IMF: 60% of Workers in Advanced European Economies Are Highly Exposed to AI

IMF: 60% of Workers in Advanced European Economies Are Highly Exposed to AI

An IMF background note warns that the gains from artificial intelligence in Europe will be unevenly distributed across countries, regions and workers, while data centers strain power grids and Europe risks new technological dependence on the US and China.

EU finance ministers have a new number to reckon with in the discussion of artificial intelligence: according to a background note from the International Monetary Fund (IMF), AI could lift European productivity by around 1 percent over a five-year period. But the note, prepared for the informal meeting of EU finance ministers in Dublin on September 18–19, is far from a message of pure optimism. The gains and costs will, according to the IMF, likely be distributed unevenly across countries, regions and workers, while the growth in AI use strains power grids and increases Europe's dependence on foreign technology — unless governments deepen economic cooperation.

The figures and recommendations are reproduced here as reported by Reuters (Jan Strupczewski) and Crypto Briefing, both of which build on the IMF's note. The note itself is not available in the sources AIMag has had access to, which sets limits on what can be independently verified.

Six in Ten Workers in Highly Exposed Occupations

Perhaps the most concrete assessment in the note concerns the labor market. The IMF estimates that around 60 percent of workers in advanced European economies are employed in occupations highly exposed to artificial intelligence.

But exposure does not mean the same thing for everyone. According to the note as reported by Reuters, some will become more productive through AI tools, while others risk being displaced as routine tasks are automated. This applies particularly to occupations where, in the IMF's assessment, AI is more likely to replace labor than to complement it.

The note also points to differences between countries. AI gains will, according to the IMF, likely be unevenly distributed across and within the EU, and more advanced economies are expected to benefit disproportionately because they are better prepared for and more exposed to the technology. This means that an average productivity boost of 1 percent can conceal large differences in who actually reaps the gains — both between member states and between workers in the same economy.

Data Centers Strain the Grids Where Tech Clusters Stand

The second challenge is physical. The note states that Europe's data centers already use roughly 3 percent of the continent's electricity, and that demand will rise sharply as AI use grows.

The pressure is, according to the note, most concrete where the tech clusters already stand. Among the most affected areas are Frankfurt, London, Amsterdam, Paris and Dublin, where clusters of data centers are already straining local power grids.

The IMF's answer to this is structural: the fund recommends cross-border grid investments and deeper integration of the European energy market. The logic is that a more tightly connected European grid can distribute load and capacity better than national grids that largely handle the pressure on their own. The recommendation fits a familiar IMF framework: if AI is to deliver gains without creating new bottlenecks, the internal markets — for goods, services and energy alike — must function more deeply than they do today.

Warning of a New Strategic Dependence

The third thread in the note concerns who owns the technology. The IMF warns that Europe risks developing a new strategic dependence because the US and China dominate the development of AI models. To avoid becoming dependent on foreign technology, the note says Europe will need significant investment in its own AI industry.

The phrase "a new strategic dependence" points beyond AI itself — it places model development in the same category as other strategic inputs Europe is already vulnerable to shortages of in domestic production. For the finance ministers in Dublin, this means the AI question is not only about productivity and the labor market, but about industrial capacity and security-policy positioning.

What the IMF's Recipe Adds Up To

Taken together, the IMF's recommendations can be read as one main idea: Europe gets the most out of AI if it acts as one market, not twenty-seven. Concretely, that means:

• Deeper integration of the internal market, so that AI gains spread faster between economies.
• Deeper integration of the energy market and cross-border grid investments, so that the data centers' power needs do not crush local grids.
• Significant investment in a European AI industry, to reduce dependence on American and Chinese models.

Open Questions

Several things remain before the picture is complete. The IMF's note itself is not available in the sources this story is based on, so all figures and recommendations rest on secondary coverage from Reuters and Crypto Briefing. The two reports were published the same weekend and likely build on the same document; they do not independently confirm each other's figures.

The methodology behind the 1 percent figure — the baseline, assumptions and uncertainty ranges — is not described in the available sources. Nor is it specified which occupations or countries the IMF's displacement estimate refers to concretely. Crypto Briefing adds that similar estimates appeared in earlier IMF analyses from April and November 2025, but that context has not been verified against the IMF's own documents in this story.

Beyond Dublin

The Dublin meeting is informal, and the note is a background document — not a decision. But the timing is significant: the EU is working in parallel to position AI both as a competitive opportunity and a risk in its economic governance, and the IMF's figures give the finance ministers a common reference point for the discussion of what a 1 percent productivity boost requires in policy terms. Whether the ministers follow up on the IMF's recommendations on market and grid integration, or whether the note remains an analytical input, is an open question after the weekend's meeting.