Europe Offers €10 Billion to Build Seven AI Gigafactories

At least 100,000 advanced chips at each site, seven proposed locations and €30 billion to be raised from public and private capital. These are not the specifications of an ordinary data centre. They are Europe's attempt to move technological sovereignty from policy documents to physical sites that need electricity, water, chips and contractors.

On Thursday, the European Union opened applications for companies to build seven AI gigafactories. According to an Associated Press report, the EU and its member states are offering €10 billion in public funding, hoping the package will attract another €20 billion in private investment. The aim is to create facilities capable of training and running advanced models at a scale that Europe's current infrastructure cannot provide.

The significant figure here is not just the money, but the capacity it buys.

The plan calls for each factory to be equipped with at least 100,000 advanced AI chips. The European Commission says these facilities will be about four times more powerful than the data centres currently operating within the EU. Europe now has a network of 19 AI-focused centres stretching from Finland to Spain, but adding the seven factories would more than double its current computing capacity.

That difference matters because building an advanced model does not depend on the algorithm alone. It requires continuous access to computing, data and energy, followed by the operational capacity that allows researchers and companies to use the model after training. When foreign providers control these layers, legal and political autonomy becomes less effective, however robust the regulations may be.

European sovereignty begins with a simple question: who can run the model when policy changes?

This is the clearest lens through which to view the story. The EU is not only trying to catch up with US and Chinese companies in the model race. It also wants to reduce its reliance on hyperscale cloud providers. The report cites a Commission assessment that the five largest cloud service providers in the bloc are US companies, and that this dependence could expose sensitive uses to risks from access by other countries or service disruption.

But the plan reveals a contradiction that the word “sovereignty” cannot conceal. Europe does not currently manufacture many of the components these facilities require, while electricity there may cost two or three times as much as in the United States and China. Even if the EU controls the sites, contracts and access rules, it will remain dependent in the near term on advanced chips and supply chains whose critical parts lie beyond its borders.

Construction is possible. Ending dependence altogether is not a near-term promise.

There is also a local bill. Large data centres consume substantial amounts of electricity and water for cooling, and 40 mayors around the world signed an agreement in June to limit their impact on natural resources, energy prices and climate goals. The tender's success will therefore be measured not only by chip counts, but by whether the selected sites can provide energy and water without shifting the cost to residents and industry.

For the Middle East and North Africa, the European plan is not a ready-made model to copy, but it does offer a useful benchmark. The region is investing in data centres and computing capacity, yet announcing the number of chips or the size of a site is not enough to establish sovereignty. The harder questions are: who owns the infrastructure? Who guarantees the energy supply? Who receives priority access? And can local universities and companies turn computing power into exportable models, products and knowledge?

The tender may also create room for partnerships in energy, finance and supply chains with countries close to Europe, but that possibility requires actual contracts before it can be treated as a regional gain. The more durable lesson is that computing has become industrial policy, not a separate technical item.

The honest conclusion is that Europe has bought itself an opportunity, not a guarantee.

If the €10 billion in public funding draws €20 billion in private capital, and the sites are built on time with sustainable capacity, European companies and researchers will have a stronger alternative within their own borders. But if energy, financing or chip supplies falter, the factories may remain a bold headline over an old dependence. What has begun is a test of execution. The result will be measured by the computing capacity actually available and by who can access it, not by the value of the announcement alone.