As artificial intelligence (AI) continues its rapid ascent, the environmental costs and energy demands of this technological revolution are drawing intense scrutiny from experts, policymakers, and industry leaders alike. In recent months, the debate has heated up across both sides of the Atlantic, with growing concern over the sustainability of AI’s infrastructure and its alignment with climate goals.
Professor Benedetta Brevini, a visiting professor at New York University and associate professor at the University of Sydney, has been at the forefront of these discussions. In an interview with Reuters on October 2, 2025, Brevini called for a more rigorous accounting of AI’s environmental footprint, warning, “We need to better measure AI's environmental impact and ask if all of its uses are needed.” Her comments come as Europe braces for a record number of new data centres this year, a surge driven by the explosive growth of AI and cloud computing.
The issue is hardly confined to Europe. In the United States, the energy implications are also coming into sharp focus. U.S. Energy Secretary Chris Wright told Reuters in late September that most of the country’s coal-fired power plants are now expected to delay their retirement, in part to supply the vast amounts of electricity required by AI systems. This development underscores the paradox at the heart of the digital transition: while AI promises efficiency and innovation, it also brings a voracious appetite for energy and resources.
At the heart of the matter are the massive data centres that power AI applications. These facilities, often located in regions already struggling with scarce natural resources, have become lightning rods in the debate over sustainability. According to Brevini, “There are communities, very often in areas where you have constant droughts and constant floods, that really cannot cope with hosting data centres. And let's remember that data centres don't bring jobs either.” She urges a fundamental rethink: “We need to really think, are we using water for our citizens? Are we using it for data centres? Just because you want to transform a resume into a poem? Is this the end game?”
This tension is not lost on the energy sector itself. At a Euractiv stakeholder workshop on September 30, 2025, Lorenzo Fiorillo, technology director at Italian energy giant Eni, described how traditional energy companies are transforming into digital enterprises. “It’s not just from now that we’re considering ourselves a technology company, because this is a mindset that is in our DNA for a while, if you look at our historical decisions,” Fiorillo said. He pointed to Eni’s long-standing investment in data centres, including HPC6, one of the first major data centres in Europe, as evidence of this evolution.
Yet with opportunity comes challenge. Pablo Riesgo Abeledo from the European Commission’s energy department noted that the upcoming AI and Cloud Development Act, part of the Commission’s AI Continent Action Plan, aims to create a regulatory framework that aligns new digital infrastructure with Europe’s climate and strategic autonomy objectives. “The strategic decision with the Cloud and AI Act is clear: we want to have these data centres in the EU, but there are challenges in having them,” he said. “Sometimes they have huge power needs, and the evolution is that they’re getting bigger and bigger.” Unlike traditional industries such as steel plants, which take decades to build, data centres can be connected to the grid in a matter of years, placing enormous strain on energy systems.
To tackle these pressures, industry and policymakers are looking at innovative solutions. Waste-heat recovery is one such approach: data centres generate substantial heat, which could be repurposed for district heating, thereby improving overall energy efficiency. Strategic siting is another focus, with experts emphasizing the need to locate data centres near abundant renewable energy sources and robust grid infrastructure. However, as Fiorillo cautioned, “The well-located areas are limited – those that are close to renewable sources with available grid power, with no environmental constraints, you get a situation where that available area is saturated very soon.”
To bridge the gap when renewable energy isn’t continuously available, Eni is exploring technologies to produce power while capturing and storing CO2 emissions. This approach, Fiorillo explained, is about ensuring a continuous energy flow for data centres even when renewables fall short.
Permitting and regulation also loom large. The European Commission is working to streamline the permitting process to avoid “stranded assets or speculative projects,” as Riesgo Abeledo put it, where companies apply for multiple projects but only proceed with a select few. This regulatory clarity is seen as crucial for both the public and private sectors as they navigate the digital and energy transitions.
Importantly, the environmental conversation extends beyond energy consumption. Brevini argues that the entire lifecycle and global supply chain of AI must be considered, from the carbon footprint and rare metals used in hardware like Nvidia chips to the mounting problem of electronic waste. “You must follow everything: the carbon footprint, the number of rare metals used in a new application, the water usage and where these devices will end up. You need a strategy on deciding what kind of development Europe needs and to look at the infrastructures – taking them from the 'digital lords' – and make them public,” she said to Reuters.
Research into greening software and data centres is underway, with a growing push for what Brevini calls “green AI literacy.” Training data scientists and engineers to question the sustainability of their computing choices could yield significant energy savings. “Very often these people will tell you that they're given infinite resources, but they only needed one third of this. That is a very good way to reduce energy use,” Brevini explained. The goal is to foster a culture where sustainability is a first consideration, not an afterthought.
Meanwhile, the European Commission is also looking at the positive side of AI and data centres. Juan Pelegrin from the Commission’s digital department highlighted how supercomputing can improve the accuracy of fossil fuel exploration and optimize wind farm placement, potentially boosting the efficiency of the entire energy grid. “With supercomputers we can do the modelling, improve the precision of forecasts. This has a lot of policy implications; it can even save lives,” he noted.
To help small and medium-sized enterprises (SMEs) tap into these opportunities, the Commission has launched the AI Factories Initiative, which provides access to supercomputing resources and support for algorithm development. Vincenzo Renda, Director for Digital Transformation Policy at DIGITALEUROPE, welcomed these efforts but warned that regulatory complexity could hinder AI adoption. The EU AI Act, while providing legal certainty, may slow uptake, prompting calls for regulatory simplification in the upcoming digital package expected in December 2025.
As the world races to harness the power of AI, the challenge remains clear: balancing digital innovation with environmental stewardship. The choices made now will shape not only the future of technology but the health of the planet itself.
