On 3 June, the European Commission presented its 'Tech Sovereignty Package'.[^1] A key part of that package is a proposal for a Cloud and AI Development Act ("CADA"), which has as its main objective to "triple (data center) capacity over the next 5-7 years, and to ensure that the EU has the necessary capacity to meet its needs by 2035".[^2] At Leitmotiv, we wrote a lengthy submission to the Commission last year to challenge this objective to triple data center capacity.[^3] Unfortunately, the Commission continues to steer in exactly the same direction. The question we at Leitmotiv have is: where is the evidence that Europe needs three times more data center capacity in the coming years?

The EU's own Guidelines require an evidence-based and transparent approach to policy-making

For context, it is relevant to point out that the Commission adopted internal 'Better Regulation Guidelines' ("Guidelines") in 2021.[^4] The Commission's 'better regulation' system relies on a number of key concepts and principles, including "an evidence-based approach", meaning that "policy decisions need to be informed by the best available evidence (including scientific evidence, where available)".[^5] Another key concept is "transparency": "preparing laws and regulation in a transparent way (i.e. openly documenting the process, making available the evidence underpinning political decisions and explaining the underlying rationale) improves the legitimacy and accountability of EU action".[^6]

One of the key instruments to ensure 'better regulation' in the EU is the "impact assessment".[^7] This instruments aims to "collect evidence (including evaluation results) to assess whether future legislative or non-legislative EU action is justified and, if so, how it can best be designed to achieve relevant policy objectives. They must: – identify and describe the problem to be tackled;...".[^8]

What evidence is underpinning the goal of tripling capacity?

How did the goal of tripling Europe's data center capacity in 5-7 years come about and what is the evidence base underpinning this goal? As so much of the Commission's work in this domain, it begins with the 'Draghi report', published in September 2024.[^9] According to the explanatory memorandum accompanying CADA, the Draghi report called on the European Commision to "expand... domestic computational capacity".[^10] Interestingly, looking at the Draghi report,[^11] there is no analysis of existing data center capacity in Europe and Draghi merely recommends in general terms that the EU should 'increase its computational capacity'.[^12]

The explicit goal of tripling Europe's data center capacity was only explicitly formulated in April 2025 in the Commission's 'AI Continent Action Plan'.[^13] In this plan, the Commission first claims that the "EU currently lags behind the US and China in terms of available data centre capacity". It then frames CADA as the solution to this problem, by tripling "the EU’s data centre capacity within the next five to seven years and bringing it to a level that meets the needs of EU businesses and public administrations by 2035."[^14] Any explanation of how the Commission arrived at this precise goal or evidence to back up this objective is missing from the AI Continent Action Plan as far as we can tell.

On 3 June 2026, the Commission finally presented a legislative proposal for CADA to the European Parliament and Council. The goal of tripling Europe's data center capacity remains in place, although the date by which this goal needs to be achieved is somewhat unclear.

In the Explanatory Memorandum attached to CADA, the Commission writes that it aims to "triple EU capacity in the next five-to-seven years and reach the needed capacity by 2035".[^15] However, in the attached 'Legislative Financial and Digital Statement', the Commission writes: "By 2030, the EU should at least triple its current data centre capacity, prioritising energy-efficient technologies in new installations. As demand continues growing, this should be considered an intermediate objective so that by 2035, the computing capacity in the EU should meet its needs.".[^16] It therefore remains unclear, at least to us, whether data center capacity needs to be tripled by 2030 or by 2031-2033 (5 to 7 years from now).

An explicit justification for the goal to triple Europe's data center capacity is also missing from the text of CADA. As we saw above, impact assessments accompanying legislative proposals are meant to assess whether legislative action is justified and to identify and describe the problem to be tackled. Yet, the impact assessment accompanying CADA lacks clear evidence on why Europe's data center capacity needs to triple.

The sole source to support this objective in the impact assessment appears to be a report by Technopolis Group, Wavestone, Timelex, STL Partners, OpenForum Europe and KAPA Research from 2025: "Study: Cloud and AI" ("The Technopolis Report").[^17] While the report itself does not appear to have been made public, which is problematic in itself as an intransparent practice, the Commission does cite from it in the impact assessment.

Did the Commission base itself on an unpublished study which acknowledges we simply don't know current data center capacity in Europe?

What is very interesting is that the Commission provides a massive caveat when introducing the Technopolis Report, which is the "main basis" for the Commission's impact assessment.[^32] The Commission writes about the report: "In spite of extensive efforts to collect relevant data through surveys, interviews and workshops certain gaps remain. There are a number of points for which projections and estimates had to be considered. For instance, there is no clear objective value on the deployed data centre capacity in the EU."[^18] What the Commission is admitting here is that its only basis to justify its goal of tripling the EU's data center capacity – the unpublished Technopolis Report – had no access to objective data on current data center capacity in the EU, the logical baseline for such a study.

Indeed, the section of the impact assessment that describes 'estimates related to data centres' starts as follows: "There is no single, EU-wide dataset produced by regulators or energy system planning exercises that consistently locates historical data centre capacity at site level in terms of commissioned IT load."[^33]

The absence of transparency in this market and the inability of the European Commission so far to adequately regulate for such transparency means that we simply do not know where we are at at present. Instead, the Technopolis Report had to rely "primarily on the Data Centre Map dataset" to estimate current capacity. However, anyone who has looked at this website,[^34] knows that it does not contain precise information on matters such as total electricial capacity, total IT demand, or utilization rates. In fact, if we look at the sites of some of the major data center operators in a major data center market like the Netherlands, it appears that this website has no information at all on those sites.[^35]

Problems with the Technopolis Report: too many 'coincidences'?

Apart from the fact that they are included in an unpublished study that acknowledges we do not know the current baseline of data center capacity in Europe, there are a number of problems with the the Technopolis Report. These problems are relevant, because this one study appears to be the sole evidentiary basis for the Commission to conclude we need to triple capacity in Europe.

According to the estimates of the Technopolis study, Europe had 12.4 GW of installed IT capacity in 2025.[^19] Without further policy action, the Technopolis Report estimates that installed capacity will reach 28.1 GW by 2030.[^20] At the same time, Technopolis expects that installed capacity demand will reach 37.3 GW by 2030.[^21] While this is not stated explicitly in the Commission's impact assessment, we assume that these figures are meant to back up the goal of tripling Europe's data center capacity. In short: the current capacity figure of 12.4 GW needs to triple by 2030 to reach the demanded capacity in that year of 37.3 GW (12.4 times 3=37.2).[^27] There are a few problems with these calculations, however.

First, it is striking that the estimated demand figure of 37.3 GW for 2030 is almost exactly three times the current estimated actual capacity in Europe of 12.4 GW. As we understand the timeline, the Commission settled on the objective of tripling EU data center capacity in the AI Continent Action Plan that was published in April 2025. This objective is now almost precisely confirmed in estimates of the Technopolis Report.

A second problem relates to data the European Commission itself collects and publishes based on the Energy Efficiency Directive ("EED") of 2023 and a delegated regulation adopted by the Commission in 2024.[^26] The Commission has recently developed a dashboard based on this data, which includes figures up until the reporting year of 2024.[^22] According to the dashboard, the total IT power demand installed in Europe in 2024 was 4.5 GW.

How do we square this figure of 4.5 GW of IT power demand installed in 2024 in the Commission's own database, with the figure of 12.4 GW for 2025 apparently used in the Technopolis report? What is interesting is that in an assessment for the European Commission of the first reporting period of the EED (2024), it was concluded that "approximately 36% of eligible EU data centres or around 770 facilities participated in the reporting scheme...".[^23] Calculating backwards, 4.5 GW in the Commission's dashboard represents 36% of 12.5 GW, almost precisely the installed IT capacity assumed by Technopolis for 2025. Like above, we wonder whether this is merely a coincidence?

If the EED data collected by the Commission would indeed be the basis for assuming that the current maximum data center capacity in Europe (we do not know this for sure, of course), that would be a significant problem from the perspective of evidence-based and transparent policymaking.

Not only did 64% of data center operators not 'participate' in the EED scheme for the reporting year 2024 (meaning they did not meet their legal obligations under this law), but – as Leitmotiv has shown for a major data center market in Europe – compliance with the EED has been especially low among non-European hyperscalers.[^24] In short: the data underlying the EED database of the Commission is neither complete nor representative.

A third and final problem with the unpublished Technopolis Report is that it appears to assume that 'installed IT capacity' indicates the maximum capacity of data centers in Europe at present. That is not true, however. There may be (much) more room in existing data centers to install additional IT capacity in.

Leitmotiv showed this for the Netherlands,[^28] using data that is collected based on the EED. According to reports received from data center operators, data centers in the Netherlands operate at only 1/3rd of their total electrical capacity (in other words: they are 2/3rds empty).[^38]

The Netherlands is a highly relevant market, as it is one of the biggest data center markets in Europe according to a report written for the European Commission[^36] and the Technopolis Report itself.[^37] What is more, the IT capacity that is currently already installed may be underutilized.[^39] If we are able to improve the utilization of existing IT equipment, we may not have to build that many new data centers or install new IT equipment.

Should Europe copy an American strategy for data center build-out or have its own leitmotiv?

Europe is on course with the CADA goal of tripling data center capacity to emulate the American approach to AI development: scaling.[^29] But, as we wrote earlier,[^25] it is simply unrealistic and unnecessary for the EU to copy the American way of AI development.

In essence, Europe has already outlined an alternative. It has committed itself to the Energy Efficiency First Principle.[^30] It has set out to create decentralized data center capacity through the European Cloud Continuum and funded this through the IPCEI instruments.[^31] What these initiatives are based on is building smaller, smarter and more distributed to reduce pressure on the energy system. It also seeks to create a level-playing field in Europe by focusing on integration of existing capacity, rather than supporting the build-out of hyperscale AI factories.

In our submission to the European Commission on CADA last year, we furthermore presented alternative policy ideas, including the establishment of a European Exchange for Digital Resources; a marketplace for digital resources that enables price discovery, transparency, and competition in European digital infrastructure markets.

As calls for data center moratoria are heard around the world and in Europe,[^40] and with alternative ways forward possible, why rush into tripling our data center capacity and deviate from long-established evidence-based and transparent approaches to policymaking to fit a square into a circle?

References