AI Surge to Double Data Centre Electricity Demand by 2030

AI Surge

According to the International Energy Agency (IEA), electricity consumption by data centres is projected to more than double by 2030, primarily due to the rising AI surge in applications, software, and programs. This surge in usage presents fresh challenges for both global energy security and carbon emission reduction targets. However, the AI surge holds potential for optimizing electricity production and consumption, as noted in the IEA’s inaugural report focused on AI’s energy impact.

• Goldman Sachs Research currently estimates the global data centre sector's power usage at approximately 55 gigawatts (GW). This total breaks down into 54% used for cloud computing tasks, 32% dedicated to conventional business operations like email and data storage, and 14% consumed by AI workloads.

• In 2024, data centres, due to the AI surge, accounted for about 1.5% of the world's electricity consumption. This figure has seen an annual increase of 12% over the past five years. The rapid rise is largely fuelled by generative AI technologies, which demand immense computational resources to process vast amounts of data.
• Looking ahead, the IEA forecasts data centre electricity consumption to hit nearly 945 terawatt hours (TWh) by 2030—a volume that slightly exceeds Japan’s current electricity usage. The report emphasizes that AI is the primary force behind this increase, in addition to expanding demand for other digital services.
• Currently, around 30% of the energy powering data centres comes from coal. However, more cost-effective and widely available alternatives—such as renewable energy and natural gas—are expected to take on a greater share soon.
• With this continued growth, emissions associated with data centre electricity use are expected to climb from 180 million tonnes of CO₂ today to 300 million tonnes by 2035. Still, this remains a relatively small portion of the 41.6 billion tonnes of global emissions projected for 2024.
• Data centres designed to deal with the AI surge are also emerging as a new category of infrastructure. Although only a few are operational today, they are engineered to accommodate the unique demands of AI, such as higher total power loads, denser rack configurations, and specialized components like liquid cooling systems. These centres are typically developed by hyperscale cloud providers or wholesale data centre operators.