Data Center Energy Demand Distribution (2023)
Source Research:
Building Greener Data Centers to Minimize AI’s Carbon Footprint
Energy consumption for data centers comes from two main processes—computing and cooling, accounting for around 40% each of total consumption, with the final 20% coming from associated Information Technology (IT) equipment. Integrating new technology to reduce demand on the grid and updated energy efficiency regulation is essential to restrain the surge in energy demand.
AI applications are estimated to use 10% to 20% of the data center electricity today, with AI training becoming a critical concern for data center energy consumption. AI server racks require considerably more energy, cooling, and space compared to traditional racks.
While some AI innovations boost computing speed at a faster rate than energy consumption, the globally increasing use of AI will drive the need for more data centers with higher energy requirements, putting pressure on grid infrastructure, and increasing the operational carbon footprint.
In 2022, data centers consumed 460 TWh of electricity, around 2% of the global usage, according to the IEA. It is expected that this consumption could more than double by 2026, reaching 1,000 TWh, led by growth in AI and the cryptocurrency sector.
At the current rate of energy consumption, ABI Research projects that, by 2030, the projected energy consumption of a data center will reach 2,477 TWH.
Over the next few years, technological innovation and updated regulations, including those on energy efficiency and responsible compute, will be vital to moderating the surge in energy consumption from data centers.
