Artificial Intelligences Energy Paradox 2025

Page 15 of 28 · WEF_Artificial_Intelligences_Energy_Paradox_2025.pdf

As data centre demand grows, firm, dispatchable generation is needed. Given generation infrastructure challenges and net-zero ambitions, this requires innovation and investment in carbon-free solutions. Renewables, such as wind and solar, support sustainability, but without sufficient energy storage capabilities, their intermittency can cause availability fluctuations, which are troublesome for data centres given their need for consistent, high-demand energy with reliable power quality. Other relevant options include nuclear, geothermal and long-duration storage, as well as harnessing energy sources (such as natural gas and bioenergy) alongside carbon capture and storage (CCS).18 Within nuclear, several companies are following through on the momentum built during COP28 and the declaration to triple global nuclear energy capacity by 2050 by harnessing nuclear solutions to support data centre growth with carbon free energy. This includes a variety of strategies built around the use of both large-scale plants as well as small modular reactors (SMRs), which offer flexible, low-carbon energy with reduced land use and construction time. In the US, for example, Amazon, Google and Microsoft have all explored nuclear options, each with varying benefits and complexities.19 In addition to the intermittency issue, other relevant mismatches exist between data centres and power generation/grid infrastructure that make clean energy procurement challenging. Included in this list is the geographical disconnect between data centre locations and renewable generation sites, which can result in inefficiencies and high transmission costs. Beyond energy, broader understanding and research efforts on natural resource concerns are also critical for achieving sustainable AI. A non-exhaustive list of several enablers that can help address the challenges outlined above is explored in the following section.Global electricity generation by source and scenario, 1990-2050 FIGURE 5 2040 1990 2023 2050 2023 2050 2023 205060STEPS APS NZEAfter decades of fossil fuels generating most of the world’s electricity, renewables are set to become the main pillar of electricity supply Solar PV Wind Hydro Other renewables Nuclear Other low-emissions Other Unabated coal Unabated natural gas80 0Thousand TWh Note: TWh = terawatt-hours; STEPS = Stated Policies Scenario; APS = Announced Pledges Scenario; NZE = Net Zero Emissions by 2050 Scenario. “Other renewables” includes bioenergy, geothermal, concentrating solar power and marine; “Other low-emissions” includes fossil fuels with CCUS, hydrogen and ammonia. “Other” includes non-renewable waste. Source: International Energy Agency (IEA). (2024). World Energy Outlook. Enabling sustainable AI will require a multifaceted approach consisting of actions across four key areas: –Regulatory and policy enablers for establishing policies and frameworks that promote responsible AI development and use, ensuring compliance with environmental standards and energy policies –Financial incentive enablers that can provide funding and investment mechanisms to support sustainable AI initiatives –Technological innovation enablers focused on promoting research and development (R&D) to drive cutting-edge technologies that enhance sustainability in AI applications3.3 Overview of ecosystem enablers Artificial Intelligence’s Energy Paradox: Balancing Challenges and Opportunities 15
Ask AI what this page says about a topic: