software as a Service Providers, and these SaaS-companies in turn are deeply locked in to exclusively US cloud providers.

The way back is to build/get/stimulate SaaS-providers that can run on what we do offer in Europe. This is feasible, they only have to want it (and get business that way).

And meanwhile, we should try to cooperate and get European providers that look sufficiently like American ones that we can shame at least governments into migrating to such European clouds.

Further reading

A comprehensive summary of my earlier writing can be found in the Cloud Overview.

Of specific note: