The Crisis of European AI Lag: Draghi Calls for Data Centers, but Is There Enough Money, Power, and Talent?
Summary of the Main Points
In short, this article discusses how Europe is falling far behind in the race for artificial intelligence (AI) and how former European Central Bank President Mario Draghi has once again issued a “red alert.”
Draghi points out that Europe currently accounts for less than 5% of the world’s AI computing power, while the United States holds 75%. If Europe does not quickly establish its own large-scale AI data centers, it will not only lose momentum for economic growth but also risk losing control of its “data sovereignty” to the United States—meaning that Europe’s economic lifeline could be in the hands of American tech companies.
Although Europe has a vast amount of data resources, it lacks the infrastructure (data centers) to process this data. Draghi suggests that Europe must build its own data centers and proposes a solution: through policy guidance, large European companies should join forces to purchase computing power collectively to attract investment. However, the reality is challenging: slow approval processes, high electricity costs, dispersed demand, and conservative societal attitudes make it difficult to change the situation.
---
Detailed Analysis
1. The Harsh Reality: Europe’s AI Computing Power Accounts for Only 5% of the World’s
First, we must face the stark reality that Europe is significantly behind China and the United States in terms of AI hardware and computing power.
Draghi provides the numbers: the U.S. accounts for 75% of global AI computing power, while Europe has less than 5%. This means that while American and Chinese companies are using AI to accelerate research and development and optimize production, European companies are lagging far behind.
What’s more concerning is the gap in productivity. The productivity gap between the eurozone and the United States has widened from $9 per hour in 2018 to $21 per hour in 2025. It’s like two runners: the gap was small before, but now the American runner is not only faster but also accelerating, while Europe has almost stopped progressing (with near-zero growth in total factor productivity).
Draghi believes that AI is Europe’s only hope for a turnaround. If Europe can quickly adopt AI, it could increase its economic growth rate by 0.3 to 0.4 percentage points per year. But there’s a prerequisite: you need the computing power, the electricity, and the data centers. Without these infrastructures, talking about AI applications is meaningless.
2. The Core Pain Point: Data Is Europe’s Asset, but Control Is Not in Its Hands
Many think Europe’s lack of success is due to poor technology, but that’s not the case. Europe’s greatest asset is its data, and its greatest weakness is the lack of control over it.
Europe has decades of medical records, extensive data from statistical institutions, and massive amounts of industrial information from its highly automated manufacturing sector. By 2030, the European data economy is expected to exceed 800 billion euros. It’s like Europe holding a huge gold mine, but most of the tools (computing power) and trucks (data centers) for extracting the gold are in the hands of Americans.
Draghi warns that this is a “catastrophic dependency.” If Europe’s economy, healthcare, and defense rely on AI, and the underlying computing power is controlled by the U.S., future U.S. governments could link the use of AI to the imposition of “digital taxes” or changes in digital regulations.
For example, imagine living in a house built by Americans and using their heating system. What if the landlord says, “If you don’t obey, I’ll cut off the heat?” Europe must have its own data centers, especially for handling sensitive data (such as in defense and healthcare), which must be managed entirely within Europe.
3. Practical Barriers: Why Doesn’t Europe Build More Data Centers? (High Cost, Slow Approval, Scattered Demand)
If Europe knows it needs to build data centers, why isn’t it doing so? Draghi identifies three main obstacles:
- Slow Approval Processes: Building a data center takes 24 months in the U.S. but 42 months in Germany. Europe’s cumbersome approval procedures and grid connection processes slow down projects.
- High Electricity Costs: Electricity prices in Europe are much higher, doubling the construction and operating costs compared to China. Although Sweden has lower costs, overall, Europe lacks a competitive advantage.
- Scattered Demand: In the U.S., giants like Amazon, Microsoft, and Google can sign multi-decade contracts, providing a stable source of demand that attracts investment. In Europe, demand is spread across millions of small and medium-sized enterprises, and large companies prefer annual purchases rather than long-term contracts. Without stable long-term buyers, no one is willing to invest billions in data centers.
As a result, most of the computing power in European data centers like Nscale is sold to American buyers because they are willing to sign long-term contracts. European companies can’t afford or dare to use these centers.
4. The Solution: European Companies Buying Computing Power Collectively
To address these issues, Draghi suggests policy guidance, company alliances, and long-term commitments. He proposes that large European companies like ASML, Capgemini, and Amadeus form purchasing alliances to pool their moderate demands and create large, long-term contracts. This would give investors confidence and encourage investment in new, AI-specific data centers. This would create a positive cycle: orders → investment → data center construction → cost reduction → more usage → more orders.
Some European companies have already committed to providing computing power for Mistral AI (a French AI company), with the goal of reaching 1 gigawatt (GW) of computing power by 2030. Draghi hopes to expand this model to involve more companies and cloud service providers, allowing Europe to truly establish its own “computing power pool.”
5. The Deeper Concern: Conservative Societal Attitudes Hinder Future Competitiveness
Beyond technical and economic issues, there’s a deeper, more challenging problem: European society’s overall attitudes are not keeping up with the times.
Researcher Zhao Yongsheng from the University of International Business and Economics notes that Europe is very “traditional and conservative” when it comes to AI. This is reflected in legislation (favoring privacy and restricting data flow) and public opinion. Many people oppose data center construction due to concerns about environmental pollution and increased electricity costs.
This conservatism will not change in the short term, as it reflects the values of most Europeans. However, the next generation in East Asia and North America is already using AI ahead of Europe. If Europe continues to limit AI development due to lagging societal consensus, its competitiveness will be severely impacted in 10 years.
Draghi also mentions that environmental concerns around data centers can be addressed by using renewable energy, recycling waste heat, avoiding resource-scarce areas, and sharing benefits with local communities. The key is whether European society can reach a new consensus and embrace the changes brought by AI, rather than being constrained by conservative attitudes.
Conclusion
Draghi’s article is both a technical report and a political statement. He calls on Europe to stop being a mere user of AI and become its controller. For ordinary Europeans, this means that AI will profoundly affect employment, healthcare, and education. If Europe can successfully establish its own AI infrastructure, it could create new economic growth opportunities; otherwise, it may be marginalized in the global tech race, becoming merely a market for American and Chinese AI technologies rather than a source of innovation.
This competition is not just about computing power but also about systems, attitudes, and societal consensus. Whether Europe can break free from its conservative constraints will determine its future fate.