On the 11th of December 2025, a morning of discussion at the E.ON “Energy Playhouse” on the EU approach to AI in times of intense global competition and global power shifts took place. The event was also a timely opportunity to discuss how the EU could maintain its human-centric approach while fostering innovation, retaining talent and attracting investments, with the distinguished speakers Gaspard Demur, Deputy Head of Unit, AI Innovation and Policy Coordination, EU AI Office, European Commission; Sergey Lagodinsky MEP (Greens/DE); Quang-Minh Lepescheux, Director of European Government Affairs, Microsoft; Ray Pinto, Senior Director, DIGITALEUROPE and Sebastian Weber, Chief Information Officer, E.ON.
Barbara Moens, EU correspondent, Financial Times moderated the discussion.
Sebastian Weber delivered an introductory speech and set the scene for the panel discussion. He referred to the energy industry and the use of artificial intelligence (AI), emphasising the need for the EU to foster new technologies to manage its energy grids. He then proceeded by adding that the energy sector has produced a large amount of data over the last years, which is now being managed by algorithms, also called artificial intelligence. He highlighted the epochal transition that has been taking place for decades and which has now its current peak with the AI rollout. He also emphasised the importance for energy suppliers to use and develop artificial intelligence applications in order to be able to manage data and ensure the stability of the grid, as well as to provide solutions to customers, as in the case of solar panels or battery storage.
Mr Weber then expressed the opinion that, if the technology sector does not scale up, society as a whole will not be able to scale up. For this reason, data centres are now becoming a crucial infrastructure for Europe to catch up with the new trends and make sure that the old continent is competitive across all industries. He subsequently reiterated how E.ON takes these matters in high consideration, and added that Europe needs policy frameworks to master the use of AI. He concluded by saying that, despite the challenges ahead, he believes Europe will find its way through this aspect of the digital transformation and benefit from the advantages that artificial intelligence can provide.
Barbara Moens then proceeded to welcome and introduce the panellists on stage. Consequently, the moderator kicked off the discussion by asking Gaspard Demur how EU institutions are helping to create an environment which fosters AI-driven innovation.
Gaspard Demur began his reply by emphasising the European Commission’s commitment to boosting the uptake of Artificial Intelligence, especially in the energy sector, in order for Europe to increase its strategic autonomy. He continued by saying that, over the last year, the European Commission had launched several AI initiatives, such as the AI continent, the Apply AI Strategy and, more recently, the Digital Omnibus proposal.
He then proceeded by giving an overview of these initiatives, and added that the European Commission was actively working on European Digital Innovation Hubs, AI factories, data labs and Testing and Experimentation Facilities. He explained that there are already nineteen AI factories, but the EU is now ready to take the next step, by facilitating the creation of AI gigafactories.
Mr Demur then mentioned the Cloud and AI development Act, planned in March/April with the objective of tripling the number of data centres in the next five to seven years. This action will have an impact on the energy sector as well, he stated, as data centres have an impact on energy consumption. The speaker then explained that the EU needs to use the latest standards for the development of data centres, as well as to improve the efficiency of the architecture of the algorithm.
Mr Demur subsequently explained the rationale behind the Apply AI Strategy by stating that Europe needs specialised and dedicated smaller models for the industry to boost economic competitiveness. In this context, the European Commission has selected ten sectors such as health, energy and automotive in order to understand their specific needs and act accordingly. He then noted that the Apply AI strategy has a dedicated section to energy, aiming to ensure a higher quality of data, with the idea of producing home-grown models.
He also highlighted the importance of driving innovation in the distribution of grids and improving the efficiency of their different elements, as renewable energy is changing the way we are producing electricity, especially in terms of grid flexibility. He then concluded by saying that it is fundamental that the EU invest in AI as a solution to the energy transition.
Finally, Mr Demur provided insights into the Digital Omnibus package. He reiterated the importance of ensuring that stakeholders understand what is now available to them under the package. He also clarified that the European Commission introduced the AI Act Service Desk to ensure that all provisions are properly explained.
The moderator then turned to Ray Pinto, inviting him to comment on the European Commission’s actions and to outline, from his perspective, if anything further needs to be done.
Ray Pinto began by noting that several points raised by Mr Demur were of particular importance. He agreed that allowing sufficient time to develop clear standards for the implementation of the AI Act is necessary. In this context, he emphasised the need to recognise the environment in which EU companies operate, where sector-specific legislation is already in place. He further underlined that avoiding regulatory duplication when implementing the AI Act is crucial, and warned that the over-classification of activities as high-risk could be detrimental to the EU’s economic competitiveness.
Building on Mr Demur’s assessment, Ray Pinto acknowledged the European Commission’s positive approach to the benefits of AI and stressed that the swift adoption of these technologies is essential, particularly to ensure that both large and small companies can benefit and drive innovation across Europe.
He also addressed the need for a new, more efficient and sustainable grid infrastructure. While recognising that upgrading the grid would be costly, he pointed out that it could unlock up to €150 billion in renewable energy investments, given that the grid currently lacks the capacity to distribute much of the energy it is already capable of producing. This, he emphasised, is an area where AI could play a transformative role in the energy sector. The speaker added that it is equally crucial to shorten approval timelines through digitalisation for renewable energy projects, noting that, in many cases, up to half of a project’s overall duration is taken away by the permitting process. He also touched upon the discussion around the European Commission’s current framework for renewable energy, which provides targeted funding for both initial capital investment (CapEx) and ongoing production costs (OpEx). He suggested that it should be considered simply as total expenditures in order to ensure long-term financial sustainability of expansion and renovation of the grid.
The speaker concluded by stating that, in his view, the European Commission’s work is moving in the right direction. Nevertheless, he emphasised that further efforts are required for Europe to fully realise the potential of artificial intelligence.
As a follow-up on the discussion, the moderator asked Quang-Minh Lepescheux whether he agreed on Mr Pinto’s previous assessment.
Quang-Minh Lepescheux opened his remarks by expressing his agreement with the points made earlier. He observed that a shift is currently under way, as evidenced by the fact that technology companies are investing substantial resources in the development of sustainable solutions and the construction of energy infrastructure. He also expressed a positive view of the European Commission’s work, with special regard to the gigafactories initiative, and stated his belief that both EU institutions and industry are moving in the same direction.
Mr Lepescheux went on to argue that partnerships with international companies such as Microsoft can be complementary to the European Commission’s sustainability and innovation objectives. He noted that building infrastructure and data centres is becoming increasingly costly and complex, as the environmental sustainability of data centres has become an imperative. By way of example, he referred to the construction of a data centre in the Aragon region of Spain to illustrate the complexity of sustainability by design. He also agreed with Mr Pinto that permitting remains a challenge, but expressed confidence that the European Grids Package will help to accelerate approval processes.
The speaker added that companies such as Microsoft should take responsibility for supporting renewable energy by committing to the purchase and distribution of clean energy through Power Purchase Agreements (PPAs). He explained that this approach would enable technology companies to provide financial guarantees to energy providers, thereby facilitating the development of renewable energy projects to supply data centres.
He concluded by referring to the critical moment currently facing the EU, with significant developments under way in both the AI and energy sectors. He noted that technology companies are investing substantial resources, particularly in Europe, where industries are undergoing a rapid transformation. At the same time, he reiterated the importance of close cooperation with local governments and European institutions to ensure that such partnerships are conducted responsibly and deliver tangible benefits to local communities.
The moderator then invited Sergey Lagodinsky MEP to comment, shifting the focus of the discussion from an industry-oriented perspective to a political examination of AI-driven innovation and of the measures being undertaken by the European Union in this field.
Sergey Lagodinsky MEP began his reply by highlighting that the European Commission had removed the energy-efficiency requirements for artificial intelligence previously established by the European Parliament. He went on to argue that relying on private companies to adopt voluntary measures regarding AI energy consumption and sustainability in general is neither an effective nor a prudent approach.
MEP Lagodinsky then drew attention to what he identified as the core issue of the discussion. He cautioned against framing AI development merely as a competition between business models, stressing instead that it constitutes a competition between societal models. This, in turn, raises the fundamental question of what kind of artificial intelligence is desirable for EU citizens.
He also warned that increasing pressure from both Eastern and Western global actors risks diverting the EU from its commitment to value-based policies, particularly with special regard to environmental protection, energy considerations and regulatory standards. In this context, he highlighted the importance of smart, value-driven regulation as a means for the EU to avoid repeating mistakes already made elsewhere in the development and ownership of AI technologies.
Against this backdrop, the MEP questioned whether the European Union could establish a viable business case for a democratic and sustainable artificial intelligence. Referring to points raised by other panellists, he explained that he envisages the EU’s AI transition as being grounded in democratic and sustainable principles rather than centred on the construction of gigafactories.
He further noted that, for several stakeholders, the business rationale behind gigafactories – as distinct from AI factories – remains unclear and unrelated to considerations regarding strategic autonomy or sovereignty. Instead, he proposed that the resources earmarked for investment be channelled into the creation of a Sovereign Wealth Fund, through which the European Union could invest in and acquire shares in companies that are either already thriving or demonstrate strong growth potential.
From a broader perspective, MEP Lagodinsky stated that the European Parliament is broadly supportive of the European Commission’s initiatives and looks forward to complementing forthcoming proposals expected in 2026, including those related to quantum computing and the establishment of the so-called “28th regime”, with the aim of fostering a more favourable environment for European companies. He also emphasised the continued importance of the General Data Protection Regulation (GDPR), noting that several companies have acknowledged that compliance with this framework constitutes a competitive advantage for Europe.
In conclusion, he reaffirmed his position by stressing that transforming Europe’s certainty, security, freedom and democratic values into a credible business case is both achievable and desirable, and argued that this objective should represent a shared responsibility of EU institutions and the private sector actors alike.
Building on the points raised by MEP Lagodinsky, the moderator returned to Ray Pinto to invite his views on the role of trustworthy artificial intelligence as a potential differentiating factor for the European Union vis-à-vis the United States and China, and to ask him how this perspective is reflected within the association he represents.
Ray Pinto responded by stating that he did not disagree with the points previously made by MEP Lagodinsky. However, he argued that, in his view, trustworthy artificial intelligence is already a reality, as no company could remain viable if its AI products were based on bias or discrimination. He further clarified that this does not imply that the European Union does not require clear AI regulation. On the contrary, regulatory clarity and harmonisation are essential, he stated.
Turning to a different aspect of the discussion, Mr Pinto emphasised the importance of avoiding the misconception that artificial intelligence or data centres are inherently incompatible with sustainability, noting that such an assumption fails to take into account the benefits they generate. He referred to the 2009 COP in Copenhagen, in which data centre CO₂ emissions were projected to account for 4% of global energy consumption, whereas they currently represent approximately 1%. While data centres’ energy use could rise to 3% by 2030, he argued that, when the innovation enabled by these technologies is taken into consideration, the net impact may ultimately be significantly lower.
The speaker also highlighted the need to fund innovative and ethical business models and cautioned against reverting to what he described as a mid-nineteenth-century misconception that technology is inherently harmful, while society is inherently virtuous. He stressed that technology and society are deeply interconnected, and that technological progress is contributing to a safer and more efficient world, particularly in sectors such as healthcare.
Mr Pinto concluded by shifting the focus to international competitiveness. He observed that Europe is comparatively less competitive in the field of business-to-consumer technologies. However, in areas such as business-to-business solutions such as AI applications in the energy sector and AI-driven healthcare, European companies demonstrate particular strengths and are well positioned to compete globally. He eventually warned about the risks associated with overregulation and stressed the importance of ensuring an adequate level of investment.
The moderator invited Mr Demur to comment on the issues addressed by Mr Pinto.
Gaspard Demur explained that, in preparing the Digital Omnibus proposal and the Apply AI Strategy, the European Commission extensively engaged with both EU and non-EU companies. According to Mr Demur, none of these stakeholders questioned the importance of fundamental rights or reliability. On the contrary, they consistently emphasised the need for clear rules on the question of ensuring trustworthy artificial intelligence. He further noted that establishing this principle as a foundational element of AI regulation in the European Union has been crucial.
Mr Demur then shifted the focus to the sustainability dimension of the discussion, and reiterated the European Commission’s strong commitment in this area, notably through the provisions of the AI Act aimed at assessing and measuring the environmental impact of algorithms.
He also highlighted the importance of the “cloud dimension”, describing it as a critical component of the broader AI ecosystem. In this context, he explained that the European Commission is developing the Cloud and AI Development Act, through which EU institutions will be able to push standards and requirements for the energy-efficiency of European cloud infrastructure.
He stressed that this objective lies at the core of the European Commission’s approach, which involves close cooperation between DG CONNECT and DG ENER, as well as with all the stakeholders.
In conclusion, Mr Demur drew attention to the challenges facing the energy sector and highlighted the need for the European Union to accelerate its efforts. He emphasised the importance of early collaboration, data sharing and ensuring that the European Union develops its own tools for managing energy efficiency within the grid.
Sergey Lagodinsky MEP directly followed up on Mr Demur’s remarks, questioning why the European Commission had removed the energy-related provisions of the AI Act from the Code of Practice and asking what would be different under the forthcoming legal framework for clouds.
In response, Gaspard Demur explained that in the context of the Cloud and AI Development Act the European Commission is working on the link between the permitting procedures and the quality of project proposals, with special regard to energy efficiency and overall environmental impact.
Barbara Moens then took the floor once again, maintaining the focus on the Digital Omnibus proposal. She asked Gaspard Demur how the EU can demonstrate to the rest of the world that, despite postponing certain elements of the AI Act, it continues to uphold its principles and remains committed to its core values.
Gaspard Demur responded that this is ultimately a question of balance. He stressed that the EU intends to establish clear, proportionate and implementable standards – developed through extensive consultations – before rolling out the high-risk provisions of the AI Act.
He further explained that the European Commission is actively engaging with specialists across the field, and that this approach requires time and careful consideration. Nevertheless, he recalled that the Omnibus proposal sets a deadline of maximum sixteen months to avoid legal uncertainty. He concluded by affirming the importance of clearly communicating to companies what the AI Act offers and requests, notably trough the AI Act Service desk.
The moderator then steered the discussion further into the energy sector, inviting Sebastian Weber to share his expectations regarding the digitalisation of the energy sector and the related initiatives.
Sebastian Weber immediately noted that, in the context of the energy sector, E.ON has been relying on AI and data centres for more than a decade. He emphasised the importance of maintaining a stable grid and of providing customers with solutions that enable them to participate in the energy transition, notably through the use of technologies such as artificial intelligence.
He went on to state that he expects both the EU and the energy sector to take practical realities into account when shaping regulatory frameworks. While certain edge cases undoubtedly require regulation, he observed that these represent a minority of situations. The majority of cases, by contrast, involve relatively low levels of risk. On this basis, Mr Weber expressed his hopes for a nuanced regulatory approach, distinguishing between areas that require stricter oversight and those where risks are minimal and greater technological freedom can be preserved.
Barbara Moens followed up by asking Mr Weber how he approaches the debate surrounding the rising energy demand associated with data centres and the related concerns of consumers.
From E.ON’s perspective, Sebastian Weber explained that grid expansion is an essential part of the company’s responsibilities, although it is inherently time-consuming. He outlined two crucial dimensions of this challenge. First, it is necessary to develop a comprehensive understanding of customer demand across Europe in order to plan infrastructure needs, such as power lines, over the next decade. Second, there is a need to apply intelligent systems to manage currently scarce energy resources in the most efficient way possible.
The moderator then turned once again to Quang-Minh Lepescheux, asking whether he believed the narrative of “Energy versus AI” was exaggerated.
Quang-Minh Lepescheux responded that he did not consider this narrative to be exaggerated, describing it instead as a genuine issue that must be addressed. He highlighted, however, the positive dimension of this challenge, noting that it presents a significant opportunity to leverage AI for beneficial outcomes, such as more efficient energy use. He cited initiatives such as those undertaken by E.ON, where drones and AI are used to inspect power lines, enabling the anticipation of potential issues and the reduction of energy-related costs. He also strongly emphasised the importance of further developing such projects.
He went on to state that, when considering the European Commission’s work on the European Grids Package, he believes that all stakeholders are broadly moving in the same direction, offering a positive outlook for future developments.
Finally, Mr Lepescheux observed that the discourse on AI in the EU often centres on the need to develop new technologies in order to remain competitive with the United States and China, which are currently more advanced in this field. However, he stressed that an equally important and strategic dimension is the diffusion of AI. While Europe must strengthen its infrastructure and support AI companies, it is also essential to accelerate the adoption and integration of AI across EU industries and society as a whole, he concluded.
The moderator then took the floor by asking Mr Lepescheux what else could be done in this regard, and whether the issue was primarily one of mindset or of incentives.
Quang-Minh Lepescheux responded that the challenge relates both to regulation and to the need for a pragmatic approach. He stressed the importance of realism, noting that, while the EU may legitimately aspire to sovereignty in the field of AI, the immediate priority must be pragmatism. He explained that the EU needs to build partnerships with actors willing to comply with EU rules and respect European values, and to use these partnerships as a foundation for progress.
Sebastian Weber stepped in to build on the preceding discussion, reiterating his belief that the application of AI technologies represents Europe’s greatest opportunity. He argued that European businesses should focus on leveraging new technologies and, in particular, on accelerating the deployment of investments that others have already made. He concluded by stating that he did not believe in full European technological independence, emphasising instead that the priority should be the effective application of new technologies.
Sergey Lagodinsky MEP took the floor immediately thereafter, stating that he also did not subscribe to a vision of European sovereignty based on isolation. However, he argued that the core issue lies not in how AI is diffused, but in who controls the underlying infrastructure for AI and data. This, he warned, is becoming a significant risk for Europe.
He continued by explaining that the EU has historically relied on its US partners, but that the trust underpinning this relationship has been eroding as several initiatives coming from the United States, including the stances taken by President Donald Trump or by Elon Musk, in his view, have contributed to fundamentally undermine this trust.
MEP Lagodinsky also argued that, if the EU were to partner with an American company on grid infrastructure, there would be a risk that a future US administration could require the shutdown of grid-related data hosted on Microsoft’s cloud. Such a scenario, he suggested, would be sufficient to undermine trust in this type of partnership. He also added that the willingness of major technology companies to comply silently with the current US administration has weakened Europe’s trust not only in US political leadership, but also in large technology firms themselves.
He concluded by acknowledging that there is currently no clear solution to this dilemma. Nevertheless, he stressed that, if the EU is to pursue any initiative related to tech companies, it must do so in a manner that ensures the EU technology independence.
Quang-Minh Lepescheux then took the floor, pointing out that MEP Lagodinsky’s concerns are understandable. He agreed that the current geopolitical context is highly complex and acknowledged that the identified risks do indeed exist. He also added that Microsoft is making significant efforts to continue operating in Europe, including substantial measures aimed at protecting European customers from potential US sanctions.
Gaspard Demur joined the discussion by stating that the European Commission supports increasing strategic autonomy, while doing so pragmatically and without isolation. As an initial step, he noted, the Commission must ensure the availability of infrastructure capable of training and developing state-of-the-art models.
He concluded by highlighting the importance of generating added value from the training of models for specific applications, in the context of the Apply AI strategy.
Sebastian Weber stepped in the debate again, addressing the EU–US technological relationship more broadly. He argued that scenarios depicting a complete technological decoupling from the United States are not actually debates about AI. He pointed out that focusing narrowly on large language models represents only a small part of a much larger issue, given that much of the global internet infrastructure relies on either American and Chinese technologies. He also cautioned against expending excessive effort on solving only the LLM dimension of what is, in effect, a far broader challenge.
Barbara Moens followed up on this point by asking Mr Weber whether he is genuinely concerned that such a scenario could materialise.
Sebastian Weber replied that it is a concern, while also stressing that Europe, in practical terms, has very limited alternatives.
Ray Pinto concluded this segment of the discussion by raising an issue that had not yet been addressed. Rather than focusing on hypothetical scenarios, he argued, attention should be directed towards a reality already unfolding in Europe, namely Russian attacks on critical infrastructure. He questioned whether Europe is currently capable of adequately protecting its energy systems, suggesting that the answer remains uncertain. He noted that the cost–benefit balance of this confrontation currently favours Russia, and emphasised the urgency of addressing this challenge by accelerating the deployment of digital technologies across Europe.
The Q&A session covered the following topics: The possibility that LLMs may not be the solution and the justification for investments in LLM infrastructure; the lack of applicable AI models for use within LLM systems; the creation of business models around sectoral AI development; the need for thoughtful investment packages tailored to the specific needs of sectors; the Apply AI Strategy and the creation of foundational models for sectors; combining LLMs with the specific requirements of companies; data labs and gigafactories and the potential to adapt AI to sector-specific needs; public funding and the uncertainty regarding its actual allocation; dual-use in European research and science; Europe’s leadership in many business-to-business areas due to R&D investment; the need to accelerate innovation rapidly in Europe; dual-use research with civil applications as the primary focus under Horizon Europe; the importance of securing societal support for these initiatives by explaining that not all funding will flow into military or defence projects but also into civil infrastructure; the necessity of EU investment in dual-use research; Europe’s weak computing power and the difficulty of competing with the US even considering existing investments; the development of European-based technologies; the importance of AI for addressing cyber threats; the lack of continuity in European research; the need for coherent and practical cross-sectoral regulations in Europe; the European Commission’s AI office’s role in ensuring consistency as part of the Digital Omnibus; the role of EU Digital Innovation Hubs in guiding sectors to meet digital challenges; the need to clarify the definition of “high-risk” to avoid penalising small EU companies with excessive regulations; the Digital Omnibus deleting Article 4 on AI literacy and the resulting disinformation about the AI Act; the European Commission’s recognition of energy infrastructure as critical, and the importance of understanding which components are safety-critical.
Do you wish to know more about the issues discussed in this debate? Then check out the selected sources provided below!
AI Act Single Information Platform, European Commission
The General-Purpose AI Code of Practice, European Commission
Digital Omnibus Regulation Proposal, European Commission
Most interest in AI gigafactory builds is European, Commission says, Euronews
Cloud and AI development act, European Parliament Think Tank
The 28th regime, European Parliament Think Tank
Large language models (LLM), European Data Protection Supervisor
AI Act, Article 4: AI literacy, EU Artificial Intelligence Act
DIGITALEUROPE position paper on the Roadmap for digitalisation and AI in energy
DIGITALEUROPE AI and Tech Declaration
DIGITALEUROPE, The European CEO AI and critical tech declaration
