Sovereignty can wait. Competitiveness can't

Why European enterprises should adopt the best AI today, and build sovereignty alongside: guest blogpost by Rad Parvin, co-founder and senior value partner, Valliance.

The debate about technology sovereignty is real and it matters. The recent Fable shut-off, when access to a frontier model was severed at the policy level, is exactly why people are right to take it seriously. Acting on that debate today, though, is a luxury most European enterprises cannot afford.

In boardrooms across the continent, a familiar scene is playing out. A CIO hears a pitch for an enterprise operating system that could reshape how the company runs. One data architecture spanning the silos, agentic workflows running across functions, AI models finally given the business context to be accurate and useful. The CIO is convinced. So are the stakeholders.

Then someone asks the question that stops the room. Do we want to be associated with this provider, on sovereignty grounds or reputational ones?

That tension, between the business value AI creates and the values of the company providing it, is becoming the dilemma that blocks enterprise adoption. Whether the provider is OpenAI, Google, Microsoft or even Palantir, it is fair to ask about the principles guiding the people building the world’s most powerful AI. However, waiting for a clean answer carries a real cost.

It’s difficult to buy into the current geopolitical discourse, and it pains me to write that sovereignty matters more now than it did – we simply cannot fully count on those we once relied on. That shift is both real, and uncomfortable. Even so, the competitive disadvantage of waiting is the consideration I keep coming back to.

We weren’t early enough then. We can’t afford to be late now.

Europe’s most consequential technology decisions were made a generation ago. As cloud computing and large-scale platforms emerged from Silicon Valley, local businesses chose to integrate rather than own. That path cannot be undone, and trying to reverse it with AI carries serious drawbacks.

What we are seeing now is a costly overcorrection. Policymakers and compliance teams are fixed on where technology is built, who controls it, and whose values it reflects. These are valid concerns. Yet when they become the primary lens for every decision, the result is friction, and friction is what erodes competitiveness.

Regulatory uncertainty is already raising the cost of building here. When OpenAI paused Stargate UK, it pointed to energy costs and an unsettled rule-making environment. Ambition was never the issue. The UK government’s own figures show AI adoption across business remains limited, and even the best-funded players hesitate when conditions stay unpredictable.

At the same time, competitors abroad are moving quickly. They are deploying new capabilities without the internal constraints that slow teams in Europe. If we spend the coming years building local capabilities just to reach parity, we will spend them watching those competitors extend their lead.

What leaders are really weighing

In practice, enterprises are making a larger trade-off than they tend to admit. They are weighing the business value AI generates against the corporate values of the firms providing it.

Stripping away the political noise to assess technology providers on what they can actually deliver, and the conclusion is usually the same. The answer to their problem exists, it just won’t come from Europe yet. The values question is thus worth asking, but the stakes with AI are simply too high to let it stall action that needs to happen now.

European businesses need to sit with an uncomfortable truth. Getting near-term value from AI will mean working with firms owned and based outside the continent. Even Europe’s own enterprise software giants reflect this. SAP customers are facing one of the most complex platform migrations in the industry as they move to S/4Hana, and SAP has endorsed Palantir as a validated path for the most complicated of those migrations. That is a notable signal from a company with every incentive to keep the work inside its own ecosystem.

European alternatives might be coming, but not in time

There is a strong case for building sovereign alternatives, and that work is already underway through investment and partnership. The reality is that a true competitor to the US AI labs is two to three years away at best. That will not cut it for enterprises that expect measurable results in half the time.

So European organisations face an intimidating choice. Wait for sovereignty, or compete with the tools available today, wherever they were built. Those who wait risk more than delay. They fall further behind competitors who are already building AI capability, developing organisational muscle memory, and embedding data-driven ways of working.

The fundamentals have not changed. Putting the best technology against the right use cases is still the surest route to value. The noise around sovereignty has simply grown loud enough to make leaders doubt the approach that has always worked.

Progress first, sovereignty alongside

The biggest success stories will come from businesses that accept today’s constraints and put strong governance around them, so they can work with the most capable tools available.

Adopt the most effective technology for the problem in front of you, and adopt it on your terms. Contract for exit and portability. Keep ownership of your data and the models built on top of it. Build the internal capability to walk away if you need to. The provider may sit outside Europe, but dependency does not have to be permanent.

Sovereignty should not be a precondition for progress. It is something to build alongside, at the level of policy and over time. The enterprises that understand this difference, and act on it, will be the ones still competitive when Europe’s sovereign capabilities finally arrive.