.png)
The largest technology companies are contributing to a surge in data centre investment, as their capital expenditure exceeded USD 400 billion in 2025 – and is expected to jump by another 75% in 2026. Capital expenditure of just five technology companies is now larger than global investment in oil and natural gas production. Many jurisdictions are seeing project pipelines accelerate dramatically, although not all projects will come to fruition. Those that are moving forward are doing so at pace: the IEA’s unique satellite-based tracking shows that “artificial intelligence (AI) factories” – cutting-edge data centres specifically designed for AI – have more than tripled in capacity in the past 18 months. Meanwhile, the capabilities of AI are improving quickly, increasing the likelihood that it will reshape economic growth, innovation and competitiveness and disrupt established industries and jobs.
In April 2025, the IEA published its landmark Energy and AI report, which provided first-of-its-kind global analysis on the links between AI and energy. Since then, the field has evolved rapidly: new questions have emerged and new data has come to light. This report builds on the foundation of previous work, providing fresh analysis on key issues. It covers the outlook for data centre electricity demand considering recent market developments; innovations in powering data centres; and the implications of these trends for energy security, affordability, competitiveness and overall energy demand.
Measured per individual task, the energy efficiency of AI is improving at a rate unprecedented in energy history. Software and hardware advances have resulted in the energy use per AI task dropping by at least an order of magnitude annually in recent years. Simple text queries now typically consume less electricity than running a television over the same period of time. If all conventional internet searches were performed with simple AI text queries, it would consume less than 4 terawatt-hours (TWh) of electricity annually, equivalent to less than 1% of total data centre consumption today.
However, new energy-intensive AI applications are increasingly being launched and used, such as those for video generation, reasoning and agentic tasks. These kinds of tasks can consume hundreds or thousands of times more energy per query than simple text generation. The energy demand of AI is therefore the result of three rapidly evolving and uncertain trends: improvements in efficiency, surging uptake, and changing model capabilities, which can unlock new and, in many instances, far more energy-intensive use cases. To improve the robustness of the outlook for AI’s energy demand, close monitoring, frequent updates and cooperation with the tech sector, including more systematic energy consumption disclosures, will remain important.
The global electricity demand of data centres – the critical infrastructure for training and running AI models – grew by 17% in 2025, in line with IEA projections. Electricity consumption from AI-focused data centres grew even faster, surging 50% in 2025. While there are no comprehensive statistics on the frequency and depth of AI usage around the world, major model providers reported a threefold increase in active users and a fivefold increase in revenue over the past year, highlighting the rapid growth of demand.
The speed of the AI revolution is increasingly contrasting with the speed of the physical, social and economic systems that underpin it. Bottlenecks across energy supply chains and advanced chip manufacturing have tightened since our last report. Planning and regulatory systems are being stretched by the wave of project applications for data centres, amid a broader trend of rapid load growth and electrification. Social acceptability is also a growing issue, as communities push back against data centre projects, and concerns about affordability and environmental impacts rise. Essential elements within the IT industry are currently facing limitations; notably, a shortage of high-bandwidth memory – integral to AI chip production – has developed over the past six months and is anticipated to persist through at least the end of 2027.
Data centre investments have grown too large to be funded from company balance sheets alone, and large amounts of funding from capital markets will be critical for their buildout. As a result, the pace of data centre growth, and the resulting increase in energy consumption, will be sensitive to market sentiment, including expectations for returns on investment in data centres and AI deployment, as well as to broader macroeconomic and financing conditions. Understanding the energy implications of AI therefore also means following closely the technology’s economic trajectory.
The central projection for electricity demand from data centres remains close to the trajectory set out in the IEA’s 2025 report. Our updated projections see electricity consumption from data centres roughly doubling from 485 TWh in 2025 to 950 TWh in 2030, accounting for around 3% of global electricity demand by that date. Electricity consumption from AI-focused data centres grows much faster than overall data centre electricity consumption, tripling in this period. Bottlenecks across the value chain, however, are reducing the likelihood of more aggressive near-term scenarios, despite booming investment and surging project pipelines.
The mid- to longer-term outlook for data centre electricity demand sees a possible upside. Investments in relieving bottlenecks across energy equipment and chip manufacturing, and rapid uptake of energy-intensive use cases of AI, raise the prospect that there could be an even higher upside case after 2030. The IEA will continue to update its projections regularly.
An individual server rack within an advanced data centre is only the size of a large refrigerator, but by 2027 it could have peak power demand equivalent to that of 65 households. The speed of this shift is remarkable: between 2020 and 2025, the power density of AI servers increased by 11 times; by 2027, it is set to see a further fourfold increase. This will test the capacity to ramp up supply chains for key electricity technologies such as power electronics and transformers. Some of these technologies also depend on critical inputs from a small number of producers, notably China. Care is therefore needed to ensure that supply chains for the critical emerging technologies going into data centres are diverse and resilient.
The tech sector remains a major driver of renewables procurement, accounting for around 40% of all corporate renewables power purchase agreements (PPAs) signed globally in 2025. Renewables PPAs signed by data centre companies are equivalent to almost half the sector’s current consumption. But procurement innovation extends beyond renewables. The tech sector has become a major driver of momentum behind both conventional and new nuclear plants, as well as next-generation geothermal. Since the IEA’s 2025 report, the pipeline of data centre offtake agreements with small modular reactors (SMRs) has grown from around 25 GW at the end of 2024 to 45 GW by the end of 2025. Nonetheless, the first projects are not expected to come online until around 2030.
Unlike traditional data centre operations, AI training and model use induce large and rapid power swings, making energy storage critical to ensure that electricity is always supplied reliably. By 2030, around 20-25 GW of battery storage could be installed in data centres globally, potentially making them a grid asset if the incentives are right. Recent months have seen a data centre operator sign an agreement for the largest battery project ever by energy capacity (four times larger than the previous record-holder), helping to commercialise long-duration energy storage.
Constrained by slow grid connections, data centre developers in the United States are pushing forward projects with onsite natural gas-based power generation. Satellite tracking of these projects indicates that around one-fifth have started land clearing or construction. This highlights that onsite gas power is an emerging solution, with key design, supply chain, regulatory and financial questions remaining. New IEA analysis shows that providing reliable onsite gas-fired electricity to meet critical and variable data centre load requires overbuilding onsite generation infrastructure by 30% to 70% relative to demand. Yet in the context of a supply crunch for gas turbines, it is not clear that onsite generation necessarily promises a faster route to market for data centres at scale. Though uncertainties are high, around 15-27 GW of onsite natural gas may power data centres by 2030, mostly in the United States. However, this does not remove the urgency of addressing grid bottlenecks, as most data centres prefer connecting to the grid.
The AI boom could therefore accelerate deployment and innovation in the electricity sector, if spending continues and if government support is also aligned. In this report, we analysed share price movements of AI and energy companies to understand how financial markets expect AI’s energy demand to impact energy companies. On the one hand, financial valuations since the launch of ChatGPT do not suggest that AI demand will provide a generalised uplift to the energy sector; it is simply too small in the context of the energy sector as a whole. In contrast, manufacturers of gas turbines and electrical equipment, some nuclear companies, and some energy startups have seen their valuations become more strongly linked to AI, highlighting the pull that data centre demand is providing for innovation and growth in their businesses.
Initial signs indicate AI is boosting productivity in some sectors, which may push up overall economic growth. There are a wide range of projections of the impact of AI on GDP. Based on cooperation with the OECD and economic modelling of the possible task-by-task productivity boost coming from AI, this report provides a first-of-its-kind analysis of the implications of an AI-driven GDP boost on the energy sector.
Stronger economic growth from AI does not translate one-to-one into higher energy demand. It is concentrated in knowledge-intensive services and high-income countries, where the elasticity between energy demand and economic activity is lower. Estimates show that, depending on the scale of uptake, an AI-driven growth boost could raise the level of global energy demand by between 1-4% in 2035 compared with trends without this AI boost. Effects are concentrated in advanced economies, particularly the United States, although emerging and developing economies also benefit from increased economic activity.
Ultimately, what matters most for energy demand are energy policies and energy technologies. In our analysis, the impact of energy policies and energy technology developments on energy demand is much larger than the potential impact of an AI-driven economic growth boost.
Over the past year, energy and technology supply chains have become further strained. Trade restrictions have targeted key components going into data centres, such as the critical minerals needed for advanced power electronics, or the batteries and battery components needed to smooth AI loads. A 70% surge in gas turbine orders in 2025 has highlighted chokepoints in energy technology supply chains. And data centres themselves have been targeted in conflict zones, underscoring their role as critical infrastructure. The broader implications of the energy crisis tied to the conflict in the Middle East are still unknown, but they could extend to the choices that countries and companies make about the fuels and technologies used to power data centres, and where sites are located.
AI has the potential to be an important tool to enhance energy security and sustainability. For example, AI technologies monitor grids, transformers and other energy equipment to reduce unexpected failures and outages, and AI and digital grid-enhancing technologies are key to optimising the use of existing grid capacity, helping to offset lengthy and costly grid expansions. At the same time, AI risks making cyberattacks easier and more powerful, and an increasingly digitalised energy sector presents new vulnerabilities.
With increasing deployment of AI-enabled robotics, automation and efficiency solutions, AI will be critical to industrial competitiveness. Advances in data, models and hardware now allow for broader automation in industrial design and production, speeding up development, boosting innovation and lowering costs. The AI-enabled optimisation of production processes could reduce energy costs by 3-10 percentage points in energy-intensive industries, where energy is a critical production input and margins tend to be low. Well-documented AI use cases have the potential to save over 13 exajoules (EJ) of energy by 2035, equivalent to 3% of global final energy consumption, if barriers to wider uptake are overcome. Longer-term, AI could unlock another wave of industrial innovation and productivity: the almost two-fold jump in venture capital going into this field in 2025 is a marker of the potential. The race to develop more powerful AI models is being run alongside another race to apply AI across the economy for innovation and productivity, with China particularly focused on the latter.
An IEA survey of energy companies reveals that the lack of digital skills is the single largest barrier to greater AI adoption in the energy sector. The adoption of AI is also limited by fragmented data and concerns related to data protection, privacy, and cybersecurity. For example, only 10% of global electricity consumption is covered by open electricity data policies. The lack of adequate digitalisation of equipment can also be a barrier. Globally, less than half of energy demand is covered by policy frameworks aiming to promote the uptake of AI in the energy sector.
Data centres are a highly visible flashpoint for concerns around energy prices and the environment. Energy prices have risen up the agenda since the inflationary shock of Covid-19 and the 2022 global energy crisis, with the energy crisis sparked by the conflict in the Middle East compounding concerns. In addition to the potential impact of data centres on electricity prices, surveys increasingly show that citizens are concerned about AI’s effect on jobs, the economy, the environment, and society more broadly. The emissions associated with data centres double in IEA projections, reaching around 350 million tonnes in 2035, but still make up about 2% of global electricity sector emissions by that date.
The impact of electricity demand growth on electricity prices depends on a combination of fundamental factors and policy choices. In systems with tighter demand-supply balances, load growth can indeed trigger a need for new investments, potentially raising prices. On the other hand, in systems with excess electricity supply, load growth can result in more efficient capital utilisation and lower prices. The shape of the load growth matters too. In the same way that high and reliable passenger numbers can allow airlines to lower average ticket prices on established routes, predictable baseload electricity demand can increase the efficiency of the use of capital-intensive power plants and grids, potentially lowering electricity prices.
Data centres can create special challenges for electricity affordability. Data centres are large, concentrated, rapidly developed infrastructure that are likely to trigger a need for new generation and grid investments in systems that host them. Actual peak load from data centres is often uncertain, as data centres are filled progressively with operating servers and often oversize their grid connections initially. The mismatch between fast-moving data centres and slow-moving energy investments, and the uncertainty over their load, creates risks of misalignment between electricity system investment and data centre demand that could raise electricity prices in some places if unmitigated.
The IEA proposes three principles to help ensure AI deployment is leveraged for the benefit of the energy sector, and that data centres minimise any adverse impacts on electricity systems.
Write a comment
Các trường bắt buộc được đánh dấu *