AI's Advance and the EU's Warning on New Inequality: Limits of Single Market Integration and Implications for Middle Powers

Executive Summary

In a background note for a meeting of EU finance ministers, the IMF warned that while AI could boost European productivity by about 1% over the next five years, its benefits and costs would be unevenly distributed across countries and social strata. The proposed solution, completing the EU single market, has a low probability of success due to the structural constraint of dispersed fiscal sovereignty among the 27 member states. The most likely path (a 55% probability) involves continued partial integration and uneven diffusion. As Europe's structural lag behind the U.S. and China in the competition for data centers and power grids becomes entrenched, some member states are pursuing niche strategies, such as Greece's effort to leverage the backlash against AI in the U.S. to attract investment. The dichotomy identified by ECB President Christine Lagarde—losing growth by shunning technology or becoming dependent on the U.S. and China by embracing it—suggests Europe may remain a rule-taker in discussions on AI safety norms. This implies that middle powers, including South Korea, must manage the risks of dependency on great powers for computing resources and technical standards and explore cooperative regional frameworks in response.

I. Analysis of the Current Situation

Concerns Over New Forms of Inequality Caused by AI Development: An Analysis of the Current Situation

1. Background and Developments

This warning came in the context of an informal meeting of EU finance ministers in Dublin, Ireland, on September 18-19. In a background note for the meeting, the IMF assessed that AI could boost European productivity by about 1% over the next five years [1]. However, it also pointed out that the benefits and costs would likely be unevenly distributed across countries, regions, and classes of workers [1]. The IMF proposed completing the EU single market as a solution, arguing that without deeper regional integration, the adoption and gains of AI would not spread evenly among the 27 member states [1].

This diagnosis aligns with an ongoing debate within Europe. The Italian daily Il Sole 24 Ore, citing the same IMF report, reported that 60% of European jobs are exposed to specific risks from AI [9]. This suggests the report submitted to the finance ministers focused more on risk management than on growth prospects [9]. Around the same time, European Central Bank (ECB) President Christine Lagarde explicitly articulated Europe's dilemma: a choice between losing growth opportunities by shunning technology and becoming dependent on tools developed by the United States and China by embracing it [14]. The Guardian framed this within the context of Europe's effective marginalization in the AI safety debate [14].

2. Current Situation

The three risks raised by the IMF—widening inequality, strain on power grids, and dependence on foreign technology—are each being discussed individually in current European policy discourse. The power grid issue is directly linked to the competition to expand data centers. Roland Berger has analyzed that Europe is falling behind in the global data center expansion race [4], reflecting a structure where the electricity, land, and capital needed to support AI infrastructure are concentrated in the United States and China [4][8].

The issue of dependence on foreign technology is paradoxically highlighted by the case of Greece. The U.S.-Greece Trade & Investment Council (USGTIC) has suggested that Greece can leverage the growing backlash in the U.S. over the energy consumption and economic benefit distribution of AI data centers as an opportunity [12]. The logic is that this AI backlash within the U.S. could encourage geographically dispersed investment [12]. This demonstrates that Europe is not merely in a passive position of dependency on U.S. technology; rather, certain member states are actively seeking niche opportunities amid this realignment.

The link to populist sentiment is detailed in an analysis by the Council on Foreign Relations (CFR). The CFR pointed to the possibility that AI development could repeat the winner-take-all structure of past globalization, likening the similarity to déjà vu [2]. It also noted that recent remarks by Anthropic CEO Dario Amodei have helped spread this debate beyond Silicon Valley into the broader political sphere, with both Senator Bernie Sanders and former President Donald Trump having stated their positions on the issue [2].

3. Key Actors and Positions

The IMF has taken on the role of both a warner and a promoter of integration on this issue. While acknowledging the potential for productivity gains, it has clearly stipulated that EU-level policy coordination is essential to realize them [1]. This is based on the judgment that responses at the individual member state level alone could actually widen the gap.

EU and Member State Financial Authorities began to address this issue as a formal agenda item following the Dublin meeting. However, the interests of individual countries are not uniform. There is a difference in perspective between countries like Italy, with a high proportion of exposed jobs [9], and countries like Greece, which approach the issue as an investment opportunity [12].

The ECB (Lagarde) is playing a role in framing Europe's strategic dilemma. By publicly addressing the trade-off between technological adoption and autonomy, she has sparked a debate on how Europe should manage its dependence on the U.S. and Chinese tech ecosystems [14].

U.S. Policy Circles (CFR, Brookings, Carnegie) view this issue through the expanded frame of a compute competition between the free world and authoritarian regimes. The Carnegie Endowment for International Peace has assessed that AI infrastructure will determine the future global balance of power, diagnosing that the window for democratic nations to take the lead is limited [10]. Brookings has conceptualized this as a "token divide," arguing that while the past digital divide was a matter of access, the new North-South divide in the AI era is emerging from gaps in computational power and access to tokens [6].

The Global South and ASEAN are structurally marginalized in this framework. A joint report by DBS and Bain & Company projected that within ASEAN, Singapore and Malaysia are likely to be beneficiaries of tech investment, while Thailand and Indonesia will face internal risks [16]. This shows that the problem of imbalance within Europe, as pointed out by the IMF, is a pattern that is being repeated globally, beyond the regional level.

The Bank for International Settlements (BIS) has stated its position on this gap, acknowledging that the concentration of technology at the national and corporate levels may be a natural phenomenon, while also emphasizing the need for international cooperation to ensure that the benefits are distributed evenly [15].

4. Summary of Key Issues

The first key issue is the feasibility of deepening EU integration. While the IMF has proposed completing the single market as a solution [1], the divergent interests among member states [9][12] suggest that this process could be politically challenging.

The second key issue is the way in which the gap in power and infrastructure is transforming into a matter of technological sovereignty. If Europe falls behind in the data center competition [4], the 'technological dependence' scenario mentioned by Lagarde [14] is more likely to become a reality.

The third key issue is how this structure of inequality will manifest politically. The parallel between globalization and AI drawn by the CFR [2] suggests that a backlash against the winner-take-all structure could spread beyond domestic politics to fuel transnational populist sentiment. This has implications for middle powers, including South Korea. Countries positioned as consumers rather than providers in the AI infrastructure competition [11] must now design their own response strategies, whether that means pursuing a European-style integrated approach or a Greek-style niche strategy.

II. In-Depth Issue Analysis

Concerns Over New Forms of Inequality Caused by AI Development: An In-Depth Analysis

1. Analysis of Root Causes

Underlying the IMF's recent warning is the structural diagnosis that the diffusion of AI technology cannot be achieved evenly through market mechanisms alone. AI infrastructure requires vast capital, electricity, and a high-skilled workforce—three resources that are all regionally concentrated. According to a Roland Berger analysis, data center investment is already showing a pattern of concentration in specific regions [4], and Europe is on the side where the gap is widening [4][8].

BIS General Manager Pablo Hernández de Cos explains this phenomenon as an inevitability inherent in the technology's characteristics, noting that concentration in a few countries and companies is a natural outcome [15]. At the same time, he stressed the need for international cooperation to distribute the benefits evenly across economies [15]. This logic is identical to the solution the IMF presented to EU finance ministers—deepening regional integration [1]. The shared diagnosis is that policy intervention is necessary because market forces alone will only intensify concentration.

The causes lie on both the supply and demand sides. On the supply side, giant U.S. and Chinese tech companies are leading the development of computing infrastructure and models. The Carnegie Endowment for International Peace has analyzed that the distribution of these computing resources will shape the future global balance of power [10]. On the demand side, differences in countries' absorptive capacities determine the outcomes. An EAI Special Report notes that developing countries often "have low levels of digital infrastructure and technological innovation, relatively insufficient capacity for digital transformation, and scarce resources to leverage other than labor" [7]. The "token divide" concept from the Brookings Institution follows the same logic: just as internet access created the digital divide of the past, the gap in access to AI models is now creating a new dividing line [6].

2. Structural Context

Political Structure: The Limits of EU Integration

The EU is a system where the fiscal sovereignty of its 27 member states is dispersed. This political constraint is the backdrop for the IMF's prescription of completing the single market [1]. In a structure that has a single currency but lacks a unified fiscal or industrial policy, decisions on AI infrastructure investment and regulation are made by each member state individually. Some member states are exploiting this gap to pursue their own positioning. For instance, Greece is attempting to leverage the backlash against AI data centers in the U.S. as an opportunity to attract investment [12]. This highlights the political reality that the interests of individual member states often take precedence over a coordinated EU-level response.

The dichotomy presented by President Lagarde—losing growth by shunning technology or becoming dependent on the U.S. and China by embracing it—exposes the structural vulnerability of European integration [14]. The Guardian has pointed out that Europe lacks a voice in the AI safety debate itself [14], suggesting that Europe may remain a rule-taker in the arena where technical standards and safety norms are established.

Economic Structure: Redistribution of Profits and the Labor Share of Income

An EAI report defines the nature of the change brought by AI based on Bain & Company's estimate of $4.7 trillion. It states, "The $4.7 trillion figure presented by Bain & Company is not new value-added, but the scale of redistribution of the existing industrial profit pool" [11]. The report further analyzes that "the approximately $7 trillion in AI infrastructure investment by cloud providers, combined with the all-time low U.S. labor share of income (52.8%) and all-time high corporate margins (14.9%), shows that this realignment is already proceeding in a way that distorts the distribution between capital and labor" [11]. This indicates that the "widening inequality" the IMF warned about is already being confirmed by data in the U.S. market.

The Italian outlet Il Sole 24 Ore, citing the same IMF report, reported that 60% of European jobs are exposed to AI-related risks [9]. This figure was presented with the diagnosis that Europe is more vulnerable than other regions [9]. This vulnerability is linked to the structure of the European economy, which has a high proportion of service-sector jobs.

Security Structure: Technological Dependence and Geopolitics

The issue of dependence on foreign technology extends beyond economics. The concern about dependence on U.S. and Chinese technology, as mentioned by Lagarde, has the potential to become a security issue [14]. The Carnegie Endowment for International Peace approaches this problem from the perspective that the distribution of computing infrastructure will determine the future of the free world [10]. This illustrates how the AI technology gap could be absorbed as a variable within the broader U.S.-China strategic competition. However, the degree of securitization is limited, as the IMF report itself focuses more on the potential for policy coordination failure within Europe rather than on the U.S.-China rivalry [1].

3. Comparison with Historical Precedents and Similar Cases

The CFR explicitly compares the current debate to the past process of globalization; the title of its analysis is "Déjà Vu" [2]. The concern is that just as globalization clearly divided winners and losers among and within nations, AI will create a similar distributive structure. During the globalization phase of the 1990s and 2000s, the benefits of trade liberalization were concentrated among capital and high-skilled labor, leaving low-skilled labor and certain regions behind. This experience led politically to the rise of protectionism and populist parties. The CFR suggests that if AI replicates this pattern, a similar political backlash could follow [2].

The precedent of the digital divide, as presented by the Brookings Institution, offers another comparison. In the early days of the internet, a gap became entrenched between countries that had access infrastructure and those that did not [6]. Brookings warns that this time, the gap could be replicated, centered on the AI computational unit known as a 'token' [6]. The argument is that while the divide of the internet era was a problem of telecommunications infrastructure, the divide of the AI era has shifted to one of access to models and computational resources.

At the regional level, the case of ASEAN is instructive. A DBS-Bain report cited by Nikkei Asia projected that Singapore and Malaysia would benefit from technology investment, while Thailand and Indonesia would face internal risks [16]. This case demonstrates how, even within the same regional economic bloc, the benefits of AI can diverge based on differences in national absorptive capacity. The situation is structurally similar to that of the EU, in that even with an integration mechanism like the single market, differences in member states' digital infrastructure and industrial structures can lead to different outcomes.

4. Key Variables Shaping the Issue's Development

The first variable is the actual progress of EU single market integration. The key question is whether the solution proposed by the IMF will remain a mere recommendation at the finance minister level or lead to concrete legislative action [1]. The EU's decision-making structure, which often requires near-unanimity, makes swift policy coordination challenging.

The second variable is the pace of power grid expansion. Europe's lag in the data center competition depends on the speed of investment in energy infrastructure [4]. This is a structural constraint that is difficult to resolve in the short term, as it is linked to national energy policies and the pace of the transition to renewable energy.

The third variable is the extent of the AI backlash within the United States. As the case of Greece illustrates, greater opposition to data centers in the U.S. creates an incentive for investment to become more geographically dispersed [12]. While this could be an opportunity for individual European countries, it is also linked to the risk of a growth slowdown in the U.S. AI industry itself. As an analysis by Mint points out, there are concerns that if the overheating of AI-related investment enters a correction phase, it could trigger a chain reaction of shocks [5].

The fourth variable is the actual rise of populist politics. For the winner-take-all structure feared by the CFR to translate into a political backlash, the shock to the labor market must become tangible, and voters must clearly attribute it to AI [2]. This is a variable more likely to unfold over the medium term than immediately. However, the fact that 22 countries have already adopted a joint position to keep AI under human control [18] indicates that political actors have begun attempts to manage this issue preemptively.