The increasing demand for artificial intelligence (AI) is expected to significantly impact global energy consumption due to several factors:

Massive Energy Requirements: AI data centers, which are pivotal for processing and storing AI data, are expected to double their electricity demand by 2026. In 2022, sectors like AI and cryptocurrencies together consumed about 460 terawatt hours (TWh) of electricity globally, which is around 2% of total global demand​ (The Street)​.
Comparison to National Energy Consumption: The electricity demands of AI are growing to the point where they could soon be comparable to the energy consumption of entire countries. For instance, if every Google search used AI, the energy required would be equivalent to the annual electricity consumption of Ireland​ (euronews)​.
Growth in Server Demand: The number of AI servers shipped is expected to rise dramatically, with projections indicating at least a tenfold increase in energy consumption by these servers between 2023 and 2026. In 2023 alone, Nvidia shipped 100,000 AI servers which consumed an average of 7.3 TWh annually​ (The Street)​.
Operational Energy Intensity: High-performance AI models like those used for large-scale language processing or generative tasks consume substantial amounts of energy per operation. For example, a single ChatGPT query uses significantly more electricity than a typical Google search, and widespread usage could lead to substantial increases in power demand​ (The Street)​.
Future Projections: It is estimated that worldwide AI-related electricity consumption could increase by 85 to 134 TWh annually by 2027, based on projections for AI server production. This level of consumption would be on par with the annual electricity needs of countries such as the Netherlands, Argentina, or Sweden​ (euronews)​.

source