Electrification and AI demand growth

The promise of electrification

Electrification is a central driver for global emissions reductions, sitting at the heart of the COP 31 agenda to transition energy, industry and transport systems. This pivot towards electrification is underscored by the COP31 Presidency, which has set out an ambitious target to increase the global electrification rate from just over 20 per cent today to 35 per cent by 2035.

Achieving this target is within reach. Sustained policy support for renewables over decades has created a virtuous circle of technological innovation and economies of scale, driving down wind and solar unit costs, displacing fossil fuels and further incentivising the electrification of end-use demand. Moreover, the target is less daunting than it first appears: given the lower conversion losses in electrical systems, less energy overall is required to heat homes, power industry, and transport people and goods using electricity compared to burning fossil fuels. Bolstered by these cost and efficiency advantages, global electricity demand is now growing faster than overall energy demand – two or three times faster since 2019.

AI and demand acceleration

However, electrification does not automatically equate to displacement of fossil fuels. A key question in the energy transition is the relative rate of growth between electricity demand and clean power generation. If demand rises more quickly than the deployment of clean power, it could lead to increased reliance on fossil fuels such as coal and gas. The promise of electrification may not be realised if demand growth outstrips the system’s ability to build and integrate clean power sources.

Equal attention must therefore be paid to demand as to the supply of clean power. Particularly concerning from a climate change point of view are emergent uses of electricity which may compete for renewable supply and slow down the displacement of fossil fuels in transport, heating and energy-intensive industry. The rise of data centres is one prominent example. Electricity consumption in this sector has surged in recent years, with particularly rapid growth in the US, where it increased by more than 25 per cent in 2025 alone.

Data-intensive economic development is a long-term structural trend which has been characterised by successive waves of investment: first, with the expansion of the internet in the 1990s, and later with the rise of cloud computing and hyperscale data centres in the 2010s. AI represents a third wave, and the impacts of this technological revolution on electricity demand are highly uncertain.

The International Energy Agency’s 2024 scenario report illustrates a wide range of possibilities looking out to 2035: from 700 TWh of electricity consumption from data centres in its Headwinds case, increasing to 1,700 TWh (around 4.4 per cent of global electricity demand) in its highly ambitious Lift-off case.

Energy system impacts of data centres

Whatever the scenario, the centralisation of computing services in ever-larger data centres will require significant electricity capacity to power servers and cool equipment. A key challenge has been the rapid growth of data centre clusters in high-growth cities and regions, which often has the effect of accelerating demand in already constrained electricity systems. This scaling and clustering effect can result in increased output from fossil fuel generation and increased emissions. It can also drive up energy prices, potentially creating new inequalities in how the costs and
risks of the energy transition are allocated across society.

There are already well-known cases: In Virginia in the US, rapid demand growth has required major new investment in gas-fired generation and grid infrastructure. In Ireland, data centres accounted for more than 20 per cent of national electricity use in recent years, raising concerns about security of supply and the pace of renewable deployment. In the UK, while data centres currently account for a relatively modest share of electricity demand, at 2.5 per cent, this is forecast to quadruple by 2030.

A vision for green data centres

There is a growing need to regulate data centre developments more stringently to ensure that decisions about their scale, location, energy procurement and operations align with decarbonisation goals. To fully align with the COP Global Action Agenda, policy and regulatory frameworks should also emphasise circular economy principles, to encourage the recycling of waste heat, water and IT hardware.

Scotland offers a potential case study for this systemic vision. It has a relatively cool climate and abundant renewable resources, generating far more renewable electricity than it consumes (38.5 TWh of renewable output and 20.8 TWh of net exports in 2025). Data centres of an appropriate scale, in the right locations and operated as ‘grid-interactive assets’, could improve the utilisation of wind power in Scotland, which is often curtailed because of congestion on the grid.

At the University, we are working closely with policymakers, industry and other stakeholders to develop a framework for green data centres in Scotland. This is a complex systems challenge, requiring detailed energy systems modelling and analysis of the social and environmental contexts where data centres could be located in the coming years.

Whole systems and circular economy principles should underpin this ‘twin transition’ of AI and renewables, with data centre developments appropriately sized and located so as not to damage local environments and exceed the operating limits of key infrastructures.

This research programme is showing how universities, industry and policymakers can come together to forge a sustainable path for AI.