Artificial Intelligences Energy Paradox 2025

Primary challenges and ecosystem enablers3 As AI demand grows, companies developing data centres may face challenges in sourcing sufficient volumes of power, particularly from carbon-free generation sources. New, large-scale data centres often have significant power requirements, which, depending on the target build jurisdiction, may require costly, time-consuming new infrastructure solutions. Integrating renewable energy is also challenging due to variability and storage issues, which hinder consistent power delivery. Transmission additionally brings complexity as high-voltage lines are near capacity in some regions. In the US, some utility companies have even halted new service requests or begun rationing power.13 Hyperscalers are exploring opportunities to use renewables to power data centres, an option that may not be feasible for some industry players. According to industry feedback, an increased interest has been observed from oil and gas players meaning to invest in renewables to meet the demand for clean energy and growing AI power needs. Given infrastructure upgrade needs, financing concerns have also arisen as stakeholders debate fair cost allocation across customer groups to support data centre expansion.14,15 Accordingly, utility companies face significant challenges in designing customer rate pricing structures that can support this growth while balancing factors like fairness, affordability and sustainability. Grid impacts must also be analysed, as managing data centre loads may require advanced demand response and load balancing. Regulatory, environmental and supply chain issues, such as delays in key grid assets, can also extend approval and construction timelines.3.1 Infrastructure challengesBalancing AI’s potential with its growing energy needs will require multistakeholder collaboration and scalable solutions to challenges. Multistakeholder collaboration is needed to address challenges across industries and enable sustainable AI. To balance AI’s transformational value with costs and negative impacts, two key elements must be addressed: infrastructure and environmental challenges. Another fundamental challenge in expanding AI solutions is addressing their environmental impact. All energy, even clean energy, has an environmental impact.16 As AI becomes increasingly integral across varied aspects of life, it’s crucial to consider energy scarcity as a key design principle (rather than assuming unlimited resources) when developing AI’s future infrastructure. This approach will help ensure AI supports the energy transition.To facilitate sustainable AI, it will be crucial to maintain a balance between optimizing the speed of progress on market goals and prioritizing net- zero emissions targets or 24/717 carbon-free energy targets. Growing global data centre demand, alongside other emerging energy market factors, could potentially leave a gap between forecasted emissions in 2050 and net-zero targets.3.2 Environmental challenges Given infrastructure upgrade needs, financing concerns have arisen as stakeholders debate fair cost allocation across customer groups to support data centre expansion. Artificial Intelligence’s Energy Paradox: Balancing Challenges and Opportunities 14