Energy Consumption of AI-Powered Chatbots: A Closer Look at ChatGPT and Deepseek

Energy Consumption of AI-Powered Chatbots: A Closer Look at ChatGPT and Deepseek

Understanding the Impact of Data Centres on Electricity Demand

Data centres are becoming significant contributors to electricity consumption worldwide. A report indicates that nearly one-third of Ireland’s electricity demand could arise from data centres by 2026. This shows how crucial these facilities are becoming in our energy landscape.

The Energy Use of AI and Data Centres

Artificial Intelligence (AI) applications, which heavily rely on data centres, are increasing electricity demand. For instance, an inquiry made through ChatGPT is estimated to consume approximately 10 times more electricity than a standard Google search. More specifically, a single request to ChatGPT uses about 2.9 watt-hours, while a typical Google query consumes around 0.3 watt-hours.

With around 9 billion searches happening daily, this translates to an additional electricity requirement of nearly 10 terawatt-hours (TWh) each year. The growing AI industry not only leans on data centres for training and operating its models but also contributes to the global emissions of greenhouse gases. Notably, Microsoft reported a rise of nearly 30% in its CO2 emissions since 2020 due to data centre expansion, and Google’s emissions in 2023 were nearly 50% higher than those in 2019, predominantly due to the energy needs tied to data centres.

Although energy consumption by AI accounts for only 2-3% of total global emissions, experts predict that this figure will climb as more organizations harness AI for efficiencies and productive output.

Data Centre Locations in the EU

Globally, there are over 8,000 data centres, with around 16% situated in Europe. A majority of these facilities are located in major financial hubs, including Frankfurt, London, Amsterdam, Paris, and Dublin. According to the International Energy Agency (IEA), the electricity consumption of data centres in the European Union is expected to reach nearly 150 TWh by 2026.

Specifically, in Ireland, the data centre sector consumed 5.3 TWh of electricity in 2022. This figure comprises 17% of the nation’s overall electricity use, equivalent to what all urban residential buildings consume. If the trend of expanding AI applications continues, data centres could be responsible for up to 32% of Ireland’s total electricity demand by the year 2026.

Data Centre Consumption by Country

Denmark hosts about 34 data centres, with half located in Copenhagen. Just like Ireland, Denmark is expecting its electricity need to surge mainly due to the growth of data centres, with an anticipated consumption of 6 TWh, accounting for close to 20% of its total electricity demand by 2026.

The Nordic countries, including Sweden, Norway, and Finland, are also players in the data centre landscape. They benefit from reduced electricity costs, particularly because their cooler climates lower the cooling demand. Sweden, the largest player in this region, has about 60 data centres, with half of them in Stockholm.

Both Sweden and Norway are poised to expand their involvement in the data centre industry due to their commitment to decarbonisation. Nearly all of their electricity is generated from low-carbon sources, aligning with global sustainability targets.

Summary of Key Data Centre Insights

  • Electricity Demand: Data centres are expected to control significant portions of electricity demand in various countries.
  • AI Impact: AI applications are consuming more electricity, intensifying the reliance on data centres.
  • Geographical Concentration: Major data centres are primarily found in global financial hubs and are growing rapidly in EU countries.
  • Future Predictions: Electricity consumption from data centres is set to rise sharply, prompting conversations about sustainability and energy efficiency.

By understanding these factors, we can grasp the complexities of energy consumption associated with data centres and AI, paving the way for informed discussions around sustainability and efficiency.

Please follow and like us:

Related