France has ordered its government ministries to migrate from Windows to Linux. The directive, issued by DINUM on 8 April 2026, requires every ministry and public operator to submit a formal exit roadmap by autumn 2026. DINUM itself will migrate first, serving as the proof of concept before the broader rollout begins. This is not a pilot. It is not a strategy document about a strategy document. It is a line in the sand. This is what sovereignty looks like when it stops being a conference topic and becomes policy.
Meanwhile, the Netherlands is still exploring options, running pilots, and writing strategy documents about strategy documents. France moves. We don’t.
The Illusion of Choice in Enterprise#
For decades, the enterprise environment has lived as a hostage to the Microsoft stack. Not because it is always the best choice, but because it is the easiest one to defend in a meeting. No one ever got fired for buying Microsoft. Entire governments got locked in because of it.
The pricing trap is structural: foreign vendors set the bill, and you do not negotiate, you absorb. France’s Gendarmerie Nationale demonstrated what walking away actually looks like. Running Linux on over 100,000 workstations since 2008, the force records roughly two million euros in annual licensing savings and a 40% reduction in total cost of ownership. That is not a projection, it is a measured outcome across nearly two decades of operation.
Beyond cost, there is a harder problem. Telemetry you cannot verify is not just metadata. It is dependency. If you cannot audit it, you do not control it, and if you do not control it, someone else does. One vendor, one ecosystem, one geopolitical decision away from operational paralysis. That is not architecture. That is dependency dressed up as convenience.
The Admin’s Scream: Technical Reality vs. Management Inertia#
Every engineer has had this conversation. It always ends the same way.
“Interesting point. But we’re going with the safe option.”
Safe means familiar. Familiar means vendor-controlled. VBA macros from 2009 are still dictating 2026 infrastructure decisions. Engineers flag systemic risk and management hears inconvenience. Open source is still dismissed as amateur hour by people whose entire production environment runs on it.
This is not a knowledge gap. It is willful avoidance.
The Dutch Paradox: Endless Pilots, Zero Commitment#
The Netherlands does not lack expertise. It lacks decisions. We pilot. We evaluate. We iterate. We never commit.
In April 2026, Amsterdam, Ede, ’s-Hertogenbosch, and Zaanstad launched the DAWO pilot: a Linux-based workplace environment that meaningfully challenges Microsoft 365. That is good. The question is what happens next. MijnBureau is assembling a European productivity stack from French, German, and Dutch components. The blueprint exists. So why is it still classified as an experiment?
BNR has already said out loud what most IT departments know quietly: reliance on US cloud infrastructure is not just an IT choice, it is economic exposure. At some point, “pilot” becomes an excuse. A comfortable place where nothing has to change.
The Swiss and French Roadmap: Policy, Not PowerPoints#
France and Switzerland proved that the problem was never technical. It was political.
Switzerland’s EMBAG law mandates open source first across the federal government. Not a recommendation, not a preference: a legal requirement. The cantons of Bern and Vaud followed with their own mandates at regional level. France’s DINUM did not ask ministries nicely. It set a deadline and told them to produce exit plans. The scope goes beyond desktops: collaboration tools, antivirus software, AI platforms, databases, virtualisation, and network equipment all fall under the mandate. Germany’s state of Schleswig-Holstein completed nearly 80% of its 30,000-workstation migration by early 2026 and recorded 15 million euros in licensing savings in that year alone.
The difference between these countries and the Netherlands is not better engineers. It is decisions.
La Suite Numérique: Replacing, Not Just Exiting#
France is not just leaving. It is replacing.
La Suite Numérique, developed and hosted entirely on ANSSI-certified SecNumCloud infrastructure operated by Outscale, a Dassault Systèmes subsidiary, already has 40,000 regular users across departments. Visio replaces Teams and Zoom for 2.5 million civil servants by 2027. Tchap handles secure messaging without WhatsApp or Slack touching a government conversation. The national health data platform is migrating off Azure entirely before the end of 2026, explicitly to eliminate exposure to extraterritorial laws like the US CLOUD Act.
This is what sovereignty looks like in practice: controlled, auditable, replaceable.
The Hard Truth#
Control is the currency. If you cannot inspect it, you do not own it. If you do not own it, you depend on it, and dependency compounds over time until the exit cost becomes the argument against leaving.
Migration hurts. Dependency hurts more, just slower and less visibly. The Gendarmerie has been running Linux for eighteen years. Schleswig-Holstein migrated 44,000 inboxes and saved fifteen million euros in a single year. These are not theoretical outcomes from a vendor whitepaper. They are documented results from governments that made a decision and held to it.
Open standards flip the power dynamic. When you require interoperability, vendors adapt or they get replaced. That is how you negotiate from strength instead of absorbing whatever the renewal cycle brings.
France pulled the trigger. Switzerland backed it with law. Germany is executing. We are still writing memos about whether pulling the trigger is feasible.
The technology is here. The expertise is here. The examples are here. What is missing is the decision.
Stop piloting. Start cutting. Stop renting. Start owning.
