Powering Intelligence: How Energy and AI Are Transforming Each Other
The International Energy Agency’s landmark report “Energy and AI” presents a sweeping and detailed global analysis of how artificial intelligence (AI) and the energy sector are becoming deeply intertwined. As AI becomes a general-purpose technology comparable to electricity itself, the energy implications are immense and far-reaching. The report, led by Thomas Spencer and Siddharth Singh under the direction of Laura Cozzi, offers the first comprehensive study on how AI is both consuming energy at unprecedented levels and transforming how energy is produced, managed, and optimized.
One of the central findings of the report is the soaring energy demand of AI, particularly through the rise of AI-dedicated data centers. As of 2024, these centers already consume roughly 415 terawatt-hours of electricity, accounting for 1.5% of global electricity consumption. With AI applications expanding rapidly, this demand is expected to more than double by 2030 and could reach 1,200 TWh by 2035 in the IEA’s Base Case scenario. This would rival the current electricity usage of entire industrial sectors. The United States, China, and Europe lead in data center deployment, with American centers projected to consume more electricity than the country’s entire manufacturing sector by decade’s end.
The report emphasizes that meeting this rising demand will require a mix of energy sources. Renewable energy will play a dominant role, contributing more than half of the growth in electricity generation for AI, while natural gas, nuclear energy (including small modular reactors), and advanced geothermal systems will also be crucial. Importantly, the infrastructure must keep pace: grid bottlenecks and equipment shortages threaten to delay up to 20% of planned data centers, especially in high-density locations like Northern Virginia.
Beyond its energy consumption, AI is already revolutionizing the energy sector itself. In oil and gas, AI tools are being used for predictive maintenance, leak detection, and methane monitoring. In electricity systems, AI supports grid balancing, enhances forecasting for renewables, and can unlock up to 175 GW of capacity through smarter management—without building new transmission lines. In industries, AI improves energy efficiency, productivity, and innovation, potentially cutting emissions equivalent to Mexico’s annual energy use. In transport and buildings, AI could save energy equivalent to that used by 120 million cars and deliver annual electricity savings on par with Australia's total generation, respectively.
Innovation is another core theme. AI is emerging as a critical accelerator for energy technology R&D, from battery chemistries to carbon capture materials. Yet, only a small share of energy sector venture capital is currently directed toward AI-enabled innovation. To bridge this gap, the report calls for policy measures that promote AI in science and commercialization.
The report doesn’t shy away from risks. The energy requirements of data centers could strain grids, delay electrification, and potentially shift energy priorities. Supply chains for AI-critical components like gallium are highly concentrated, raising geopolitical and supply security concerns. Moreover, while AI can reduce emissions through efficiency gains, the rebound effect—such as increased personal mobility due to autonomous vehicles—could offset these benefits.
In emerging economies, the challenges are different. These countries host half the world’s internet users but less than 10% of global data center capacity. With the right infrastructure, these regions could leapfrog into AI-powered development. But power reliability remains a major barrier, and inclusive policies are needed to close the digital divide.
Ultimately, the IEA concludes that the global energy and AI sectors are on a converging path. Policymakers, technologists, and energy leaders must work together to manage risks, accelerate beneficial innovation, and ensure that the growing energy needs of AI do not come at the cost of climate and energy security goals. The authors argue that uncertainty should not prevent action. Rather, collaboration and robust data should be used to shape a balanced, secure, and sustainable energy-AI future.
Disclaimer: This blog post is a summary of the International Energy Agency’s report titled “Energy and AI.” While efforts have been made to ensure accuracy, this post is not guaranteed to be free of errors and does not constitute legal or professional advice.
