A new white paper from NTT DATA, titled “Sustainable AI for a Greener Tomorrow,” warns that artificial intelligence’s rapid expansion is placing unprecedented strain on the planet’s natural and digital resources. The report — released simultaneously in Tokyo and London — highlights that AI workloads are projected to consume over 50% of global data center power by 2028, driven by massive model training, inference, and deployment demands.
Beyond power consumption, researchers warn of the ripple effects: escalating water usage for cooling systems, rising e-waste from hardware turnover, and growing reliance on rare-earth minerals used in GPUs and semiconductors. “The resource consequences of AI’s rapid growth are daunting,” said David Costa, head of sustainability innovation at NTT DATA. “But if we act now, AI itself can be part of the solution.”
Balancing AI Innovation with Environmental Stewardship
The white paper calls for a paradigm shift in how organizations measure AI performance — urging them to move beyond accuracy and speed to include efficiency and sustainability as core design principles. It advocates introducing global standards such as an “AI Energy Score” and “Software Carbon Intensity for AI” to quantify energy, carbon, and water impacts.
NTT DATA researchers emphasize that most enterprises currently overlook critical aspects of AI’s environmental cost. While carbon emissions receive attention, few companies measure or mitigate water use, resource depletion, or hardware waste. The report recommends that sustainability become an integral part of AI system design — from hardware manufacturing to deployment and eventual recycling.
A Lifecycle-Centric Model for Sustainable AI
NTT DATA proposes a lifecycle-centric approach, embedding sustainability across the entire AI value chain. This includes extending hardware lifespans, optimizing data center cooling, leveraging renewable energy scheduling, and promoting modular, upgradable components to reduce e-waste.
The report also advocates green software engineering practices, encouraging developers to design AI models that dynamically align workloads with periods of renewable energy availability. The goal is to reduce total environmental impact while maintaining computational performance.
“AI can be a powerful force for good,” Costa said. “It can model climate risks, optimize energy use, and drive sustainable urban planning. But it must first clean up its own footprint.”
A Global Call to Action
NTT DATA’s findings echo growing concerns among policymakers and sustainability experts that the AI boom could undermine net-zero commitments if left unchecked. The paper urges international collaboration to develop verifiable sustainability metrics, accelerate renewable integration into data centers, and promote circular-economy models for AI infrastructure.
With the AI race accelerating globally, “Sustainable AI for a Greener Tomorrow” serves as a timely reminder that the future of intelligence — both artificial and human — depends on our ability to balance innovation with responsibility.
