The rapid acceleration of Artificial Intelligence is driving a critical resource crisis that demands immediate, systemic action, according to a stark new white paper from global digital business and technology services leader, NTT DATA. Titled Sustainable AI for a Greener Tomorrow, the report highlights that AI’s escalating appetite for power, water, and rare materials is on an unsustainable trajectory, potentially undermining corporate and global climate goals. Researchers predict that AI workloads will account for over 50% of data center power consumption by 2028, a surge driven by the computational demands of training massive language models and maintaining continuous inference services.
Beyond the enormous electricity required, the industry’s focus on hardware performance over longevity is exacerbating the twin problems of water usage for cooling and the generation of significant e-waste. This resource drain poses a severe challenge to the positive potential of AI technology.
NTT DATA’s paper issues a global call for action to embed sustainability into every stage of AI development, arguing that resource efficiency must be a core design principle, not an afterthought. David Costa, Head of Sustainability Innovation Headquarters at NTT DATA, emphasized that while the resource consequences are daunting, AI also offers solutions to the very environmental problems it creates, such as optimizing energy grids and improving water conservation. To bridge this “responsibility gap,” the paper outlines several key solutions. The firm urges organizations to expand their focus from conventional performance metrics, such as speed and accuracy, to include holistic sustainability goals. This includes the adoption of standardized, verifiable metrics like the “AI Energy Score” and “Software Carbon Intensity (SCI) for AI” to quantify energy consumption, carbon emissions, and water footprint.
Furthermore, a lifecycle-centric approach is essential, moving from raw material sourcing and hardware production through to system disposal, embracing circular-economy principles, and reducing e-waste by prioritizing modular, upgradable components. The ultimate solution, the paper concludes, requires shared accountability and cross-sector cooperation among hardware manufacturers, data center operators, software developers, and policymakers to redesign the entire AI lifecycle.
