AI for Decarbonisation: Driving Solutions

AI's Dual Role in Decarbonisation

The world faces mounting pressure to decarbonise rapidly as deadlines loom. Major economies have committed to net-zero emissions by 2050, with interim targets requiring emissions cuts of around 45% by 2030 from 2010 levels to align with the 1.5°C Paris goal. Yet progress lags, and current trajectories point to continued rises in emissions in many sectors without accelerated action.

Artificial intelligence has emerged as both a powerful enabler and a complicating factor in this transition. On one hand, AI optimises energy grids, improves renewable forecasting, accelerates materials discovery for clean technologies, and enhances efficiency in heavy industries like chemicals and manufacturing. Studies suggest AI could reduce global greenhouse gas emissions by several gigatonnes annually in areas such as transport and agriculture by the mid-2030s, potentially outweighing the emissions from data centres.

On the other hand, the explosive growth of generative AI and large-scale models has triggered a power super-cycle. Data centres supporting AI are driving unprecedented electricity demand, with projections indicating AI-related consumption could exceed that of entire countries by 2026. This surge has prompted utilities to plan new capacity, often leaning on natural gas for reliability while clean energy ramps up, creating short-term emissions risks. Tech giants face scrutiny over whether their net-zero pledges hold amid this demand, with some achieving renewable matching milestones but broader concerns about additionality and grid strain persisting.

Non-obvious tensions include the paradox of AI's energy intensity potentially delaying clean transitions in regions with constrained grids, while simultaneously catalysing investment in renewables and nuclear. Claims of AI averting climate breakdown often conflate efficient machine learning with power-hungry generative tools, leading to accusations of greenwashing. Meanwhile, policy responses, such as US efforts to streamline large-load interconnections, aim to balance AI growth with decarbonisation imperatives, but outcomes depend on how swiftly low-carbon sources scale.

In the UK, initiatives like the ADViCE programme reflect efforts to direct AI toward high-impact decarbonisation challenges, highlighting the need for targeted tools to map solutions to pressing needs in the race to net zero.