Experts skeptical

Can the EU still catch up with AI?

At the most recent AI summit in Paris, EU Commission President Ursula von der Leyen announced her intention to mobilize 200 billion euros for the development of AI technology.

(Bild: mariiaplo (KI-generiert))

By building huge data centers, the European Union wants to have a say in the development of artificial intelligence in the future. However, experts doubt that the domestic economy will be able to catch up with the USA and China in terms of technology.

"Even if we were to build large data centers in Europe, and even if we were to train a model on this infrastructure, what would we do with it when it's finished?" asks expert Bertin Martens from the think tank Bruegel.

It's the typical chicken-and-egg problem: on the one hand, these "AI gigafactories" could help companies such as Mistral from France or Aleph Alpha from Germany to develop AI models that meet the high European security and data protection standards compared to the USA and China.

At the same time, the absence of large and financially strong cloud providers such as Amazon Web Services or Google, as well as globally successful AI providers such as OpenAI, makes the construction of European data centers a risky undertaking.

Draghi report and "Stargate" project
The planned expansion of the AI infrastructure is a reaction to the report by the former head of the European Central Bank, Mario Draghi, on the competitiveness of the European Union. In it, he called for additional annual investments of up to 800 million euros. However, the plans are also a response to the "Stargate" project, under which the USA wants to pump 500 billion dollars into the construction of new AI data centers.

At the most recent AI summit in Paris, EU Commission President Ursula von der Leyen announced her intention to mobilize 200 billion euros for the development of this technology. She described the initiative as a "public-private partnership that will enable all our scientists and companies - not just the biggest ones - to develop the most advanced very large models needed to make Europe an AI continent."

According to von der Leyen, "AI gigafactories" would contain computers with a total of 100,000 of the latest high-performance processors. That would be more than four times as many chips as the largest European supercomputer to date, "Jupiter", which is currently being built at Forschungszentrum Jülich. The AI chips alone would cost several billion, as they cost tens of thousands of dollars each.

Compared to the USA, however, this is peanuts: Facebook parent company Meta alone wants to build a data center with 1.3 million AI processors.

Hurdles on the way to AI gigafactories
However, data center operators on both sides of the Atlantic are facing the same two main problems, emphasizes expert Kevin Restivo from CBRE, a consultancy firm specializing in this sector. These are the availability of AI chips in sufficient quantities and the enormous energy requirements of the servers. In the USA, the latter is giving nuclear power a renaissance.

AI expert Martens from the think tank Bruegel does not believe in financing data centers with public funds as a matter of principle. "The life expectancy of such factories is around one and a half years." After that, the processors are technically obsolete and need to be replaced with newer models.

In any case, since the triumphant advance of DeepSeek, the question has arisen as to whether Europe should rather invest the money in the development of software instead of pumping billions into the expansion of infrastructure. The AI from the Chinese start-up of the same name makes do with less computing power than its Western competitors while offering comparable performance.