AI-driven productivity gains found to increase fossil fuel emissions
New research published in Nature has revealed the unseen climate impact of AI, with the technology found to be responsible for so-called ‘enabled emissions’. According to the study, emissions enabled by AI-driven fossil fuel productivity gains outweigh the emissions avoided through AI applications in renewable energy, even under scenarios where both technologies are adopted in parallel.
These additional greenhouse gas emissions are the result of oil and gas companies using AI to enhance fossil fuel productivity across the value chain. Examples include AI systems that optimise drilling operations, improve reservoir modelling, predict equipment failures, reduce downtime and increase recovery rates from existing oil and gas fields.
The study was published by the Enabled Emissions Campaign (EEC). According to their modelling scenarios, AI is enabling between 0.47 and 1.8 gigatonnes of CO₂ annually (equivalent to 1.2- 4.8% of 2024 global energy-related emissions)[i]. In fact, AI's fossil fuel applications enable more emissions than its renewables applications avoid, even under parallel adoption.
Conventional assessments of AI's climate impact focus on data centre energy demand, AI-optimised renewables, and demand-side efficiency, but largely omit AI's significant effect on the economics of fossil fuel supply. The researchers argue that this creates a significant blind spot in climate assessments of AI. While much attention has focused on emissions associated with data-centre electricity consumption, the study suggests that AI's role in increasing fossil-fuel productivity may have a far greater impact on global emissions.
Taken in isolation, enabled emissions are 3.3 to 13.3 times larger than today's data centre emissions, and up to 8 times larger than projected 2035 data centre emissions. For AI's climate benefits to outweigh its fossil fuel effects, renewables gains would need to be 4 to 5 times larger than fossil fuel gains.
AI’s growing environmental toll
AI is already under significant scrutiny regarding emissions, with concerns around soaring energy needs and water requirements. The IEA previously reported that growth in the use of artificial intelligence will lead to global electricity demand from data centres more than doubling in the next 5 years, reaching 945 terawatt-hours (TWh) in 2030. Notably this figure is equivalent to Japan’s current annual electricity usage and represents an almost 128% increase from the 415 TWh of power used by data centres in 2024[ii]. The IEA has credited AI as "the most significant driver of this increase, with electricity demand from AI-optimised data centres projected to more than quadruple by 2030".[iii]
Likewise, with many data centres using fresh water to keep servers and equipment cool, there is growing stress being placed on water courses and supply networks particularly as most centres source their water in the form of potable (drinkable) water from utility companies. The research paper “Making AI Less ‘Thirsty’: Uncovering and Addressing the Secret Water Footprint of AI Models” released in October 2023, found that globally, AI has the potential to account for between 4.2 billion - 6.6 billion cubic meters of water withdrawal by 2027. This is equivalent to roughly half of the UK's annual freshwater withdrawals[iv].
While the Nature paper focuses on AI-enabled fossil fuel emissions, data-centre expansion remains the most visible aspect of AI's environmental footprint. Big tech in particular has received criticism for its part in the rapid roll-out of data centres. The ‘big three’, Microsoft, Amazon, and Google have seen their emissions rise collectively by almost a fifth in the past year, with the bulk of this growth attributable to data centre build-out.
References
[i] ENABLED EMISSIONS CAMPAIGN | Hold Big Tech Accountable Today
[ii] Data centre energy requirements to double in the next 5 years, as AI demands soar
[iii] Executive summary – Energy and AI – Analysis - IEA
.jpg)
