Strengthening the European cloud and AI ecosystem
Summary
The impact assessment (IA) identifies two problems affecting the EU cloud computing ecosystem: limited availability of computing capacity in the EU, and excessive dependency on cloud and AI computing services supplied by non‑European providers. It defines four specific objectives that are consistent with the identified problems and drivers. It discusses a relevant set of policy options for each problem, including various specific policy measures.
Each option's main economic, social and environmental impacts are assessed compared with the baseline. Whereas the potential economic impact is thoroughly assessed, including through a quantitative analysis of costs and benefits for the key stakeholders, the analysis is less detailed for social and environment impacts. The IA neither addresses the potential impact on fundamental rights – despite acknowledging that several policy measures have such an impact (e.g. on data protection) – nor assesses the impact of policy options on grid capacity.
The IA adequately assesses each policy option according to the criteria of effectiveness, efficiency, coherence, subsidiarity and proportionality. However, the wide ranges of cost and benefit estimates for the policy options make it difficult to assess their effectiveness. The identified preferred options are consistent with the analysis comparing and ranking the policy options. The IA presents a set of relevant quantifiable monitoring indicators linked to specific operational objectives. Most of these indicators include specific targets, allowing the achievement of the objectives to be measured.
The IA makes use of a wide variety of data and sources, including a supporting external study and feedback from all major stakeholders. It explains in sufficient detail the methodological approach, including data sources, modelling approach, assumptions and limitations. The text is generally well‑structured, and the information is presented clearly.
The Regulatory Scrutiny Board (RSB) issued a 'positive with comments' opinion on the draft IA after initially issuing two negative opinions. The revised draft appears to have addressed the RSB's comments. The proposal appears to generally follow the preferred policy options identified in the IA. One notable omission is the removal from the proposal of the small and medium‑sized enterprises (SME) cloud and AI support scheme included in the preferred policy package.
Purpose statement (this title will not be shown in final layout)
This briefing provides an initial analysis of the strengths and weaknesses of the European Commission's impact assessment (IA) accompanying the proposal for a regulation establishing a framework of measures for strengthening Europe's cloud and AI ecosystem (Cloud and AI Development Act), submitted on 3 June 2026. The proposal was referred to the Committee on Industry, Research and Energy (ITRE) and the Committee on Internal Market and Consumer Protection (IMCO). This briefing analyses whether the principal criteria laid down in the Commission's own Better Regulation Guidelines, as well as additional factors identified by the Parliament in its Impact Assessment Handbook, appear to be met by the IA. It does not attempt to deal with the substance of the proposal.
Background
The proposal sets out a regulatory framework for a single market in cloud computing services, aiming to incentivise investment in cloud infrastructure and artificial intelligence (AI) research and development (R&D), which should boost the Union's competitiveness and strategic autonomy. The proposal is part of the tech sovereignty package, which also includes a proposal on a chips act 2.0, an open‑source strategy, and a strategic roadmap for digitalisation and AI in the energy sector. It is in line with the conclusions of the Draghi report on the future of European competitiveness, which strongly linked access to cloud services and increased computational capacity with the EU's ability to develop and adopt AI. The proposal also aligns with the EU Competitiveness Compass, which calls for adequate cloud and data infrastructure to ensure the EU's AI leadership. The initiative supports the goal set in the AI Continent action plan to triple the EU's data centre (DC) capacity within the next five to seven years. Moreover, it supports the target set in the Digital Decade policy programme, according to which 75 % of EU businesses should adopt cloud services by 2030.
The initiative is one of the flagship actions included in the European Commission's 2025 work programme. In its own‑initiative report on European technological sovereignty and digital infrastructure, adopted in January 2026, the European Parliament expressed concern about the EU's excessive dependency on non‑European actors in critical areas such as cloud infrastructure, and called on the Commission to introduce an EU cloud and AI development act to strengthen European data infrastructure and support European cloud providers. In its conclusions of December 2025 on European competitiveness in the Digital Decade, the Council called for the proposed act to include common criteria for sovereign cloud services, allowing for addressing market transparency and risks associated with dependencies.
Problem definition
The impact assessment (IA) describes the EU legal frameworks on different digital areas, highlighting gaps relating to fostering strategic investments in computing capacity and incentivising the uptake of European cloud and AI computing services. It identifies two major problems (Ps) and four main problem drivers (PDs) (see Table 1).
| Identified problem (P) | Problem driver (PD) |
|---|---|
| (P1) Limited and geographically concentrated availability of computing capacity in the EU | (PD1) Lack of scale and scope of European cloud and AI computing service providers |
| (PD2) Bottlenecks slowing down DC expansion | |
| (P2) Dependency on cloud and AI computing services supplied by non‑European providers | (PD1) Lack of scale and scope of European cloud and AI computing service providers |
| (PD3) Limited public sector uptake of cloud and AI computing services supplied by European providers and diverging procurement practices | |
| (PD4) Absence of clarity around the concept of sovereign cloud and AI computing services |
Subsidiary/proportionality
The IA identifies two possible legal bases for the proposal: Article 114 of the Treaty on the Functioning of the European Union (TFEU) for legislative measures aimed at improving the functioning of the internal market through the approximation of the provisions laid down by law, regulation or administrative action in Member States; and/or Article 173(3) for non‑legislative measures to enhance the EU's competitiveness and innovation capacity. The Commission's proposal was adopted as a single legislative instrument under the cumulative legal basis of Articles 114 and 173(3) TFEU.
The IA does not include a subsidiarity grid, as recommended by the Task Force on subsidiarity, proportionality and 'doing less more efficiently'. Nevertheless, it includes two subsections on subsidiarity (IA, p. 27), where it explains the necessity and added value of EU action. It argues that closing the existing computing capacity gaps cannot be sufficiently achieved by Member States' measures alone. Creating a common approach to accelerating DC deployment would enable the coherent planning and deployment of computing capacity in a geographically balanced way, while addressing the dependency on cloud and AI computing services by non‑European providers would deliver EU‑wide benefits that exceed what the Member States could achieve individually. Proportionality aspects are sufficiently discussed when comparing different policy options (under the criterion of 'subsidiarity and proportionality').
The deadline for the submission of reasoned opinions on the grounds of subsidiarity is 12 November 2026.
Objectives of the initiative
The general objective of the policy action is to ensure the functioning of the internal market for cloud and AI computing services, and to secure the conditions necessary for the Union's competitiveness and strategic autonomy. The IA identifies four specific objectives (SO) of the initiative:
-
(SO1) Increase computing capacity deployed in the EU through innovative and sustainable technologie s. By 2030, the EU should at least triple its current DC capacity, prioritising energy‑efficient technologies in at least 80 % of new installations (by 2035, the computing capacity in the EU should meet the bloc's needs).
-
(SO2) Ensure attractive conditions for the deployment of sustainable and innovative computing capacity. By 2030, operators should be able to obtain all permits to build and run a DC in less than 18 months throughout the EU, including access to land, permits for energy access, and connectivity.
-
(SO3) Decrease the overall reliance on non‑European cloud and AI computing services. By 2035, the market share of European cloud and AI computing service providers in the European market should reach 30 %.
-
(SO4) Contribute to the protection of public order by enhancing the resilience of supply of cloud and AI computing services, in particular in the public sector. By 2035, 100 % of the highly critical use cases in the public sector should be operated using sovereign cloud and AI computing services to ensure data confidentiality and operational autonomy, and prevent harms that could undermine public order.
The general and specific objectives are consistent with the problems and drivers identified in the IA. The links between objectives, problems and drivers are clearly presented (IA, p. 29). The specific objectives are sufficiently distinct and follow the SMART (specific, measurable, achievable, relevant and time‑bound) criteria. The IA also sets out operational objectives and performance thresholds, listing a set of relevant possible indicators for monitoring and evaluation, including data sources and data collection methods (IA, pp. 96‑97).
Range of options considered
The IA presents baseline scenarios showing the future evolution of the two problems over a 10‑year horizon (Annex 9 further explains the underlying assumptions of the baseline scenario). It presents a series of policy options (POs) and policy measures (PMs) for each problem (see Table 2). Although the two problems are interrelated (they concern successive steps of the same value chain), the IA identifies separate sets of POs and PMs in order to address the distinct regulatory and economic mechanisms affecting the two problems. The relationships between POs and PMs and the linkages between PMs and PDs are mapped clearly and appropriately (IA, pp. 50‑54).
The policy options for the first problem combine measures of different regulatory intensity and different calibrations of EU versus national delivery. Whereas the soft measures under PO1‑A (i.e. stronger collaborative mechanisms) could be combined with measures from the other options, the legislative and financial measures listed in PO1‑B and PO1‑C are mutually exclusive (in that they require enforcement at either national or EU level).
The options for the second problem are structured along a continuum of intervention intensity, starting from supporting measures to increase transparency and visibility of sovereign cloud and AI computing services, to establishing an EU‑coordinated procurement and support framework for such services. The non‑binding measures included in PO2‑A can be implemented together with the more ambitious measures under PO2‑B and PO2‑C. The measures under PO2‑B and PO2‑C are not mutually exclusive, yet some measures in PO2‑C represent a different gradient of intervention from similar measures in PO2‑B (e.g. voluntary award criteria (PM16) in PO2‑B vs mandatory award criteria (PM19) in PO2‑C). Under PO2‑A, PM11 establishes a harmonised Union‑level framework for sovereign cloud and AI computing services (not covering AI systems, which are subject to Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence – AI Act) that distinguishes between four levels of sovereignty assurance. The Commission has tested a sovereign cloud framework through its own tendering for €180 million worth of sovereign cloud services in 2026.
The IA briefly discusses the options discarded at an early stage, and provides sufficient justification for these choices (e.g. some of discarded options could be tackled better at other levels or might lead to unwanted consequences). For instance, it discarded measures tackling access to grid and the lack of grid capacity (a bottleneck for the deployment of DCs), because these were covered in the European grids package. However, the IA accompanying the proposal on the European grids package does not analyse in sufficient detail the issues of grid access and grid capacity needs for DCs.
Assessment of impacts
In accordance with the requirements of the Better Regulation Guidelines, the IA assesses the main economic, social and environmental impacts of each option compared with the baseline (IA, pp. 56‑77).
The policy options' economic impact concerns the impact on key stakeholders (industry, including SMEs, public authorities, and the European Commission) and the wider economic effects (impact on innovation and technological sovereignty). The IA quantifies the expected costs and benefits of each policy option for the main stakeholders over the next 10 years. In line with its sensitivity analysis, the Commission provides ranges of cost and benefit estimates for most policy measures. However, the resulting estimates span a range of uncertainty so wide that it is difficult to draw meaningful conclusions about the options' cost‑effectiveness.
Under PO1‑B, the estimated benefits for DC operators are much larger than the costs (€8 billion – €27 billion compared with €7 billion – €31 billion) (see Table 3). Under PO2‑C, the estimated benefits for cloud and AI computing service providers also largely outweigh the estimated costs (€233 million – €484 million compared with €2 billion – €5.1 billion). Under PO2‑C (PM21), private‑sector essential entities identified under Annex 1 of Directive (EU) 2022/2555 of measures to ensure a high common level of cybersecurity in the Union (NIS2) are expected to incur costs to comply with the obligation to carry out sovereignty risk assessment (between €486 million and €2.6 billion for 25 600 entities).
For national public authorities, PO2‑B and PO2‑C would bring much greater direct cost savings as compared with costs. Under PO2‑B, the costs for the public authorities who would voluntarily opt for porting critical cloud applications to sovereign levels 3‑4 (PM15) is estimated at €620 million – 4 billion (over three years from 2030). Similarly, under PO2‑C, which provides for mandatory sovereign risk assessment for the use of cloud and AI computing services in the public sector (PM21), the estimated porting costs for national public authorities amount to €3 billion – €15 billion. These costs are further explained in Annex 12; however, they are not discussed in Annex 4 (Section 3), which details the costs and benefits of each policy measure and allows for a comparative overview of costs and benefits across the key stakeholders. The justification provided for this omission is that 'these represent planned expenditure that would be incurred in the future as part of the regular cloud contact renewal cycles and independent of the present policy measures'. Nevertheless, the IA includes these costs in the assessment of the economic impact of policy options on national public authorities. The administrative and adjustment costs for the European Commission are estimated to raise according to the level of regulatory ambition (reaching €17 million – €19 million under PO1‑C and €470 million –€760 million under PO2‑C).
| PO
| DC operators | Cloud and AI computing service providers | National public authorities | European Commission | |||
|---|---|---|---|---|---|---|---|
| Costs | Benefits | Costs | Benefits | Costs | Benefits | Costs | |
| PO1‑A | 8 – 18 | 1 – 3.5 | Not applicable (n/a) | n/a | 1 – 2 | 0.1 – 0.6 | 0.5 |
| PO1‑B | 7 – 31 | 8 000 – 27 000 | n/a | n/a | 106 – 236 | 166 – 277 | 1 |
| PO1‑C | 7 – 27 | 5 000 – 12 000 | n/a | n/a | 84 – 90 | 6 – 19 | 17 – 19 |
| PO2‑A | n/a | n/a | 16 – 19 | 2 – 15 | 14 – 15 | 0 | 7 |
| PO2‑B | n/a | n/a | 67 – 160 | 200 – 700 | 820 – 4 400 | 8 000 – 17 000 | 130 |
| PO2‑C | n/a | n/a | 233 – 484 (+ 484 – 2 600 for private‑sector essential entities identified under Annex 1 of NIS2) | 2 000 – 5 100 | 4 000 – 18 000 | 21 000 – 61 000 | 470 – 760 |
The IA discusses, in qualitative terms, the impact on technological sovereignty and innovation, highlighting the great potential of PO2‑C to improve the EU economy's overall level of sovereignty, and to provide additional opportunities and incentives to procure from European providers. The policy options are expected to generate wider economic effects mainly due to boosting EU computing capacity and the adoption of sovereign cloud and AI computing services. As these effects are difficult to quantify, the IA relies only on qualitative analysis and evidence from the literature. Whereas the literature suggests that new DCs have positive economic impacts in the short to medium term, the magnitude of these effects depends on the context (e.g. local procurement patterns, labour market conditions, and the degree to which these expenditures are retained domestically) (IA, pp. 69‑70). Higher benefits may arise from additional computing capacity through the effects of increased cloud computing, AI and other data‑intensive activities on productivity and innovation. The increased adoption of sovereign cloud and AI computing services would result in enhanced autonomy, improved resilience, reduced dependency, stronger legal and operational control, and better conditions for AI‑enabled innovation (IA, p. 71). Under the new framework, a higher market share for EU providers is expected to shift cloud and AI spending towards services produced within the Union (opening a public‑sector cloud market of up to €4.5 billion for EU‑controlled providers by 2030). The IA also discusses negative effects, such as potential price premiums, whereas additional costs linked to providing sovereign services would be passed on to customers through higher prices.
The IA assesses the social impact of policy options focusing on the limited employment opportunities provided by the new DCs, and the shortage in information and communications technology (ICT) skills that can hamper the development of cloud and AI ecosystems in the EU. The assessment is rather brief, lacks quantitative evidence, and does not distinguish sufficiently between the impacts generated by different policy options (except for a potential contribution of PO2‑B and PO2‑C to the development of much‑needed ICT skills in Europe). The IA addresses the issue of public resistance against the construction of DCs. Whereas under PO1‑B, the greater role for local authorities would increase social acceptance of DCs, PO1‑C would have detrimental social impact, as the EU‑level decision‑making on DC deployment would be perceived as being further away from the citizens.
The environmental impact is assessed in terms of the impact of different policy options on electricity use, CO₂ emissions, and water consumption as compared with the baseline. All policy options would lead to increased installed capacity, and therefore to greater carbon emissions. However, under PO1‑C, higher efficiency and gradual adoption of sustainable‑energy integration measures means that CO₂ emissions per unit of added capacity would decline much faster than under the baseline. The cross‑border reuse of cloud and AI capacity among Member States, under PO2‑B and PO2‑C, would further lead to environmental savings. Although it acknowledges that 'grid capacity limitations are also real and increasing barriers for the development of data centre capacity', the IA does not assess the impact of policy options on grid capacity.
As stated in the IA, DCs are becoming a growing source of water consumption. Depending on the cooling technology used, DCs can consume large volumes of freshwater, which adds pressure on local water supplies, particularly in drought‑prone or water‑stressed areas. Although DCs represent a relatively small share of overall water pressure in the EU (with an estimated 76 billion litres of water consumed by DCs across the EU in 2025, i.e. 0.04 % of overall consumption), increased DC capacity is expected to lead to greater water consumption (IA, p. 76). The IA mentions that measures that incentivise resource‑efficient investments, technological improvements (e.g. adoption of advanced cooling technologies) and a balanced distribution of new DC capacity across Member States would reduce the environmental impact of water consumption. However, there is no further assessment or quantification of the impact of different policy options on water consumption.
The IA does not include an assessment of impacts on fundamental rights. This is despite the fact that several policy measures, such as the new framework for sovereign cloud computing services, require strong safeguards to ensure the lawful processing of personal data involving EU citizens. For example, in the definition of the second problem, the IA notes that 'dependence on US [United States] providers also exposes EU user data to extraterritorial laws and potential US government access resulting from the US Clarifying Lawful Overseas Use of Data (CLOUD) Act in the context of criminal proceedings, or Section 702 of the US Foreign Intelligence Surveillance Act' (IA, p. 14). However, the analysis of impacts does not consider these issues. Annex 3 (Section 1), which provides an analysis of who is affected by the preferred policy option, includes 'enhanced protection of fundamental rights, especially privacy and data protection' on the list of impacts affecting costumers and citizens, but without further discussion.
The IA assesses each policy option according to the criteria of effectiveness, efficiency, coherence, subsidiarity and proportionality. The effectiveness of policy options is assessed against the specific objectives. The efficiency of policy options is assessed with respect to costs and benefits borne by the main stakeholders. The policy options are compared against each other in terms of their main impacts, relative to the baseline, and against the five evaluation criteria. The IA ranks different policy options clearly and transparently. The potential economic impacts are assessed and compared; however, the wide ranges of cost‑and‑benefit estimates for the different policy options makes it difficult to assess their effectiveness.
The IA identifies preferred options. For the first problem, PO1‑B (legislative and financial intervention enforced nationally) appears as the most effective policy option for increasing the computing capacity in the EU. However, it also leads to the highest absolute increases in electricity demand and CO₂ emissions. To improve the environmental impact, the IA defines the preferred option for this problem as a package consisting of PO1‑B and two policy measures from PO1‑C, namely PM8 (identification of key EU‑level R&D funding challenges) and PM9 (EU funding for strategic DC deployment projects). Adding EU‑level funding to R&D and innovation into the mix is expected to bring 'a favourable balance between supporting deployment of capacity and better energy efficiency and environmental performance per unit of new capacity' (IA, p. 92). However, since the environmental benefits of PM8 and PM9 are not quantified, it is not clear to what extent these measures outweigh the negative environmental impacts of PO1‑B. For the second problem, the preferred option is PO2‑C (EU‑coordinated procurement and support framework for sovereign cloud and AI computing services), which is the best‑performing option across the comparison criteria.
Regarding alignment with the Sustainable Development Goals (SDGs), Annex 3 (Section 3) identifies contributions to six SDGs: SDG 4 (quality of education ), SDG 7 (affordable and clean energy), SDG 8 (decent work and economic growth), SDG 9 (industry, innovation and infrastructure), SDG 12 (responsible production and consumption), and SDG 13 (climate).
SMEs/Competitiveness
The initiative is considered relevant for SMEs, particularly for SMEs providing cloud and AI services. An SMEs test is provided in Annex 6. The initiative is expected to directly affect about 150 SMEs across the EU that are involved in DC construction and operation. The specific policy measure on the support for SME cloud and AI adoption (PM23) is expected to affect 65 000 SMEs. Finally, European SMEs in general (about 1.6 billion) would benefit through broader availability of cloud and AI computing services in the EU. The SME test provides qualitative and quantitative assessments of impacts (costs and benefits) of the preferred policy options on SMEs. While all options would generate benefits for SMEs, PO1‑B and PO2‑C would entail the greatest direct and indirect net benefits. Under PO2‑C, the SME cloud and AI support scheme (PM23) is expected to bring SMEs overall benefits worth of €328.4 million. The estimated costs for all SMEs preparing project proposals are estimated at €55.4 million.
The initiative is in line with one of the recommendations from the Draghi report and is part of the EU Competitiveness Compass. Annex 5 provides a competitiveness check analysis. The preferred policy package (PO1‑B) is expected to increase the number of providers, enhance competition, and reduce market fragmentation. In addition, R&D funding and innovation under PM8 and PM9 are expected to increase DC efficiency with spillover effects across the European cloud and AI value chain. The preferred package on cloud and AI computing services would initially generate additional costs for providers (which may be passed on to costumers), but once established, the framework is expected to reduce transaction costs and increase trust and market access. Internationally, DC operators in the EU are expected to expand their capacity faster and at lower costs.
Simplification, burden reduction and other regulatory implications
The IA discusses and quantifies the expected administrative costs and savings generated by the preferred policy options. It calculates the costs and savings for specific activities to be carried out by business operators (DC operators, cloud and AI service providers, independent auditors, SMEs, and private sector entities operating in sectors listed under Annex I to NIS2), and public administrations, broken down by policy measure. However, the table summarising these figures (IA, Tables 16‑17, pp. 94‑95) does not sufficiently outline the costs and savings for each policy measure (costs and savings are listed in the same column) The costs and benefits for the key stakeholders (excluding private auditors) are nevertheless listed in Annex 4 (Section 3). Annex 3 (Section 2) includes an application of the 'one in, one out' (OIOO) approach, with quantifications of new administrative burdens and removed administrative burdens for businesses and national public administrations. The total administrative burdens (for businesses and national public administrations) generated by the preferred options are estimated at €252.7 million to €1 billion.
Monitoring and evaluation
To monitor the implementation of the proposed regulation, the IA presents a set of relevant quantifiable indicators linked to specific operational objectives and performance thresholds. The operational objectives set out general targets allowing the success of the intervention to be measured (e.g. by 2030, overall permitting time should be less than 18 months). Annex 11 presents a coherent set of relevant indicators for output, outcome and impact monitoring. Whereas most output‑ and outcome‑monitoring indicators include specific targets, this is not the case with several impact‑monitoring indicators. For example, the target for the indicator 'lower carbon intensity of digital infrastructure' is simply 'decreasing KgCO₂ per compute unit', without indicating the level of decrease needed to establish that the impact has been achieved. An evaluation of the intervention is envisaged five years after implementation. The IA includes a set of relevant preliminary questions for the evaluation, targeting specific evaluation criteria; Annex 11 also explains the evaluation arrangements and timing.
Stakeholder consultation
In accordance with the Better Regulations Guidelines, a comprehensive summary of the stakeholder consultations, including an analysis of participants and their views, is provided in a separate annex (IA, Annex 2). The Commission organised a public consultation, which was open between 9 May and 3 July 2025, consisting of an online questionnaire and a call for evidence (CfE) for stakeholders to submit their views and recommendations on the objectives and the proposed actions.
| Stakeholder Consultation | Period | Total replies received |
|---|---|---|
| Online questionnaire | 9.5. – 3.7.2025 | 243 |
| Call for evidence | 9.5. – 3.7.2025 | 193 |
The Commission also organised a series of workshops, seminars and roundtables, as well as bilateral meetings with stakeholders from industry, academia, public administrations and finance. The IA describes in sufficient detail the inputs received on the main policy challenges, as well as the stakeholders' level of support for specific policy measures. Stakeholders' views are considered when discussing and ranking different policy options. For example, the IA refers to the strong support expressed by DC operators in the CfE for measures under PO1‑B (simplification of permitting requirements and fast‑track mechanisms). Similarly, under PO2‑B, the IA acknowledges the high level of support received from national public authorities for the federation of cloud and AI computing services.
Supporting data and analytical methods used
The assessment of impacts is based on several sources, including stakeholder consultation activities, and an external study carried out by an independent consultant. Annex 4 explains in sufficient detail the methodological approach, including data sources, modelling approach, assumptions and limitations. For example, it explains how wider economic impacts and non‑market social and environmental externalities were excluded from the cost‑benefit analysis (CBA) due to data limitations and lack of robust methodologies to monetise them. Annex 4 briefly discusses the strengths and limitations of the analysis. For example, it underlines that currently no data on the deployed DC capacity in the EU exists; a novel methodology to estimate this value was developed in the context of the supporting external study. Annex 4 explains clearly and with sufficient detail the sensitivity analysis used to indicate the potential range of error and uncertainty for all selected key estimates. It also explains the multi‑criteria decision analysis (MCDA) used to evaluate the relative performance of different policy options.
The text in generally clear and well‑structured. However, the IA does not use a consistent referencing style, using both footnotes and endnotes (Annex 0), without justifying this choice.
Follow‑up to Commission Regulatory Scrutiny Board opinion
The Regulatory Scrutiny Board (RSB) issued two negative opinions on the draft IA report (on 10 December 2025 and 25 February 2026). On 8 May 2026, the RSB issued a third positive opinion with comments. In its final comments, the RSB recommended that the Commission make further improvements in order to sufficiently:
-
describe the policy measure establishing a set of non‑price award criteria and the one encouraging the use, reuse, development and sharing of open‑source assets by the public sector;
-
explain and assess the internal coherence between the policy measures addressing sovereign cloud and AI services and those addressing critical dependencies and supporting the use of open source;
-
reflect the costs relating to migration and porting and the related uncertainties in the assessment of the total costs in the main report.
The IA explains in a detailed table (Annex 1, Section 3) how the RSB's comments from the three opinions were addressed. It appears that the RSB's comments were taken broadly into account in the final IA.
Coherence between the Commissions's legislative proposal and IA
The proposal broadly follows the preferred policy option identified in the IA. One key omission is the removal of the SME cloud and AI support scheme (PM23). Instead, the preamble of the proposal states that 'Member States should therefore aspire to award at least 25 % of relevant cloud and AI procurement innovation procedures to SMEs'. Similarly, the EU cloud and AI toolbox (PM24) was not retained in the proposal.
What follow‑up services are available?
The Ex‑Ante Impact Assessment Unit prepared this briefing to provide a brief overview of the Commission's impact assessment, as well as a succinct analysis of its methodological strengths and weaknesses. We remain available for any further questions, or to undertake any follow‑up work on this IA, in the form of a more detailed appraisal of the IA or a complementary IA, as may be deemed useful by the committee in relation to this legislative proposal. Practical details on how to request IA work are described in the Conference of Committee Chairs' IA Handbook.
Main references
- Marcelin, T., Cloud and AI development act, European Parliament, EPRS, December 2025.
Classification
Policy areas: Ex-ante Impact Assessment
Regions: European Union
Committees: Industry, Research and Energy (ITRE), Internal Market and Consumer Protection (IMCO)
Statement on the use of AI
Any AI‑generated content in this text has been reviewed by the author.
Disclaimer
This document is prepared for, and addressed to, the Members and staff of the European Parliament as background material to assist them in their parliamentary work. The content of the document is the sole responsibility of its author(s) and any opinions expressed herein should not be taken to represent an official position of the Parliament.
