The rapid expansion of AI-driven data centers with growing demands for electricity, water, and critical minerals is creating a significant environmental challenge. In 2023, in the US data centers accounted for 4.4% of electricity usage with projections tripling up to 12% by 2028. 

Beyond electricity, data centers are among the largest industrial water consumers. For instance, training OpenAI GPT-3 model in Microsoft’s advanced US data centers required the evaporation of 700,000 liters of clean freshwater. Looking ahead, global AI demand is projected to account for 4.2 - 6.6 billion cubic meters of water withdrawal by 2027 — more than the total annual water use of Denmark or nearly half of the UK. 

Meanwhile, data centers hardware rely heavily on rare earth elements and critical minerals, which are often mined unsustainably, further exacerbating environmental concerns. The demand for materials like lithium, cobalt, and neodymium is expected to rise dramatically, creating supply chain vulnerabilities and intensifying mining-related environmental damage. 

Addressing these issues through climate-driven innovation presents a major opportunity to reshape the future of AI infrastructure, making it more efficient and resilient. At the same time, AI undoubtedly has the potential to accelerate the development of breakthrough climate technologies — both in mitigation and adaptation efforts. However, there is still no clear understanding of how to fully harness its capabilities for maximum climate impact.