Artificial intelligence may accelerate oil extraction, but also increase emissions
Artificial intelligence can assist the development of renewable energy, but its use in the extraction of oil, gas and coal could have a much greater impact on the climate. At the same pace of technology adoption across different sectors, global carbon dioxide emissions could increase by 0.47–1.8 billion tons per year.
This is indicated by the results of a new study published in the scientific journal npj Climate Action. The authors analyzed not only the energy consumption of data centers, but also how AI changes the productivity of the entire energy industry.
What the study showed
Researchers used a global economic model that accounts for interactions between different sectors, changes in production costs, prices, demand, and the redistribution of investments. Altogether, they examined 64 scenarios for the application of artificial intelligence in fossil and renewable energy.
In scenarios where AI is being introduced at roughly the same pace in both sectors, annual CO emissions increased by 0.47–1.8 gigatons. This corresponds to approximately 1.2–4.8% of all energy-related emissions recorded worldwide in 2024.
A separate modeling showed that increased productivity of fossil energy enterprises could cause additional emissions at the level of 0.6–2.4 gigatons of CO per year. Optimization of renewable energy compensates only part of this volume.
According to the authors' calculations, every 1% increase in fossil fuel extraction productivity must be accompanied by a 4–5% increase in renewable energy productivity for the overall emissions balance to remain neutral.
How AI helps extract more oil and gas
Oil and gas companies use artificial intelligence to analyze seismic data, find new fields, plan wells, and predict technical malfunctions. The technology also helps optimize equipment operation, reduce downtime, and lower production costs.
As a result, companies can extract more fuel from existing fields and start developing reserves that were previously considered economically non-viable. Cheaper production can also reduce prices and stimulate additional demand for oil, gas and coal.
The International Energy Agency previously estimated that the application of AI could increase the volume of technically accessible oil and gas reserves by about 5%, and reduce the cost of implementing deepwater projects by 10%.
The technology is already actively used by large market participants. Saudi Aramco uses AI in various production processes, and Norwegian Equinor has linked a number of new discoveries on the continental shelf to automated analysis of seismic data and well planning.
What benefits AI brings to renewable energy
In the clean energy sector, algorithms can predict electricity production depending on wind speed and sunlight intensity. This helps operators balance the power system and use battery storage more efficiently.
AI is also used for predictive maintenance of solar and wind power plants. Early detection of faults reduces equipment downtime, extends its operating life, and increases the volume of electricity produced by already installed capacities.
However, the study authors point to a structural inequality between the two directions. Renewable energy depends on the construction of new generating facilities, grids and storage systems, whereas oil and gas companies can apply AI more quickly at existing fields and immediately ramp up extraction.
Will a carbon emission fee help
The researchers separately tested scenarios with the introduction of a high price on carbon emissions. At a cost of $80 per ton of CO, artificial intelligence still increased net emissions by approximately 0.7 gigatons per year.
Even at a price of $308 per ton, the model showed an increase of approximately 0.1 gigatons. Thus, a carbon emission fee reduced the negative effect, but did not eliminate it entirely if fossil energy productivity continued to grow.
Improvements in the operation of power grids and increased industrial energy efficiency also could not fully compensate for the consequences of expanding fossil fuel supply.
What the model did not consider
The authors emphasize that the indicators obtained are not an exact forecast for a specific year. The model demonstrates a probable direction and approximate scale of AI’s impact given the current structure of the global economy, in which over 80% of primary energy still comes from fossil fuels.
The calculations do not include direct emissions from the operation of data centers, model training, and the production of computing hardware. Methane and other greenhouse gases are also not accounted for, so the study mainly concerns CO emissions.
At the same time, the authors believe that assessing the climate impact of artificial intelligence should not be limited to the electricity consumption of data centers. It is also necessary to account for the emissions that the technology indirectly creates by making fossil fuel production cheaper and increasing its output.
Based on articles: npj Climate Action, The Guardian
