Former European Central Bank President Mario Draghi said Europe needs to increase its artificial intelligence data-centre capacity to support economic sovereignty and productivity growth.
Writing in the Financial Times on Friday, Draghi said the productivity gap between the euro area and the US increased from $9 per hour in 2018 to $21 in 2025.
He cited European Central Bank scenarios indicating that rapid AI adoption could add between 0.3 and 0.4 percentage points annually to total factor productivity growth, which has been approximately zero since 2022.
According to figures cited by Draghi, the European Union accounts for less than 5% of global AI computing capacity, compared with 75% for the US. He said the gap between European demand and installed supply is currently approximately 3GW, equivalent to around a quarter of existing capacity, and is expected to reach 14GW by 2030.
“Being cut off from AI, once the economy runs on it, would be more like being cut off from the US financial system. The effects would be catastrophic,” Draghi said.
Draghi, who also served as Italy’s prime minister, said Europe controls a limited share of the AI value chain but has an opportunity to retain sovereignty over its data. European Commission estimates cited by Draghi put the value of Europe’s data economy at more than €800 billion, or over 5% of GDP, by 2030.
He also highlighted differences in data-centre construction times. According to Draghi, a facility takes approximately 24 months to build in the US compared with 42 months in Germany, reflecting longer permitting and grid-connection processes. He added that building an AI data centre in Sweden costs approximately 10% more than in China.
Draghi proposed that European companies combine their purchasing commitments into contracts of sufficient scale to support financing for new data centres.
He cited a group of European companies, including ASML, Capgemini and Amadeus, that have committed to multiyear purchases of Mistral’s European Compute Units. The initiative is intended to support 1GW of capacity by 2030.
Leave a Reply