A recent report by the International Energy Agency (IEA) suggests that data centers accounted for approximately 1.5% of global electricity consumption in 2024, a figure that marks a 12% increase over the past five years [1]. While 1.5% might sound small at first glance, when scaled to the entirety of global electricity production, it represents an enormous and rapidly expanding energy footprint. Much of this growth is being linked to the explosive rise of a new class of machine learning systems known as Generative Artificial Intelligence (AI) models that are behind the popular chatbots, for example. These models, which are far larger and computationally hungrier than earlier AI systems, demand significant resources for both their training and inference phases.
In Denmark it is projected that the energy consumption by data centers there will be seven times higher by 2030 compared to 2021 levels, and nine times higher by 2035 [2]. If these predictions hold, they signal not just a continuation of current trends, but an exponential surge. Denmark, with its ambitions for a green economy, thus finds itself in a peculiar and precarious position – hosting infrastructure that could substantially undermine its climate goals if left unchecked.
At the global scale, electricity generation remains one of the largest contributors to carbon emissions worldwide, despite significant advances in renewables. Thus, the operational carbon emissions associated with the rise of data centers – especially for computationally intensive tasks like Generative AI – have similarly swelled in recent years. However, operational emissions tell only part of the story. The environmental burden stretches across the entire lifecycle of these models: from the extraction and processing of silicon, rare earth minerals, and other materials needed to manufacture the specialized hardware; to the construction of sprawling data center facilities; to the ongoing energy demands of cooling systems that are indispensable to maintain operational stability; and finally, to the thorny issue of e-waste management at the end of hardware lifespans. Each of these stages introduces further ecological costs that are often omitted from simplistic accounts of AI’s climate footprint.
Given this broader view, it is increasingly apparent that the development and deployment of Generative AI systems carry a heavy resource footprint. Their climate impact is significant – and yet, frustratingly, remains poorly quantified. The crux of the problem is that tracking the full environmental cost of Generative AI development is extraordinarily difficult. The supply chains are sprawling, opaque, and involve numerous stages that span continents and regulatory regimes. From mining operations in often-overlooked regions, to electricity grids whose carbon intensity varies dramatically across geographies and timescales, the true impact remains buried under layers of complexity.
One of the major blind spots in contemporary discussions about AI and its climate impact is precisely this lack of reliable, standardized data. Bridging this gap will not happen spontaneously. It will require the deliberate development of standardized environmental reporting frameworks, ideally designed in a way that they can be operationalized pragmatically – not just aspirationally. Crucially, this effort must be a collaboration across sectors: researchers who can develop rigorous metrics, industries that can implement measurement and reporting protocols at scale, and governments that can enforce compliance and incentivize best practices.
For countries like Denmark, which have positioned themselves as attractive destinations for data center investments, there is an opportunity – and arguably an obligation – to lead by example. Denmark could set high standards by incentivizing data center operators to adopt best practices, not only in terms of transparent and verifiable environmental reporting, but also in designing operational guidelines that promote energy efficiency, prioritize low-carbon electricity sources, and foster innovation in cooling technologies. If successful, Denmark’s approach could serve as a model for how digital infrastructure growth can be reconciled with genuine climate leadership.
[1] Energy and AI. (2025) https://www.iea.org/reports/energy-and-ai
[2] Denmark’s Climate Status and Outlook 2023 (CSO23)