Until recently, the rapid growth of the digital economy had relatively few implications for global energy demand and carbon emissions. But the landscape has shifted dramatically with the explosive rise of generative Artificial Intelligence (AI), accelerated by the launch of ChatGPT in October 2022. This surge is driving a sharp increase in demand for data centres, reshaping energy systems around the world and increasing greenhouse gas emissions.

In theory, the sector’s power demands could be 100% renewably-sourced (or reasonably close). In practice, however, the rapid growth is being met largely by fossil fuels. Natural gas and even coal are often seen as the cheapest and most reliable energy sources available for data centres. In the United States, for example, 60% of additional electricity demand is currently being met by natural gas. In Germany and Poland, the growing number of data centres is extending the life of coal-powered electricity generation.

So far, governments and regulators have largely allowed Big Tech and data centre operators to set their own voluntary decarbonisation targets. But with AI adoption surging, and its future impact only set to grow, a hands-off approach is no longer sufficient. Regulation is urgently needed to reduce emissions and maintain the stability of energy systems.

The environmental justice implications are also stark. Data centre profits overwhelmingly flow to a small number of companies typically in the Global North, while local populations suffer from their high resource demands for energy and water – and see few benefits.

Our policy briefing outlines the key challenges associated with AI-driven data centre growth and sets out initial recommendations for how governments can respond.

Read the full policy briefing below.