Overview
Former European Central Bank President Mario Draghi wrote in the Financial Times that Europe must accelerate the construction of artificial‑intelligence (AI) data centres to safeguard economic sovereignty and lift productivity.
Productivity Gap
Draghi highlighted that the euro‑area versus United States productivity gap widened from $9 per hour in 2018 to $21 per hour in 2025. He cited ECB scenario analysis indicating that rapid AI adoption could contribute an additional 0.3 to 0.4 percentage points to total factor productivity growth each year, a metric that has been roughly flat since 2022.
Current AI Compute Landscape
According to Draghi, the European Union currently provides less than 5 % of global AI compute capacity, while the United States accounts for about 75 %. The shortfall between demand and installed supply in Europe is roughly 3 GW today—about a quarter of existing capacity—and is projected to expand to 14 GW by 2030.
Data‑Center Build Times and Costs
He noted that constructing a data centre takes about 24 months in the United States but up to 42 months in Germany because of longer permitting and grid‑connection procedures. An AI data centre in Sweden costs only about ten percent more than a comparable facility in China.
Funding Proposal and Corporate Commitments
Draghi proposed that European firms pool their procurement commitments into contracts large enough to finance new AI data centres. He cited a consortium that includes ASML, Capgemini and Amadeus, which have pledged multiyear purchases of Mistral’s European Compute Units intended to support 1 GW of capacity by 2030. He also mentioned that the European Commission estimates the continent’s data economy could exceed €800 billion, representing more than 5 % of EU GDP by 2030.
Implications
Draghi warned that being excluded from AI, once the economy becomes AI‑driven, would be comparable to being cut off from the U.S. financial system and could have catastrophic consequences for Europe’s competitiveness.