In 2022, the 11,000 data centers worldwide consumed approximately 460 TWh of energy, which is equivalent to the total energy consumption of France.

The International Energy Agency (IEA) reported that data centers consumed between 240 TWh and 340 TWh of power in 2022, constituting 1% to 1.3% of global energy use.

Goldman Sachs Research predicts that AI-related power demand at data centers will grow by around 200 TWh per year between 2023 and 2030, potentially reaching 19% of their energy consumption by 2028.

The energy consumption of data centers is expected to increase by 160% over the next five years, driven in part by the adoption of generative AI.

Generative AI, such as ChatGPT, consumes significantly more energy than traditional search methods, with a single ChatGPT query using approximately 3 Wh compared to 0.3 Wh for a Google search.

Generative AI relies heavily on large language models (LLMs), which consume energy during both training and processing phases, with processing now accounting for 60% to 70% of total power requirements.

While the energy impact of generative AI is currently marginal compared to overall digital technology consumption, it adds to the existing high demand driven by economic growth and other fossil fuel-reliant sectors.

A substantial portion of data stored, known as dark data, contributes to carbon emissions, highlighting the need for digital cleanup efforts to reduce unnecessary energy consumption.

Aging power grids may struggle to support the increased energy loads from data centers, which already account for significant electricity consumption in regions like Ireland and Virginia.

Nvidia currently dominates the AI chip market with a 95% market share, but resource limitations may hinder the production of sufficient chips to meet demand.

The rapid expansion of generative AI may disrupt existing forecasts, as evidenced by China's DeepSeek, which developed a more energy-efficient AI program than those in the U.S.

Quantum computing presents potential for significant energy efficiencies in data centers, where cooling accounts for 40% of electricity use.