Artificial Intelligence

Artificial intelligence (AI) is emerging as one of the most consequential technologies of our time. In recent years, the capabilities of AI systems have grown rapidly, as has early adoption by both businesses and individuals. The AI revolution is being driven by rapid improvements in the cost and performance of computing hardware, such as the falling costs of Graphic Processing Units (GPUs), the explosion of data driven by the internet, and continuous improvement in the algorithms used to train AI models. While innovation is progressing very rapidly, the outlook for the technology’s capabilities and its adoption remains uncertain.

In the energy sector, the deployment of AI could help to optimise how extremely complex and data-rich systems like electricity networks are built and managed. AI could also accelerate innovation in new energy technologies, helping to develop more efficient and cheaper batteries for electric vehicles (EVs) or catalysts for hydrogen production, for example.

At the same time, AI is energy hungry. The rise in AI deployment has led to a surge in investment in new data centres, which is raising concerns about soaring power demand. Although today's electricity demand from data centres is small in the context of the global energy system, it has increased in recent years and is expected to keep growing. This is creating challenges for utilities and system operators, particularly in areas where data centres are concentrated. At the same time, large tech companies have also emerged as major customers for clean electricity and are driving forward investment in new technologies such as small modular reactors (SMRs) and long-duration energy storage.

The IEA has long been at the forefront of understanding the links between the energy sector and digitalisation. As the only global agency tracking all fuels, technologies, sectors and geographies, it is uniquely placed to analyse the connections between AI and energy. To explore the opportunities and challenges ahead, the IEA has launched a major new initiative: Energy for AI, and AI for Energy. As part of this initiative, the IEA will organise the Global Conference on Energy and AI on 4 and 5 December, providing a platform for dialogue among governments, the energy industry, the tech sector, researchers and civil society.

Data centres serve as the backbone for storing, processing and distributing data for various applications, including websites, cloud computing and AI services. In 2022, data centres collectively accounted for about 1% of global electricity demand (excluding data networks and cryptocurrency mining). In large economies, such as China, the European Union, the United Kingdom and the United States, data centres are estimated to account for 2-4% of electricity demand. That said, these aggregate numbers hide local challenges, as data centres tend to cluster together. Large data centres today can consume as much electricity as an electric arc furnace steel mill, but they are typically much more spatially concentrated than other similarly energy-intensive infrastructure.

As a result, in at least five states in the United States, data centres have already surpassed 10% of total electricity consumption. In Ireland, the share is over 20% of all metered electricity consumption. With major data centre campuses currently under development, this could lead to considerable strain on local grids. Meeting this electricity demand with emissions-intensive sources of electricity could also throw regional energy transition targets off track. Given this concern, companies with large data centre operations have been active in procuring low-emissions sources of electricity from renewables and investing in technologies to produce nuclear and geothermal power.

The energy system is hugely complex, data-rich and dynamic. Globally, there are around 75 000 power plants, nearly 80 million kilometres of electricity distribution and transmission lines, 1.4 billion cars, 900 000 kilometres of natural gas pipelines, 1.5 billion air conditioning units and 1 200 large steel plants, for example. Assets and infrastructure such as these operate in challenging and changing environments in which they must match fluctuating supply and demand.

A number of trends are further increasing the complexity of the energy system. Variable and distributed renewables are growing their share in global electricity generation, while technologies like heat pumps and electric vehicles are affecting energy demand patterns. Meanwhile, digitalisation is dramatically increasing the amount of data generated by the energy system, and climate change is creating more frequent and intense extreme weather risks.

Artificial intelligence offers the potential for solutions to some of these problems. AI is being applied to better forecast electricity demand and renewables output, automatically manage demand to better match supply, and monitor infrastructure for faults and predict maintenance needs. And applications are not limited to the electricity sector. In the industry sector, the combination of AI and robotics could lead to more efficient and precise production processes, potentially lowering the costs of PV or battery manufacturing. There are also substantial AI applications for oil and gas production, including analysing subsurface data and optimising the operations of refineries or drilling rigs.

Many of the energy technologies required for clean energy transitions are already commercially available and increasingly cost-competitive. However, meeting countries’ goals for achieving net zero emissions will require clean energy technologies that are not commercially available today – as well as continuous improvements to those already on the market.

AI is increasingly being applied to accelerate innovation and technology development in numerous fields, including in the energy sector. For example, researchers are using AI to accelerate the discovery of promising battery chemistries. However, understanding how much AI could accelerate energy innovation, as well as for which technologies and how, will require further analysis and collaboration between industry and government.