How AI Impacts the Climate: Balancing Innovation and Environmental Costs

The rapid expansion of artificial intelligence (AI) technologies has sparked significant debate about their environmental impact. While AI has the potential to aid in combating climate change, the energy demands of AI datacentres and the associated pollution present serious concerns. For instance, a thermal imaging investigation of Elon Musk’s xAI Colossus datacentre in Tennessee revealed substantial methane emissions from gas-fired turbines powering the AI supercomputer. This pollution contributes to greenhouse gas accumulation, raising alarms about the carbon footprint of AI infrastructure. Moreover, the content generated by AI, such as controversial or harmful outputs, has led some critics to question the societal value relative to the environmental costs.

Datacentres, essential for AI operations, currently consume about 1% of global electricity but are projected to increase their share significantly. In the US, datacentre electricity use could rise to 8.6% by 2035, while in Ireland, datacentres already consume one-fifth of the national electricity supply and are expected to grow further. This surge strains existing power grids and often relies on fossil fuels, including coal and natural gas, which undermines renewable energy gains. For example, in Ireland, the datacentre boom has offset progress made through renewable energy expansion. Similarly, in countries like China and Pakistan, datacentres are predominantly powered by coal or fossil fuel plants, highlighting the global challenge of powering AI sustainably.

Despite these concerns, the energy consumption per AI query is relatively low—estimates range from 0.2 to 3 Watt-hours for simple text prompts, comparable to running a lightbulb for a few minutes. However, the sheer scale of AI usage, with hundreds of millions of users and increasing integration into digital services, amplifies the total energy demand. Experts worry about the lack of transparency from tech companies regarding AI’s full energy footprint and caution against assuming future technological advances will automatically solve these issues. The risk is that unchecked growth in AI demand could lead to increased fossil fuel consumption, especially if renewable capacity does not expand at a matching pace.

On the positive side, AI also offers tools to reduce carbon emissions in other sectors. Studies from the International Energy Agency and the London School of Economics suggest AI can accelerate clean technology adoption by optimizing renewable energy integration, improving electric vehicle batteries, and encouraging sustainable consumer behavior. Real-world examples include Google’s AI-driven datacentre cooling efficiency improvements and AI-enhanced maintenance of wind turbines and solar farms in Europe. These applications demonstrate AI’s potential to offset its own carbon footprint by enabling emissions reductions elsewhere. However, uncertainties remain due to possible rebound effects, such as increased energy use from AI-enabled conveniences or autonomous vehicles.

The fossil fuel industry is also adopting AI to enhance production efficiency and expand reserves, which could counteract climate goals. Companies like Saudi Aramco and American Petroleum Institute leaders describe AI as a transformative tool akin to a new fracking boom, increasing oil and gas output. While AI can help detect methane leaks and improve operational efficiency, critics argue that these benefits are insufficient to address intentional emissions releases. Additionally, AI-driven marketing and consumption may increase demand for goods and services, further complicating the net environmental impact.

In response to these challenges, some environmental groups and policymakers are calling for moratoriums on new datacentre development until stricter regulations are in place. Others advocate for integrating AI into climate legislation and taxing AI operations to fund sustainability efforts. Industry insiders and advocates emphasize the need for transparency, reasonable regulations, and the design of energy-efficient AI tools. The debate continues as stakeholders weigh AI’s transformative potential against its environmental costs, highlighting the importance of balancing innovation with responsible energy use and climate action.