The Cracks Are Showing in Big Tech's Foundations

Something significant happened in April 2026 that most people missed.

France’s government announced it is migrating 2.5 million workstations from Windows to Linux. Not a pilot program. Not a trial. An order to every government ministry to submit a formal migration plan by autumn, covering not just the operating system but everything — antivirus, collaboration tools, the lot.

French minister David Amiel was direct about the reasoning: France could no longer accept not having control over its data and digital infrastructure. The phrase they keep using is digital sovereignty — the idea that depending on foreign software for critical government infrastructure is a strategic risk.

They’re not alone. Germany has been moving state computers to Linux for years. The European Commission has been quietly reducing Microsoft dependency. It starts slowly, then all at once.

The Monopoly Was Never That Solid

It’s easy to forget how contingent big tech’s dominance is. Windows became the default because in the 1990s, switching costs were enormous. You needed Microsoft’s ecosystem: Office, Active Directory, certified hardware drivers, the whole stack. The cost of leaving was so high that most organisations never seriously evaluated the alternatives.

That calculus is changing.

Cloud-based collaboration tools mean your documents live in a browser regardless of OS. Linux has matured enormously — the desktop experience in 2026 is genuinely good. And the political climate has shifted: depending on American tech companies is now a boardroom-level concern, not just an IT department talking point.

The inertia was always the moat. Not the product.

AI Is the X Factor Nobody’s Talking About Enough

Here’s what doesn’t get discussed enough.

AI isn’t just making existing software better. It’s dramatically lowering the cost of building new software. A small team — or even a solo developer — can now build tools that would have required a 20-person engineering team five years ago.

For decades, organisations accepted that a product like Microsoft Office would do 80% of what they needed, and the remaining 20% wasn’t worth the cost of building something custom. Custom software was slow, expensive, and risky.

Now it isn’t.

When AI can generate working code from a plain English description, write tests, and handle all the boilerplate — the economics flip. A small software shop can build exactly what a specific client needs at a fraction of the old cost. The client gets something that fits their actual workflow, rather than forcing their workflow to fit the software.

That’s a fundamentally different world.

The Unbundling Has Started

Big tech’s business model has always depended on bundling. You don’t just buy Word — you get the whole Office suite. You don’t just get email — you get the entire Microsoft 365 stack, tightly integrated and expensive to exit.

But that lock-in is eroding. Not because the products got worse, but because the alternatives got genuinely good and the switching costs dropped.

We’re going to see a wave of smaller, specialised software companies build products for specific industries, specific workflows, specific needs. Not trying to be everything to everyone — just trying to be exactly right for someone.

The companies that will win aren’t the ones with the biggest platforms. They’ll be the ones who understand a particular domain — a specific type of law firm, a particular kind of healthcare provider, a niche manufacturing process — and build software that fits like a glove.

AI makes building those solutions tractable. That’s new.

What France’s Decision Really Means

France’s migration isn’t just about saving money on licensing (though that’s real). It isn’t just about geopolitics (though that’s real too). It’s a sign that the “default to Microsoft” assumption is no longer automatic, even at the highest levels of institutional inertia.

When governments start questioning the monopoly, the businesses that follow governments’ lead — and there are many — will start asking the same questions.

The future isn’t one platform to rule them all. It’s a lot of smaller, better-fit solutions, built by people who know their customers well. The tools to build them exist now. The motivation to look for alternatives is there. The old reasons to stay are getting weaker.

The cracks are showing. The interesting question is what gets built in the space that opens up.