AI Boosts Fossil Fuel Emissions, Study Warns
New research shows fossil fuel efficiency gains may outweigh climate benefits
A peer-reviewed study published in npj Climate Action has found that artificial intelligence could deliver a net increase in global carbon emissions. The research suggests that productivity improvements in fossil fuel extraction and processing may generate more carbon dioxide than AI saves through renewable energy applications.
The modeled impact ranges from 0.47 to 1.8 gigatonnes of CO₂ annually. To put that in perspective, the upper estimate represents roughly 4.8% of global energy-related emissions recorded in 2024. The authors describe this as a significant blind spot in current corporate carbon accounting.
Most corporate sustainability reports focus on direct operational emissions from data centers. However, the study argues that this approach misses the broader economic effects of AI deployment across energy sectors. When AI makes fossil fuel operations cheaper and more productive, it can expand commercially viable supply and stimulate additional demand.
This research adds an important dimension to the climate conversation around artificial intelligence. While many businesses treat AI as a tool for reducing emissions, the technology's impact depends entirely on where it gets deployed and at what scale.
How artificial intelligence changes fossil fuel economics
The study uses energy-economic modeling to test what happens when AI adoption rates remain similar across both fossil fuel and renewable energy sectors. Under these conditions, the researchers found that efficiency gains in fossil fuel operations create what they call enabled emissions.
When AI improves the productivity of oil and gas extraction, it reduces the cost per unit of fuel produced. Lower production costs typically lead to lower market prices. Consequently, cheaper fuel increases consumption across the economy. This cycle creates additional emissions that would not have occurred without the AI-driven efficiency improvements.
The research team compared these enabled emissions against the current carbon footprint of data centers. They found that emissions from AI-enhanced fossil fuel operations could be 3.3 to 13.3 times larger than the electricity consumption of the data centers running the AI systems themselves.
Meanwhile, renewable energy applications of AI do generate emissions reductions. The technology can improve grid management, optimize wind and solar output, and make energy storage more effective. However, under the modeled scenarios, these benefits do not offset the expansion of fossil fuel production.
The net result shows annual emissions rising by 0.47 to 1.8 gigatonnes of CO₂. This range reflects different assumptions about adoption rates and productivity improvements. Even at the lower end of the range, the impact equals roughly 1.2% of current global energy emissions.
The International Energy Agency has noted that widespread adoption of existing AI applications could deliver substantial reductions in end-use sectors. Nevertheless, the agency also cautions that these reductions would fall far short of what climate targets require. Other research has similarly found that AI's climate impact varies dramatically depending on deployment patterns across different economic sectors.
Implications for UK businesses reporting carbon footprints
This research has direct relevance for UK companies navigating carbon reporting requirements. Many organizations now measure and disclose their emissions under frameworks like the Streamlined Energy and Carbon Reporting regulations or voluntary standards such as the Greenhouse Gas Protocol.
Current reporting practices typically capture direct emissions from operations and purchased electricity. Some businesses also report supply chain emissions under Scope 3 categories. However, standard methodologies do not account for enabled emissions, which are the indirect market effects created when your technology makes carbon-intensive activities more economically attractive.
For businesses deploying AI in their operations or supply chains, this creates a measurement gap. If your AI systems improve efficiency in processes that ultimately rely on fossil fuels, the resulting emissions increase may not appear anywhere in your carbon footprint. Similarly, if you provide AI services to clients in oil, gas, or coal sectors, the downstream emissions effects remain invisible under conventional accounting.
This matters particularly for companies pursuing net zero commitments or competing for public sector contracts under Procurement Policy Note 06/21. That policy requires suppliers to demonstrate credible carbon reduction plans and increasingly detailed emissions reporting. As scrutiny of corporate climate claims intensifies, businesses may face questions about the full climate impact of their technology deployments.
The financial implications also deserve attention. If regulations evolve to capture enabled emissions, companies could face unexpected carbon costs. Furthermore, investors and lenders increasingly assess climate risk when making funding decisions. A technology investment that appears carbon-neutral today might create material climate liabilities tomorrow if reporting standards change.
For manufacturers, the issue extends into supply chain resilience. If your suppliers use AI to improve fossil fuel efficiency, that could increase your Scope 3 emissions even as your direct operations become cleaner. Consequently, businesses need to understand where AI sits in their value chain and what that means for their overall carbon trajectory.
There is also a strategic dimension for businesses considering AI adoption. The technology can genuinely reduce emissions in many applications, particularly in energy management, transport optimization, and building efficiency. However, the climate benefit depends on displacement. If AI makes your operations more efficient but that efficiency allows you to increase overall production without changing your energy mix, total emissions may still rise.
What the research reveals about carbon accounting gaps
- AI productivity gains in fossil fuel sectors could add 0.47 to 1.8 gigatonnes of CO₂ to annual global emissions, equivalent to 1.2% to 4.8% of current energy-related carbon pollution.
- Enabled emissions from AI-enhanced fossil fuel operations may be 3.3 to 13.3 times larger than the direct emissions from data centers running AI systems.
- Renewable energy applications of AI generate climate benefits, but under modeled scenarios these do not offset the emissions increase from fossil fuel efficiency improvements.
- Standard corporate carbon accounting focuses on direct operational emissions and typically excludes the broader economic effects of technology deployment across sectors.
- The International Energy Agency estimates that existing AI applications could reduce emissions in end-use sectors, though these reductions would fall short of climate targets without broader systemic change.
- The net climate impact of artificial intelligence depends heavily on deployment patterns, with significantly different outcomes based on whether the technology scales fossil fuel production or accelerates decarbonization.
What businesses should consider when deploying AI systems
The challenge for UK companies is that AI presents both opportunity and risk from a climate perspective. The technology can drive genuine emissions reductions in the right applications. However, it can also entrench carbon-intensive business models if deployed without strategic consideration of the broader system effects.
Businesses should start by mapping where AI touches their operations and supply chains. This includes direct uses like process optimization and predictive maintenance. It also includes indirect exposure through suppliers, service providers, and customers. Understanding these connections helps identify where efficiency gains might create unintended emissions consequences.
Procurement decisions deserve particular attention. If you are buying AI services or products, ask vendors about the climate implications of their technology. Specifically, question whether their solutions optimize fossil fuel operations or support decarbonization pathways. For public sector suppliers, this due diligence aligns with the carbon reduction expectations in PPN 06/21.
Carbon reporting methodologies may need to evolve. While current standards provide a framework for measuring direct and supply chain emissions, they may not capture the full impact of technology choices. Businesses preparing for potential regulatory changes should consider how enabled emissions might affect their reported footprint and what that means for net zero commitments.
Investment decisions also require climate-aware evaluation. AI projects that improve operational efficiency can reduce costs and emissions simultaneously when applied to clean processes. However, efficiency improvements in carbon-intensive operations may simply make those operations more profitable without changing their environmental impact. Therefore, project approval criteria should include explicit consideration of the carbon consequences across the full system, not just at the point of deployment.
There is also value in engaging with industry peers and trade bodies about measurement approaches. The research highlights a genuine gap in current accounting methods. Businesses that help develop practical solutions for measuring enabled emissions will be better positioned when reporting requirements inevitably tighten.
For companies pursuing science-based targets or other credible net zero pathways, the key question is displacement. AI-driven efficiency should enable you to do the same work with less energy and fewer emissions. If instead it enables you to do more work with the same total emissions, you are not moving toward net zero regardless of what unit efficiency metrics show.
Training and capability building also matter. SBS Academy offers training on carbon measurement and climate strategy that helps teams understand the difference between operational efficiency and absolute emissions reduction. This distinction becomes especially important as AI deployment accelerates across business functions.
Where to find authoritative guidance on AI and emissions
The Department for Energy Security and Net Zero provides policy updates on climate targets and carbon accounting requirements through its official guidance pages. These resources include information on reporting regulations and the UK's pathway to net zero by 2050.
For detailed technical guidance on measuring greenhouse gas emissions, the government conversion factors for company reporting offer standardized methodologies. These factors are updated annually and provide the basis for most UK corporate carbon reporting.
The International Energy Agency publishes research on data centers and energy consumption, including analysis of how digital technologies affect overall energy demand. Their reports provide context for understanding AI's role in the broader energy system.
Businesses looking for support with carbon reporting and ESG compliance can access structured guidance on meeting regulatory requirements while building credible decarbonization strategies. This becomes particularly relevant as measurement methodologies develop to capture the full climate impact of technology investments.