The expansion of AI is significantly increasing the energy demands of data centres. Major tech companies like Microsoft and Google have reported substantial rises in their electricity consumption, with Microsoft's usage tripling since 2020 and Google experiencing a 48% increase in greenhouse gas emissions since 2019. The International Energy Agency predicts that global data centre power consumption could exceed 1,000 TWh annually by 2026, which is comparable to Japan's total electricity use.

Yes, AI can serve as a powerful tool for enhancing sustainability in data centres. For instance, Google's AI-powered efficiency recommendation system has led to a 40% reduction in energy used for cooling. Additionally, Amazon's AWS claims to be up to 4.1 times more energy efficient than traditional infrastructure, and its AI tools can reduce a workload's carbon footprint by up to 99%. These advancements illustrate that while AI increases energy demands, it also provides solutions to mitigate its environmental impact.

Companies are exploring various strategies to maintain sustainability amidst growing AI workloads. This includes developing custom chips for improved computational efficiency, implementing advanced cooling systems, and utilizing renewable energy sources. For example, Meta has matched 100% of its data centre electricity use with renewable energy since 2020 and is considering nuclear power to meet future demands. The challenge lies in deploying these innovations quickly and transparently to keep pace with the rapid growth of AI infrastructure.