AI's Impact: Designing for a Greener Future

A single ChatGPT query uses 10 times the energy of a Google search, and in the US alone, data centres are expected to consume enough power by 2030 to run 26 million homes. 

With temperatures projected to exceed 2.5°C warming, we need to ask: how do we ensure AI’s growth doesn’t come at the expense of accelerating the planet’s uninhabitability? 

AI can drive positive change in climate action and sustainability, but we need to design its infrastructure to reduce its own impact. Some companies are already experimenting with repurposing waste heat and exploring water conservation, but there’s still lots more to do. What if we could design AI infrastructure to function like permaculture, where byproducts serve multiple purposes and create a more sustainable system? 

AI has the potential to accelerate solutions for sustainable development, yet it also carries significant socio-environmental consequences.

As we navigate this pivotal moment, we have an opportunity to shape AI’s trajectory as a driver of positive, sustainable innovation. 

Driving the Green Transition 

Time is critical in the fight against climate change, and AI is an accelerator—of everything. This makes it a double-edged sword for sustainable development. It is precisely for this reason that we must be strategic, rapidly assessing impacts, minimising harm, and implementing effective solutions. AI is already contributing to sustainability in several ways: 

Advancing climate research and impact analysis – AI can process vast datasets at unprecedented speeds, improving climate modelling and ESG impact assessments. Google’s Earth Engine, for example, integrates satellite imagery with geospatial data, making forest cover analysis more accessible. Similarly, machine learning enhances the organisation of unstructured earth data, helping to localise global climate models and generate actionable insights. 

• Enhancing disaster preparedness and response – AI improves the accuracy of climate-related disaster predictions, strengthening early warning systems. DeepMind’s weather prediction model, for instance, delivers rapid, highly precise forecasts, helping authorities prepare for floods, hurricanes, and extreme weather—essential for climate adaptation strategies. 
• Accelerating scenario modelling for adaptation and mitigation – AI significantly reduces the time needed for scenario modelling, aiding policymakers in strategic decision-making. It has already been applied to simulate the impact of rising sea levels on housing, providing communities with the insights needed to mitigate risks. 
• Streamlining ESG reporting and optimising resources – ESG reporting is often complex and time-consuming, delaying action. Increasingly, technology start-ups are leveraging AI to simplify data collection, management, and analysis—enabling businesses to focus on tangible action rather than lengthy assessments. 
• Driving innovation in materials science – AI is revolutionising materials science by predicting properties and configurations without exhaustive experimentation. This has accelerated the development of climate-resilient materials, such as advanced battery technologies and fire-resistant compounds, which support both climate mitigation and adaptation efforts. 

The Environmental Cost of AI 

While AI’s potential is vast, its environmental footprint is equally significant. The infrastructure powering AI—considered one of the largest technological build-outs in human history—demands enormous resources, including: 

• Energy consumption – Data centres, the backbone of AI operations, already account for 2% of global electricity use. A single ChatGPT query consumes ten times the energy of a Google search. In the United States alone, data centre energy consumption could reach 35 gigawatts by 2030—enough to power 26 million homes. 
• Water usage – AI infrastructure requires millions of litres of water for cooling. A medium-sized data centre (15 MW) consumes as much water as three hospitals or more than two 18-hole golf courses. Material demands – The production of silicon chips and cooling systems drives significant energy and water usage, further intensifying resource demands. 
• Localised resource strain – The impact of AI is particularly concerning in regions already facing resource scarcity. In the United States, one-fifth of all data centres are located in water-stressed areas, while parts of Taiwan and Arizona face mounting pressures on energy and water supplies.

This is particularly stressful as we're no longer able to remain below 1.5 degrees global warming, and neither are we at 2 degrees. This means we're heading towards a 2.5-degree warming, which means widespread ecosystem collapse, more frequent and intense extreme weather events, and catastrophic impacts on food security, sea levels, and human health. The fact that billions of dollars have been invested in AI—and AI means more data centres, which involve an acceleration of emissions—you can understand how this is quite frightening. 

A Smarter Approach to Sustainability 

Despite the challenges, some data centres have made meaningful strides in improving sustainability. Many are reducing their water and energy consumption through targeted interventions—using reclaimed and recycled water in Texas, Washington, California, and Singapore, and implementing rainwater harvesting in the Netherlands, Ireland, and Sweden. 

Beyond water conservation, some have taken an even more innovative approach. In Denmark, excess heat from data centres is repurposed to warm 6,000 homes—transforming a byproduct into a valuable resource rather than expending additional energy to cool infrastructure while simultaneously building separate heating systems. 

These efforts set a promising precedent, but they remain the exception rather than the norm. Just as health and safety protocols are now fundamental across industries, sustainability must become a standard, non-negotiable practice. 

Priorities for 2025

Whether AI becomes a positive or negative force depends on the ecosystem we create for it. These are the essential priorities for 2025:  

• Establish regulatory frameworks: Governments, businesses, and civil society must collaborate to create comprehensive policies that align AI development with ESG principles. These frameworks should address sustainability, equity, data privacy, and ethical governance.  
• Make AI purpose driven: Ensure AI applications are prioritised for sustainable purposes, leveraging their capabilities to address pressing environmental and social challenges.   
• Drive operational sustainability: Evaluate and enhance infrastructure efficiency. For example, Microsoft’s 2022 audit identified improvements that eliminated 90% of excess water use.   
• Design sustainable infrastructure: Design data centres that are sustainable from the start, situating them in resource-appropriate locations and incorporating green technologies to minimise energy and water demands.