Why Germany’s Microsoft Exit Might Fail
I live here, I want it to work, but I’m not sure it will
I’ve written about Germany and Denmark ditching Microsoft. About the EU cloud sovereignty movement. About schools replacing Office with LibreOffice.
I’m rooting for all of it.
But there’s a version of this where it doesn’t work.
Munich already tried this
In 2004, Munich switched its entire city administration to Linux.
LiMux, they called it. 15,000 computers. Years of planning. A lot of political will. It was the biggest open source government migration at the time, and the whole open source community was watching.
In 2017, Munich switched back to Windows.
The reasons were a mix of things. Compatibility problems with documents from other agencies still using Microsoft. Software that only ran on Windows. Complaints from employees who found the tools unfamiliar. A new mayor who was less committed to the project. Microsoft moving their European headquarters to Munich around the same time.
Probably some lobbying. Definitely some inertia.
Thirteen years of work, undone.
Schleswig-Holstein is now doing something similar. So is Denmark. France has been doing it. Everyone points to these as examples of what’s possible.
The compatibility problem is real
LibreOffice is a good product. It handles most documents fine.
But when you send a .docx file to a municipality still on Microsoft Office, or receive one from them, things can go wrong. Fonts shift. Formatting breaks. Tables move.
It’s not catastrophic, but it’s daily friction that makes people frustrated, and frustrated people start asking why they can’t just use the “normal” thing.
That takes a while to overcome. Years even.
Germany’s federal government still runs on Microsoft in most places. So does most of the private sector. So do most other European governments.
A state switching to LibreOffice doesn’t work in isolation. Not ideally.
It constantly exchanges documents with institutions that haven’t switched. Every one of those exchanges is a potential compatibility moment, and enough of those moments can erode the political will to continue with that.
The training cost is underestimated
Migrating software is not just migrating software. It’s migrating people. And people hate change.
It’s telling 30,000 people to change how they do their job. People who learned on one tool, who have workflows built around it, who will now need to find where things are in a completely different interface.
And more often than not, people who are not really tech-savvy.
Some of those people will adapt eventually. Some will struggle a long time. Some will complain, and their complaints might even get political.
The projected savings with open source software look good in the spreadsheet.
The training costs and productivity loss during transition are harder to quantify and easier to underestimate. Munich can probably tell a few stories about that.
Elections change things
Schleswig-Holstein’s migration plan has political backing… right now.
What happens after the next state election if a different party takes over? What if Microsoft lobbies hard, offers a better deal, points to Munich, and finds a receptive audience?
It’s happened before. It will happen again.
Long-term infrastructure decisions made by governments are only as durable as the political environment they were made in.
A three-year migration plan spans election cycles. That’s a vulnerability.
Microsoft isn’t standing still
While other European governments are planning their exits, Microsoft is building sovereign cloud infrastructure across Europe.
They don’t want to lose a humongous market. Obviously.
Data centers in Germany, in France, in the Netherlands. “EU Data Boundary” commitments. Products specifically designed to address the CLOUD Act concerns.
It’s not the perfect solution. An American company is still an American company under American law, but it’s a credible enough response to slow some of the migration momentum.
Some governments will look at the Microsoft sovereign cloud offering, compare it to the cost and disruption of a full migration, and decide the sovereign cloud is good enough.
Not all of them. But some.
I still think the direction is great
Germany’s Microsoft exit is NOT a mistake. Definitely. It’s years too late. Long overdue.
The underlying reasons are real and concerning. Data sovereignty matters. The CLOUD Act is a real problem. Reducing dependency on a single American vendor is sensible long-term policy. The open source tools have genuinely gotten better. And more abundant.
Still, wanting something to succeed and it actually succeeding are different things.
Munich wanted it to succeed too. Fortunately, this was a long time ago.
The difference this time, hopefully, is that the political pressure is broader, the tools are better, and the EU coordination is stronger.
Germany, France, Italy, and the Netherlands building shared open source infrastructure together is harder to reverse than one city doing it alone.
Hopefully, that’s enough.