EU accused of wasting €20B on AI computing dreams

BRUSSELS — A €20 billion European Union plan to build massive artificial intelligence computing hubs is drawing widespread criticism ahead of launch.

The plan to build very large computing hubs to train powerful AI models — outlined by European Commission President Ursula von der Leyen more than a year ago — will be announced this spring as Europe’s response to bold U.S. efforts to boost computing power.

But legislators and experts question whether there’s demand in Europe for the computing power these buildings will generate, and say the EU risks plowing €20 billion into an effort that will be unsuccessful in helping it compete with the U.S. and China.

With the bloc already lagging behind the U.S. and China due to a strong focus on regulating the technology, an ineffective computing power plan would set the bloc back further as big projects such as OpenAI’s massive $500 billion Stargate data center project run ahead.

“Nobody could explain to me what is the business case they are planning with these gigafactories,” German Greens European Parliament lawmaker Sergey Lagodinsky said at a recent event in Brussels.

“I talked to some who are saying: ‘we just need more compute in Europe.’ But then, when I ask them, ‘What for?’ They say: ‘It doesn’t matter, we just need more compute.’”

The European Commission said in a response that Europe needs computing power to avoid being too reliant on other continents. “This is not just about raw compute power, it’s about sovereign compute,” said European Commission spokesperson Thomas Regnier.

Stargate rival

The European Commission has been scrambling to get computing power to the bloc’s tech champions, startups and researchers in a bid to match the development of leading U.S. large language models (LLMs) such as OpenAI’s GPT and Anthropic’s Claude.

As an early pledge, the Commission in 2024 announced plans to build 19 AI-focused supercomputers across 16 countries, known as AI Factories. In February last year, von der Leyen outlined the next step: gigafactories.

The aim is to build four to five mega facilities, each powered by 100,000 graphics processing units (GPUs), a type of chip used to train AI models. They would be four times the size of the AI Factories and rival some of the world’s leading projects, such as OpenAI’s Stargate data center in Norway. 

The projects will be funded by combining public and private investment, and the EU has set up a €20 billion fund to back the plans from its side. A total of 76 bids to establish 60 sites across 16 countries were filed during an informal sense-check by the Commission last year, with established companies such as French Scaleway among them.

“This demonstrates an emerging substantial market demand,” the Commission’s Regnier said.

As an early pledge, the Commission in 2024 announced plans to build 19 AI-focused supercomputers across 16 countries. | Nicolas Economou/NurPhoto via Getty Images

Efforts are ongoing to merge and finalize some of those bids and to iron out the practical details between EU countries that support the bids and the vehicle running the EU’s network of supercomputers, called EuroHPC. A formal call for proposals has been delayed twice, and is now expected to be launched this spring.

The idea is clear: to build very large facilities to train very large AI models, which requires vast amounts of computing power.

But critics say it is unclear which companies will use the power from the mega facilities compared to smaller facilities. “It’s way less clear what the audience for the gigafactories is,” said Nicoleta Kyosovska, a research assistant at the Brussels-based think tank Centre for European Policy Studies, in an interview.

Kyosovska co-authored a report on gigafactories with the headline: “Sanctuaries of innovation, or cathedrals in the desert?”

“We don’t have that many AI companies, we only have Mistral,” she said.

Europe’s reliance

Besides the French AI front-runner, Europe has no AI companies capable of developing models that need this amount of computing power to train. But Mistral isn’t waiting for the AI gigafactories to launch: It’s building its own infrastructure. 

In February, the company announced an investment of €1.2 billion in data centers in Sweden. At the end of March, it raised $830 million to finance a data center with close to 14,000 GPUs near Paris.

That has fueled calls for the Commission not to focus on providing computing power to developers of large language models that can rival ChatGPT, but to double down on its industrial strengths. 

“We can never compete with the investments that were already done in the U.S.,” said the Bulgarian European Parliament lawmaker Eva Maydell at a panel in February. “There’s an entire world beyond LLMs, where our strategic advances as Europeans lie,” she said.

She argued that LLMs can draft text, but they are not optimized for tasks such as developing new battery technologies that the EU’s industrial base can benefit from.

She stopped short of saying that the gigafactories lacked the right focus, but urged the Commission to consider the “end state” of the project.

Mistral didn’t respond to a request for comment on whether it intends to use the gigafactories.

The headquarters of Nvidia — the chipmaking company that is the world’s leading supplier of GPU chips — is pictured in Santa Clara, California in August 2025. | Justin Sullivan/Getty Images

There are also concerns the gigafactories will increase Europe’s reliance on U.S. technology because of the dominance of Nvidia, the chipmaking company headquartered in Silicon Valley that is the world’s leading supplier of GPU chips.

A group of 18 European Parliament lawmakers warned the Commission that the data center space is “concentrated and dominated by a single supplier” and asked: “How does the Commission intend to use AI gigafactory initiatives to reduce Europe’s strategic dependencies?”

Regnier declined to respond directly to the concerns around Nvidia, but said that the gigafactories will boost European sovereignty by avoiding reliance on U.S. infrastructure.

“European industry and researchers are explicitly demanding sovereign environments where their strategic data and proprietary models are fully protected under European law, without the possibility of third country interference,” he said.

Others say the plan is too little, too late and the scale of the EU’s €20 billion investment is no match for what is being set up globally.

OpenAI launched a $500 billion compute power plan last year, with an initial focus on the U.S., while Anthropic also unveiled a $50 billion investment in infrastructure there.

“If you compare other parts of the world and the volumes that are being invested, they are on a totally different of magnitude,” said Jeff Campbell, chief government strategy officer at Californian tech giant Cisco at a panel in February. “Thinking about doubling might be a good idea.”