Why Governments Can't Quit Microsoft: Munich, Paris, Kiel, and Denmark's Quiet Retreat

Denmark's digitalisation ministry quietly killed its Collabora pilot this month, and the tech press filed it under embarrassment. But one ministry's abandoned pilot is just the newest chapter of a story Europe has been living for more than twenty years — a story with a famous failure, a quiet success almost nobody cites, and one ongoing attempt that might finally be doing it right.

Munich: the original sin

In 2004, the city of Munich launched LiMux, a project to replace Windows on its desktops with its own Debian-based (later Ubuntu-based) Linux distribution. By 2012 the city had migrated 12,600 of its 15,500 desktops. Munich said the project saved it €11.7 million and made it ISO-certified for usability — a genuine, completed, working Linux city hall.

In November 2017, the city council voted to undo it and move 18,500 workstations back to Windows by 2020, citing compatibility. The reversal is usually retold as "Linux failed," but that is not what happened. The migration itself had succeeded; what failed was the political maintenance behind it. Elections changed the council, departments wanted less friction with the state's standard software ecosystem, and keeping LiMux required continuous effort that nobody was re-elected for. By May 2020, Munich's newly elected politicians were saying they preferred free software for future projects — after having already paid to go back.

The lesson Munich taught was cynical but durable: a migration is not an event you survive once. It is a position you have to keep defending, and politics stops defending anything after the ribbon is cut.

Paris: the success nobody quotes

While Munich's retreat made global headlines, France's National Gendarmerie — a police force of tens of thousands — quietly pulled off one of the largest desktop migrations in history.

It started in 2004-2005 with the unglamorous move: replacing 90,000 copies of Microsoft Office with OpenOffice.org and making the open ODF format the nationwide standard. When Microsoft ended Windows XP's life, the Gendarmerie faced paying to retrain staff for Vista, and instead built GendBuntu, its own Ubuntu variant, first released in 2008. The goal was 80,000 migrated computers by the end of 2014, with 35,000 already deployed by late 2011, and estimated savings around €2 million per year. The project is still alive — GendBuntu 24.04 LTS shipped in December 2024.

Two details explain why it worked where Munich failed. The Gendarmerie went office-suite first and desktop second, so the hard document-format problems were solved before the operating system changed underneath everyone. And it is a military-style organisation: no elections, no council majority to flip, one budget holder for two decades.

Kiel: the blueprint, honestly audited

The current exemplar is Schleswig-Holstein, the northern German state that announced in 2024 it would leave Microsoft entirely, framing it as digital sovereignty with a hard deadline: Microsoft contracts run only until 2029.

The verified progress is real. By December 2025, LibreOffice was on roughly 80% of the state's workplaces outside the tax administration, with Microsoft Office and Outlook being uninstalled across departments. By mid-2026 the state reported moving more than 40,000 accounts — and over 100 million emails and calendar entries — off Exchange and Outlook onto Open-Xchange and Thunderbird, projecting more than €15 million a year in licence savings against a one-off cost of about €9 million.

But the part everyone headlines is the part that hasn't happened. A response filed in the state parliament this year put the actual Linux desktop rollout at up to 250 workstations by mid-2026 and 500 more by year's end — pilot scale, against a destination of roughly 30,000 administrative PCs (a fleet the state now sometimes describes as 60,000). The state did the smart thing: it dismantled the Microsoft dependencies around the desktop first — office, mail, collaboration, telephony — and left Windows itself, the hardest and most visible layer, for last.

Why exits fail

Line up the four attempts and the failure modes sort themselves out:

• The .docx moat. Denmark's pilot drowned in formatting — blank pages, broken tracked changes — because document exchange with the Microsoft-format outside world is the actual hard problem, not features. This is what our office-suite alternatives guide keeps returning to: fidelity of the round-trip, not checklists of features.
• Adjacent software. Munich's council cited compatibility; the ecosystem around a government — SAP, land registry tools, departmental add-ons — is written assuming Windows.
• Political time versus migration time. A serious exit takes five to ten years. Elections arrive every four. Denmark's pilot did not even survive to the next ballot.
• The announcement-to-deployment gap. Headlines report targets (30,000 PCs!), audits count machines (750). Every migration's true progress is only ever the audited number.

The verdict

Quitting Microsoft is not impossible — a French police force did it at a scale Denmark's ministry never attempted, and it is still running two decades later. What the record actually says is narrower and more useful: the exits that survive start with formats and mail, anchor themselves to contract deadlines rather than press releases, and outlive the governments that announced them. Denmark had none of those anchors. Schleswig-Holstein, so far, has all of them. By 2029 we will know which pattern Europe copied.