Is technology truly the solution to the environmental crisis? Digital innovation is often proclaimed by policymakers and industry alike as the ultimate catalyst for environmental progress: the promise that AI, big data, and smart devices will unlock new efficiencies and drive us toward a greener future.
Yet beneath this optimism lies an uncomfortable tension. Every data centre humming with servers, every smart device deployed in the name of efficiency, and every algorithm trained to optimise our systems demands resources of its own (energy, water, and rare earth minerals). Data centres already consumed around 415 TWh of electricity in 2024, roughly 1.5% of global demand, a figure the IEA projects will more than double to about 945 TWh by 2030. The question is not whether technology can help us achieve sustainability, but whether our solutions are quietly creating new problems.
The irresistible business case
For businesses, digital tools offer a compelling value proposition. They improve resource management, identify waste, and optimise consumption. For example, a smart building system can slash energy use by adjusting heating and lighting in real time. Technology also addresses the growing burden of compliance. As sustainability reporting requirements multiply globally, companies face an avalanche of transparency obligations. Digital tools streamline data collection and automate reporting.
The result is a compelling business case for accelerated sustainable business models, streamlined decision-making, and simplified compliance.
The critics have a point
Yet dissenting voices urge caution. Movements advocating for ‘low tech’ approaches challenge the premise that more technology automatically equals more sustainability. The central argument is provocative: what if technology, rather than solving our problems, merely replicates the unsustainable structures it claims to transform?
The lifecycle of digital devices illustrates the first issue. Smartphones and servers require rare earth minerals, often extracted under problematic conditions. They consume significant energy in manufacturing, operation, and disposal. A second problem is the ‘rebound effect’: when greater efficiency makes a resource cheaper to use, total consumption often rises rather than falls, offsetting the expected savings.
Generative Artificial Intelligence (AI) systems represent this tension. Training large models requires enormous computational resources. Data centres have become temperature-controlled behemoths housing tens of thousands of servers.
The numbers are staggering. Global electricity consumption by data centres reached around 415 TWh in 2024 (approximately 1.5% of total global consumption). By 2030, this is projected to approach 945 TWh.
The path between extremes
Must we choose between uncritical faith in technology and a wholesale rejection of digital innovation?
Policymakers, businesses, and technology designers should refuse both extremes. Rather than treating technology as a catch-all answer to broad, hard-to-measure goals such as ‘sustainability’ in abstract, they should deploy it as a precision instrument for specific, well-defined and measurable objectives.
This approach rests on two foundations. First, reorganising technological design around social values. As Andrew Feenberg has argued, technology is never neutral as it inevitably embeds the values and priorities of its designers. If we want technology to serve sustainability, we must intentionally design it with sustainability as a core value. This means asking difficult questions during the design phase like ‘who benefits from this innovation?’ or ‘what resources does it consume?’. This shifts the focus beyond energy consumption to a further dimension: the question of power, as technology is not a neutral instrument; its design encodes choices about who benefits, whose interests are served, and who bears the costs. Second, clearly establishing who and what each innovation seeks to address. Generic ‘sustainable technology’ initiatives often lack the specificity needed for meaningful impact.
Towards responsible innovation
The sustainable technology paradox is real, but it is not insurmountable. Digital innovation remains an indispensable tool for addressing our most pressing environmental and social challenges. However, its value depends entirely on how thoughtfully we deploy it, and regulators are increasingly ensuring that organisations do precisely that.
A rapidly evolving legal framework is emerging, particularly in the European Union (EU), that seeks to balance the promise of digital innovation with its environmental costs. The revised Energy Efficiency Directive mandates monitoring and reporting of data centre energy performance. The EU’s Artificial Intelligence (AI) legislation, the AI Act, requires standards for improving energy efficiency during the lifecycle of these systems. Companies within scope of the Corporate Sustainability Reporting Directive must report on climate impacts across their entire supply chain.
This regulatory landscape continues to develop. The European Commission is preparing a Data Centre Energy Efficiency Package, which will introduce a rating scheme for data centres and launch work on minimum performance standards. In this context, the Cloud and AI Development Act is expected to facilitate the tripling of EU data centre processing capacity in the next five to seven years, but only if facilities comply with stringent requirements on energy efficiency, water efficiency, and circularity.
The path we have travelled must be evaluated. The relentless pursuit of innovation in the name of efficiency has brought us to the very edge of resource exhaustion. To address the environmental footprint of technology, we are constructing an increasingly dense web of regulatory obligations, which, when taken together, create a compliance burden of considerable complexity: one that may ultimately stifle the very innovation we need to solve the problems we have created. The problem is one of cumulative overlap rather than any single rule. A data centre operator may simultaneously have to monitor and report energy performance under the Energy Efficiency Directive; satisfy the energy-reporting and efficiency obligations imposed on general-purpose AI models by the AI Act; and, if within scope, report climate impacts across its entire value chain under the Corporate Sustainability Reporting Directive. This is the second paradox of sustainable technology. In seeking to constrain harmful innovation, there is a risk that innovation could be constrained. The costs of compliance, the uncertainty of evolving standards, and the spectre of litigation exposure may deter precisely the transformative solutions that could reconcile technological progress with planetary limits. The EU has itself acknowledged this burden.
Overall, the right balance must be found. While regulation without flexibility becomes an obstacle, innovation without accountability becomes reckless. The challenge for policymakers, legal practitioners, and technologists alike is to craft a framework that demands responsibility without demanding paralysis, one that channels innovation toward genuine sustainability rather than mere compliance, and that recognises the difference between technology that serves human concerns and technology that merely serves itself.
**************************
This blog post is based on the paper to be presented at the conference Sustainability Law: Law’s Response to the Convergence of Crises, 16-18 September 2026, at the University of Oslo.
