Executive summary Artificial intelligence (AI) is facilitating a new era of innovation, with nearly three in four companies using AI for at least one business function.5 This innovation brings many benefits, including enhanced productivity, new ways of working and revenue growth. AI-related electricity consumption is expected to grow by as much as 50% annually from 2023 to 2030. AI data centre consumption, while growing rapidly, is projected to remain a small fraction of global electricity demand, starting at just 0.04% in 2023 (see Figure 4). However, when combined with other market factors (such as growing electricity demand for transport, buildings and more), AI’s accelerated adoption could potentially increase the strain on power grids and electricity providers. However, such projections can vary.6 Uncertainty remains around how profound AI’s overall energy impact will be and which strategies could mitigate challenges that arise or enable new solution opportunities. In this context, it’s essential to assess how AI could accelerate the energy transition in line with net-zero goals, as well as which supporting ecosystem enablers can support this. This paper focuses on AI’s electricity impacts while addressing the broader energy landscape, including generation and fuel sources supporting AI. Work under the AI Governance Alliance (AIGA) AI Energy Impact Initiative has surfaced key insights on these topics. The initiative collaborates with over 40 global organizations across more than nine industries driving AI adoption. This analysis highlights key findings relevant to three distinct areas related to AI’s role in transforming energy systems: 1. Electricity consumption of AI: Reviewing the AI life cycle, strategies for reducing its consumption and new opportunities for process digitalization –AI adoption varies by sector, with electricity demand expected to rise sharply. However, projections remain uncertain, underscoring a need for ongoing assessment. –Optimizing AI’s consumption includes harnessing technological innovations such as energy-efficient AI chip hardware and AI- optimized cooling solutions. –Companies are reducing data centre electricity consumption through operational strategies like AI-driven environmental controls, server virtualization and workload distribution. 2. AI-enabled energy transition: Exploring innovative, emerging company use cases and the potential for scaling across industries –Existing use cases demonstrate reduced energy consumption of 10-60% in some instances, with potential for further optimization. –AI is helping electricity providers optimize operations via energy storage, enhanced battery efficiency and smart grid. –AI can support decarbonization, helping to lower emissions, reduce waste and improve resource use. 3. Primary challenges and ecosystem enablers: Analysing regulation, policy and partnerships necessary for sustainable AI adoption at scale –Enabling sustainable AI requires a multifaceted approach spanning: regulation and policy, financial incentives, technological innovation and market development. –Regulatory, policy and financial enablers can incentivize responsible AI through compliance frameworks and funding mechanisms. –Technological innovation and market development foster research, collaboration and sustainable AI adoption. This white paper is a preliminary exploration of AI’s energy-related impact, and outlines the key challenges and opportunities that emerge as AI adoption grows across industries. It concludes by sharing four areas to monitor for continued understanding of AI’s evolving energy impact: –AI deployment for decarbonization –Transparent and efficient AI electricity use –Innovation in technology and design –Effective ecosystem collaborationArtificial intelligence presents energy opportunities and challenges – strategic mitigation can help to maximize benefits while reducing burdens. Artificial Intelligence’s Energy Paradox: Balancing Challenges and Opportunities 5

Artificial Intelligences Energy Paradox 2025