There has been a step change in the capabilities of artificial intelligence (AI), driven by falling computation costs, a surge in data availability and technical breakthroughs. AI is the science of making machines capable of learning to perform tasks that traditionally required human intelligence. AI is emerging as a general-purpose technology, much like electricity. Today, it can generate text and videos, accelerate scientific discovery in fields like medicine or materials science, make manufacturing robots smarter and more productive, drive commercial taxis in complex city landscapes, and detect threats to critical infrastructure.

In the past few years, AI has gone from an academic pursuit to an industry with trillions of dollars of market capitalisation and venture capital at stake. The market capitalisation of AI-related firms in the S&P 500 has grown by around USD 12 trillion since 2022. While there are several uncertainties about its uptake and impact, AI’s rapid development and huge potential have made it central to corporate strategies, economic policies and geopolitics.

However, there is no AI without energy; at the same time, AI has the potential to transform the energy sector. Affordable, reliable and sustainable electricity supply will be a crucial determinant of AI development, and countries that can deliver the energy needed at speed and scale will be best placed to benefit. Training and deploying AI models takes place in large and power-hungry data centres. A typical AI-focused data centre consumes as much electricity as 100 000 households, but the largest ones under construction today will consume 20 times as much.

The energy sector is therefore at the heart of one of the most important technological revolutions today. However, there is still a lack of understanding of the stakes and implications of this deepening connection between energy and AI. Consistent with its strong track record of identifying and exploring emerging issues in the energy sector, this new International Energy Agency (IEA) special report seeks to fill this gap with the most comprehensive, data-driven analysis on the topic to date. Based on a new global model and comprehensive dataset of data centre electricity demand, its analysis was also enriched by an in-depth process of consultation with policy makers, the tech sector, the energy industry and other experts.

Global investment in data centres has nearly doubled since 2022 and amounted to half a trillion dollars in 2024. This investment boom has led to growing concerns about skyrocketing electricity demand.

Data centres accounted for around 1.5% of the world’s electricity consumption in 2024, or 415 terawatt-hours (TWh). The United States accounted for the largest share of global data centre electricity consumption in 2024 (45%), followed by China (25%) and Europe (15%). Globally, data centre electricity consumption has grown by around 12% per year since 2017, more than four times faster than the rate of total electricity consumption. AI-focused data centres can draw as much electricity as power-intensive factories such as aluminium smelters, but they are much more geographically concentrated. Nearly half of data centre capacity in the United States is in five regional clusters. The sector accounts for substantial shares of electricity consumption in local markets.

Data centre electricity consumption is set to more than double to around 945 TWh by 2030. This is slightly more than Japan’s total electricity consumption today. AI is the most important driver of this growth, alongside growing demand for other digital services. The United States accounts for by far the largest share of this projected increase, followed by China. In the United States, data centres account for nearly half of electricity demand growth between now and 2030. By the end of the decade, the country is set to consume more electricity for data centres than for the production of aluminium, steel, cement, chemicals and all other energy-intensive goods combined. Uncertainties widen further after 2030, but our Base Case sees global data centre electricity consumption rising to around 1 200 TWh by 2035.

Renewables and natural gas take the lead in meeting data centre electricity demand, but a range of sources are poised to contribute. Half of the global growth in data centre demand is met by renewables, supported by storage and the broader electricity grid. Renewables generation is projected to grow by over 450 TWh to meet data centre demand to 2035, building on short lead times, economic competitiveness and the procurement strategies of tech companies. Dispatchable sources, led by natural gas, also have a crucial role to play, with the tech sector helping to bring forward new nuclear and geothermal technologies as well. Natural gas expands by 175 TWh to meet growing data centre demand, notably in the United States. Nuclear contributes about the same amount of additional generation to meet data centre demand, notably in China, Japan and the United States. The first small modular reactors come online around 2030.

Data centres account for around one-tenth of global electricity demand growth to 2030, less than the share from industrial motors, air conditioning in homes and offices, or electric vehicles. However, the significance of data centres in driving electricity demand differs by country. Emerging and developing economies are already experiencing rapid electricity demand growth. In these countries, data centres account for around 5% of the increase in electricity demand to 2030. Advanced economies, on the other hand, have seen several decades of essentially stagnant electricity demand. In this group of countries, data centres account for more than 20% of demand growth to 2030, presenting a wake-up call on the need to put the electricity sector on a growth footing again.

Electricity grids are already under strain in many places: we estimate that unless these risks are addressed, around 20% of planned data centre projects could be at risk of delays. Grid connection queues for both supply and consumption projects, including data centres, are long and complex. Building new transmission lines can take four to eight years in advanced economies and wait times for critical grid components such as transformers and cables have doubled in the past three years. Generation equipment is also in high demand. Turbine deliveries for new gas-fired power plants now face lead times of several years, potentially delaying their commissioning beyond 2030. If the electricity sector does not step up, there is a risk that meeting data centre load growth could entail trade-offs with other goals such as electrification, manufacturing growth or affordability.

Key options to mitigate these risks include locating new data centres in areas of high power and grid availability, and operating either data centre servers or their onsite power generation and storage assets more flexibly. These strategies are still underexplored. An AI- focused data centre is 10 times more capital-intensive than an aluminium smelter, which means curtailing its operations to provide flexibility to the grid is very costly. But many data centres operate with a buffer of spare server capacity. Regulators could explore measures to incentivise data centre operators to use spare server capacity or their backup power generation or storage assets more flexibly. Grid operators could also examine incentives to locate data centres in areas where grids are less constrained. We find that 50% of data centres under development in the United States are in pre-existing large clusters, potentially raising risks of local bottlenecks.

There are uncertainties in how quickly AI will be adopted, how capable and productive it will become, how fast efficiency improvements will occur, and whether bottlenecks in the energy sector can be resolved. These uncertainties are explored in sensitivity cases. A Lift- Off Case assumes higher rates of AI uptake and proactive action to reduce energy sector bottlenecks. A Headwinds Case incorporates bottlenecks – including macroeconomic headwinds – in the uptake of AI and the buildout of energy infrastructure to power it. Our High Efficiency Case highlights the potential for even stronger gains in the efficiency of AI- related hardware and AI models. In this case, electricity demand from data centres is 20% lower in 2035 than in the Base Case. By 2035, the range of data centre electricity demand across our cases spans from 700 to 1 700 TWh. The increase in gas-fired power to meet data centre demand in our Lift-Off Case is four times higher than in our Headwinds Case. Growth in nuclear output to meet data centre demand varies even more.

AI is already being deployed by energy companies to transform and optimise energy and mineral supply, electricity generation and transmission, and energy consumption. There are numerous objectives in play, including reducing costs, enhancing supply, extending asset lifetimes, reducing downtime and lowering emissions.

The oil and gas industry has been an early adopter of AI, using it to optimise exploration, production, maintenance and safety. In exploration and development, AI can make the evaluation of resources more reliable and reduce predrilling uncertainty. In operations, it is being used to optimise and automate production processes, detect leaks, predict maintenance needs, and support efforts to reduce methane emissions.

AI can help to balance electricity networks that are growing more complex, decentralised and digitalised. AI can improve the forecasting and integration of variable renewable energy generation, reducing curtailment and emissions. AI-based fault detection can help rapidly identify and precisely pinpoint grid faults, reducing outage durations by 30-50%. Remote sensors and AI-based management can increase the capacity of transmission lines. Up to 175 gigawatts (GW) of transmission capacity could be unlocked if these tools are applied, without any new lines being built. This is more than the increase in the data centre power load to 2030 in the Base Case.

The industry of the future will be increasingly digitalised and automated; countries and companies that take the lead in integrating AI into manufacturing will jump ahead. AI applications can accelerate product development, lower costs and increase quality. Widespread adoption of existing AI applications to optimise processes in industry can lead to energy savings equivalent to more than the total energy consumption of Mexico today. European companies have over half of the market share for industrial automation solutions, which are the critical enabler for industrial AI deployment.

AI applications in transport can improve efficiency and save costs, but they could also increase demand for personal mobility. AI applications are being used to manage traffic, optimise routes, predict maintenance needs and develop autonomous vehicles. The widespread adoption of AI applications across the transport sector could lead to energy savings equivalent to the energy used by 120 million cars. While autonomous vehicles operate more efficiently than conventional ones, they might also attract people away from public transport as costs fall and availability increases, leading to rebound effects.

In buildings, there is significant potential for AI-led optimisations to make heating and cooling systems more efficient and electricity use in buildings more flexible. Barriers to realising this potential include fragmented ownership of buildings, lack of digitalisation and inadequate incentives. If scaled up, existing AI-led interventions could lead to global electricity savings of around 300 TWh, equivalent to annual electricity generation today for Australia and New Zealand combined.

AI is emerging as a powerful tool for scientific discovery, helping researchers to find, test and commercialise innovations faster. In biomedicine, for example, AI led to a 45 000-fold acceleration in the mapping of protein structures – critical for designing new drugs. Innovation lead times for new energy technologies often span decades. Reducing this period will be key to achieving energy sector goals such as sustainability and competitiveness. Yet only 2% of the equity raised by energy start-ups has gone to companies with an AI-related value proposition.

Energy innovation challenges are characterised by the kinds of problems AI is good at solving. For example, only 0.01% of next-generation solar PV materials have been experimentally produced, leaving a huge set of possible materials still to be explored. AI could allow scientists to dramatically accelerate the process of finding and testing promising materials, battery chemistries and carbon capture molecules. Policy will be required to support AI-led invention and also accelerate commercialisation, which is often a bigger impediment to new products than the discovery phase. 

Energy is amongst the most complex and critical sectors in the world today, yet it can and should do more to seize the potential benefits of harnessing AI. The energy sector faces barriers to realising the widespread adoption of AI, including missing or inadequate access to data and digital infrastructure and skills, as well as persistent digital and physical security concerns, which often trump potential efficiency gains. The prevalence of AI-related skills is much lower in the energy sector compared with other sectors. Policy and regulatory changes will be needed to enable the energy sector to seize the benefits of AI.

The supply chains for the components going into data centres are complex and globalised. For example, gallium is an increasingly critical metal used in cutting-edge computer chips and power electronics, offering significant efficiency benefits compared with traditional silicon- based semiconductor designs. China currently accounts for around 99% of global refined gallium supply. Our estimates indicate that in 2030, demand for gallium for data centres could reach over 10% of today's supply.

AI compounds some energy security risks, but it also offers solutions in both the cyber and physical domains. As AI capabilities increase, so does the capacity for them to be used and misused by various actors. Cyberattacks on energy utilities have tripled in the past four years and have become more sophisticated because of AI. At the same time, AI is becoming a critical tool to defend against them. In the physical domain, AI-equipped satellites and sensors can detect incidents in critical energy infrastructure 500 times faster than traditional ground-based methods and at high spatial resolutions. As the nature of energy security evolves, the IEA will continue to monitor this critical issue.

Emerging and developing economies other than China account for 50% of the world’s internet users but less than 10% of global data centre capacity. Countries with a record of reliable and affordable power will be best placed to unlock data centre growth, localise the computing power that is critical to homegrown AI development, and spur the IT industry more generally. Data centres can also be anchors for new low-emissions power projects. However, in regions with frequent power outages or power quality issues, maintaining a data centre can be risky or costly, making overseas hosting more appealing for businesses. There have also been promising use cases of AI in developing economies that have helped unlock new efficiencies and optimise processes. Overcoming barriers to digitisation can help such economies leapfrog to AI solutions that offer cost and time savings.

Emissions from electricity use by data centres grows from 180 million tonnes (Mt) today to 300 Mt in the Base Case by 2035, and up to 500 Mt in the Lift-Off Case. While these emissions remain below 1.5% of the total energy sector emissions in this period, data centres are among the fastest growing sources of emissions.

The widespread adoption of existing AI applications could lead to emissions reductions that are far larger than emissions from data centres – but also far smaller than what is needed to address climate change. We estimate that emissions reductions from the broad application of existing AI-led solutions to be equivalent to around 5% of energy-related emissions in 2035. Various barriers to AI adoption will need to be overcome to unlock these gains. Rebound effects – for example from modal shifts away from public transport to autonomous cars – could undercut some of these benefits. AI can be a tool in reducing emissions, but it is not a silver bullet and does not remove the need for proactive policy.

The tech sector and energy industry are more intertwined than ever before. There are large uncertainties on the path ahead, but these should not get in the way of concerted action. Delivering the energy for AI, and seizing the benefits of AI for energy, will require even deeper dialogue and collaboration between the tech sector and the energy industry. Along the way, there will be risks to manage. The IEA will continue to provide data and robust analysis to inform decision making and help the energy and technology sectors be better prepared as the adoption of AI unfolds.