The rapid digitalisation of the global economy, accelerated by the widespread deployment of artificial intelligence (AI), is fundamentally reshaping patterns of energy demand. Data centres, at the core of this transformation, are experiencing unprecedented growth driven by expanding cloud services, big data analytics, Internet of Things (IoT) applications, and, most notably, the increasing computational intensity of AI workloads. This expert commentary examines the scale, drivers, and implications of this surge in electricity demand, with particular emphasis on its consequences for global power generation and the evolving role of natural gas.
Historically, improvements in energy efficiency enabled data centres to decouple workload growth from electricity consumption. Between 2015 and 2019, global data centre workloads more than doubled while electricity consumption remained relatively stable at around 200 TWh annually. However, this trend has reversed in recent years. Since 2020, electricity demand has risen sharply, reaching over 400 TWh in 2023, reflecting both the rapid expansion of digital infrastructure and the diminishing potential for further efficiency gains. The emergence of generative AI has further intensified this trajectory, as training and operating advanced machine learning models require significantly higher computational power and continuous energy supply.
Looking ahead, the report develops three scenarios, Low, Base, and High, to assess potential electricity demand growth from data centres and AI by 2030. Under the Base Case, considered the most plausible trajectory, global electricity demand is projected to increase to approximately 1,135 TWh by 2030, representing an annual growth rate of around 15.5% and accounting for 3.6% of global electricity consumption. In the High Case, demand could reach as much as 1,560 TWh, while even the Low Case indicates substantial growth to around 770 TWh. These projections highlight both the scale of uncertainty and the magnitude of the challenge facing global power systems.
Meeting this rising demand poses critical challenges for energy systems, particularly in ensuring reliable, continuous, and cost-effective electricity supply. While renewable energy sources are expected to play an expanding role, their intermittency limits their ability to independently meet the constant and high-load requirements of data centres. Nuclear energy offers reliable baseload generation but is constrained by long development timelines and high capital costs. In this context, natural gas-fired power generation emerges as a key enabler, offering flexibility, scalability, and reliability to complement variable renewable energy sources and maintain grid stability.
The analysis indicates that natural gas is expected to account for approximately 41–45% of the additional electricity generation required to meet the growth in data centre demand across scenarios. Its ability to provide dispatchable, on-demand power makes it particularly well suited to support AI-driven workloads, which are characterised by both sustained and peak demand requirements. This is especially relevant in regions such as North America and Europe, where natural gas plays a central role in balancing power systems with increasing shares of renewables.
The report concludes that the expansion of data centres and AI represents a structural shift in global energy demand, reinforcing the centrality of electricity in the future energy system. Addressing this demand will require a balanced and integrated approach that combines renewable energy expansion, continued improvements in energy efficiency, and the strategic deployment of flexible generation sources such as natural gas. In this context, natural gas is positioned not only as a transitional fuel but as an essential component of a reliable and resilient energy system capable of supporting the digital transformation of the global economy.
For further insights and detailed analysis, please refer to the full Expert Commentary below:
Electricity demand growth for data centres and AI and implications for natural gas-fired power generation