On June 12, 2026, Gaia-X welcomed the European Commission's proposed Cloud and AI Development Act. But the real story sits in two numbers: 82 percent of German companies want to end their dependence on U.S. cloud providers — and 78 percent still depend on them anyway. Between wanting and doing lies more than a political statement.
June 12, 2026, and an Uncomfortable Difference
On June 12, 2026, Gaia-X — the Brussels-based European federated data-infrastructure association — welcomed the European Commission's proposed Cloud and AI Development Act (CADA) as an important step toward European digital sovereignty. That's good news. But it sits next to two numbers that, taken together, describe the actual situation.
82 percent of German companies want to end their technical dependence on U.S. cloud providers. In practice, 78 percent still are dependent. This is not a disagreement between two camps. It's the same camp wanting one thing and doing another. The four-point spread isn't the success — it's the measure of a gap between intent and reality that has stayed almost unchanged for years.
That gap is the story. Not the announcement, not the bill, not the catalogue. The question is: why does a company fail to do what it explicitly wants?
Klingt interessant?
Wanting Is Cheap, Switching Is Expensive
The answer is unromantic. Wanting sovereignty costs nothing. Achieving it costs a migration project, a re-architecture, and the risk that the product stalls mid-rebuild.
Most mid-market stacks aren't built to be provider-agnostic. Over years, they've been hard-wired to a single hyperscaler's proprietary services — managed databases, identity services, queues, AI endpoints. Each of those has its own API, its own billing model, its own operational logic. Switching to a sovereign provider is therefore rarely a configuration toggle. It's a quarter-long project nobody prioritizes while the system is running fine.
Then there's the second problem: for a long time there simply were no serious alternatives you could move to without losing functionality. Companies that wanted out couldn't find a door. And that is precisely what the June events address.
The Supply Side: Policy and Infrastructure Are Catching Up
Two movements are closing the supply gap right now.
First, policy. The Commission's CADA proposal aims to deliberately build out European cloud and AI capacity — a regulatory answer to a market still dominated almost entirely from outside Europe. The backdrop: global spending on sovereign-cloud IaaS reaches roughly 80 billion US dollars in 2026, with Europe alone accounting for about 12.6 billion euros. This is no longer a niche. It's a market that is being invested in.
Second, infrastructure. Gaia-X is releasing its first multi-provider catalogue with around 600 services from 15 providers, targeting 1,000 services by year-end. That's more than a directory. It's an attempt to make sovereign options comparable, discoverable, and consumable — solving exactly the problem that had been stalling German companies: there was nothing concrete to switch to.
Both are right and overdue. But both address the supply side. They build doors. They don't walk through them.
What This Means for CTOs and Tech Leads
The decisive point: a growing catalogue of sovereign services helps a company only if its architecture can actually consume those services without being rebuilt. Three consequences I think are worth taking seriously.
First: sovereignty is an architecture question, not a procurement question. You can pick the best sovereign provider in the catalogue — but if the application hard-codes knowledge of the old endpoint, the switch is still a rebuild project. The gap between 82 and 78 percent doesn't close through more supply; it closes through the ability to switch supply.
Second: switching cost is the real lock-in. Nobody stays with a U.S. hyperscaler out of conviction. They stay because moving is expensive. Lowering the cost of switching — through provider-agnostic abstraction instead of hard coupling — changes the math long before you actually migrate. Optionality has value, even unused.
Third: sovereignty is not a one-time project. Providers, prices, and the regulatory landscape change. An architecture that makes switching a configuration decision survives those changes. One that makes every switch an emergency stays structurally dependent — no matter how sovereign the mission statement sounds.
This Is Exactly Where nopex Comes In
The June events solve the supply side. nopex solves the demand side — the gap between wanting and doing at the application layer.
nopex combines agent-based, fully automated software development with infrastructure on European data centers and provider-agnostic routing. The application logic never knows which vendor is behind it. Whether a service comes from the Gaia-X catalogue or somewhere else is a configuration decision, not a re-platforming. That lets a German mid-market company move from the 78-percent-dependent column to the sovereign one without rebuilding its product.
That's the whole point: sovereignty shouldn't be a strategic decision plus a migration project — it should be a property of the stack. Policy builds the doors, Gaia-X catalogues them. nopex makes sure your architecture can actually walk through — today, and the next time the landscape shifts again.
