This is an extract from a recent report “Energy and AI” by the International Energy Agency (IEA).
AI for energy optimisation: Artificial intelligence (AI) is being deployed across various parts of the global energy system, where AI applications are suited to meeting a wide variety of objectives, including cutting costs, integrating a growing share of variable renewables, making systems more efficient, enhancing fuel supply, ensuring timely maintenance of infrastructure and reducing emissions. The energy system is currently seeing rapid change that creates new challenges and opportunities – many of which are well suited to AI applications. The key trends in the energy sector include:
Rising electrification: The overall share of total final energy consumption met by electricity has been steadily rising and is projected to accelerate.
Growing digitalisation: Energy systems are becoming more digitalised and integrated through the proliferation of connected devices and appliances, electric vehicles, smart meters, and smart sensors in industrial and commercial applications.
Rising complexity: The evolution of the energy system is resulting in greater complexity in supply, demand and energy flow patterns. On the supply side, electricity generation from variable sources, such as wind and solar, is growing fast.
Pressure on costs: The last few years have been challenging for energy consumers around the world, with high energy prices putting significant pressure on the cost of living. With new entrants in the market on both the supply and end-use sides, the energy sector has also become more competitive.
In addition to these structural trends, the energy sector is subject to several important policy objectives. International targets aim to make the energy sector more efficient and sustainable. The energy sector is the largest source of greenhouse gas emissions, which cause climate change. Energy sector emissions have continued to rise, reaching 37.8 gigatonnes of carbon dioxide in 2024 – the hottest year on record. Energy also needs to be reliable, affordable, secure and resilient. AI can help advance progress on these critical challenges, but its successful deployment is likely to depend on several key criteria. Typically, for AI applications to be deployed, they require the availability of digital infrastructure and skills. Widespread use of sensors, analytics and control systems allows for the collection of the extensive datasets that AI needs, with increased scope for automation. Where advanced software systems are already in place, AI capabilities can be rapidly deployed – but this is often inconsistent with the slow turnover of capital equipment in the energy sector.
AI for energy and minerals supply
Digitalisation in the oil and gas sector has progressed rapidly in recent years. Oil and gas companies were among the earliest adopters of supercomputers to boost prospects for oil and natural gas exploration and reduce costs. Moreover, mining companies have increasingly developed digital technologies in recent years. The growth of AI opens up the potential to expand on this, helping companies to explore and identify additional volumes of oil, gas and minerals and plan their development, reduce costs, improve safety and reduce environmental impacts.
Production forecasting is a critical component of the oil and gas industry, enabling companies to optimise operations and manage resources effectively. Traditional methods rely on many assumptions and oversimplifications. AI-driven forecasting methods have been evolving to overcome these challenges and improve results. Various AI and machine learning techniques are being applied to production forecasting. For example, a hybrid AI model for oil production showed significant improvements in accuracy compared to traditional methods and a recent comparative analysis of machine learning techniques predicted oil production to a much higher degree of accuracy. Recently, ExxonMobil’s AI- powered demand forecasting model was reported to have reduced forecast errors by 25 per cent.
Additionally, the use of AI can also significantly increase the potential for operations, monitoring and control to be carried out remotely. A typical oil platform operates tens of thousands of sensors (measuring aspects such as the temperature, pressure, and flow rates of produced liquids), which generate terabytes of data. Analysing and utilising these data streams from a centralised, remote location can increase efficiency and safety and reduce the costs of operations, which AI can assist in the management of.
In the mineral mining sector, machine learning and AI techniques already play a significant role in the exploration, mine operations and extractive metallurgy. Many AI techniques in mineral exploration parallel those in upstream oil and gas industries, where machine learning has long been used for subsurface data interpretation, reservoir simulation and reducing uncertainty. However, AI can also be used to process geophysical data to improve anomaly detection and orebody prediction, lowering costs and boosting resource confidence while reducing sampling needs.
Once an ore deposit is identified, AI can contribute to improving productivity, safety and cost-efficiency in mining operations. Autonomous haulage systems allow for high-utilisation operations, reducing labour costs while increasing safety and fuel efficiency. Predictive maintenance algorithms analyse sensor data from heavy machinery to anticipate failures before they occur, helping to reduce unplanned downtime and extend equipment lifespans. AI is also being applied to ore tracking systems that monitor material movement from blasting through processing, ensuring that high-grade material is prioritised while minimising waste and environmental impacts.
Refining and metallurgical processes can also benefit from AI in driving gains in efficiency and recovery rates. Machine learning algorithms analyse real-time plant data, such as temperature, pressure and flow rates, to fine-tune processing conditions dynamically. Sensor-based sorting systems use AI to distinguish valuable ore from waste, improving pre-concentration and reducing the volume of material. Computer vision technology is being applied in flotation circuits to optimise mineral separation and recovery rates.
AI for electricity sector
The power system has become increasingly complex in many countries, as the production of electrical power has shifted from large, centralised power plants to a multitude of small, distributed sources. In parallel, a digitalisation revolution is producing large pools of data, which in turn can be used to manage the complexity of the whole system. The integration of AI into the electricity sector could bring significant system-wide benefits with its ability to process huge amounts of data and provide optimisations based on trained models rather than predetermined rules. AI has the potential to play a critical role in managing the complexities of integrating renewable energy sources into the grid. AI-enhanced control systems could allow plants and facilities to operate at their rated performance for longer periods, improving efficiency while minimising downtime.
Another important AI application at the system level is to enhance the forecasting of electricity demand and supply from variable renewables in order to optimise the use of power sector assets, including dispatchable power plants, energy storage and demand-side flexibility, and ultimately improve the overall efficiency of the power system. Additionally, advanced AI driven weather and demand prediction models allow grid operators to anticipate fluctuations more accurately, minimising the curtailment of wind and solar photovoltaic (PV) in conjunction with demand shifting or storage. AI can significantly improve the accuracy of weather forecasts by analysing vast amounts of historical and real-time meteorological data, which can also improve the resilience of energy systems. Further, machine learning models can also predict local weather conditions, such as wind speeds and solar radiation, with high precision. These accurate predictions help anticipate the output of wind and solar farms at specific locations.
AI-driven data analytics could further improve planning, project design and real-time operational decisions, resulting in reduced fuel consumption, lower carbon emissions and extended asset lifetimes. For renewable energy projects, AI is being applied to design solar and wind projects, including the selection of primary equipment (solar panels or wind turbines), the siting of the equipment (orientation of panels, available areas and spacing of turbines) and the planning of supporting infrastructure, all to optimise performance and returns on investment.
In the next phase of a project, AI can also be applied to accelerate permitting and licensing processes, which often span several years and require thousands of pages of documents to be drafted by applicants and processed by regulators. Lengthy permitting times have been flagged as a concern for countries seeking to reach their policy ambitions, including for renewables.
AI and Energy in India
India has a thriving information and communication technology sector, with the value of IT exports steadily growing to over $200 billion in 2024. By comparison, the world’s largest oil exporter earned $220 billion on export revenues that year. India is also home to around 950 million Internet users. Spurred by data localisation requirements in some sectors, India is now emerging as a rapidly growing data centre market. As of June 2024, India had 2 GW of total installed data centre capacity in operation, together consuming electricity equivalent to 6.5 million Indian households. India’s total installed data centre capacity has doubled in only four years, and over 2 GW of further maximum designed capacity is in the pipeline and planned to come online over the next two years. This means that total installed capacity is on track to reach nearly 5 GW by 2030. The government’s IndiaAI Mission, with a budget of $1.2 billion, consists of several objectives, including the development of an AI computing ecosystem with over 18 000 GPUs to support AI start-ups and research. In addition, there are incentives from state governments for data centres; for instance, Uttar Pradesh announced a 100 per cent exemption on electricity duty and transmission charges for ten years for new data centres.
Electricity consumption from data centres is contributing to India’s electricity demand growth at a time when India is already among the world’s fastest-growing electricity markets. Since India’s “open access” rules enable the direct purchase of power from generators, several technology companies are signing power purchase agreements directly with renewable energy generation companies to reduce their emissions. For example, the data centre subsidiary of Indian telecommunications major Bharti Airtel announced it would procure 140 GWh of renewable energy annually and has been working with generation companies to set up captive solar PV and wind capacity for their data centres.
To ensure that the upcoming wave of new data centre construction remains on target, India will need to address long-standing issues of grid reliability to capitalise on data centre and AI growth. In the current context, backup and captive power generation for data centres remains a critical consideration owing to the risk of power supply interruptions from the grid. Grid infrastructure creation and upgrades will also need to keep track of new data centre construction. Data centres are proving to be important energy consumers in India, creating additional demand for power generation, notably from solar PV and wind, and driving investment in power backup options (including battery storage) and transmission infrastructure upgrades.
Access the full report here
