undefined
Posts
undefined
Followers
Why AI may boost carbon emissions instead of cutting them
By Paul Arnold,
2026-08-12AI data centers get a bad rap, not least because of concerns that they drive climate change by consuming massive amounts of electricity. But one potentially larger impact of AI on global carbon emissions may be that it's helping the fossil fuel industry become more productive. That's the main finding of a paper published in the journal npj Climate Action that looked at how boosting productivity across both clean and dirty energy sources affects net CO2 emissions.
Modeling the global economy
Researchers in the U.S. built a complex computer model of the global economy to compare AI-driven productivity gains in clean power generation with those in fossil fuel production. They tested 64 different scenarios and applied various levels of AI adoption, from low to high, to fossil fuels, renewable energy, power grids, and selected areas of shipping and heavy manufacturing.
According to the model's scenarios, when AI productivity gains are applied across both fossil fuel and clean energy sectors, global annual CO2 emissions rise by 0.47 to 1.8 billion tonnes (0.52 to 2 billion tons). That increase roughly equals 1.2% to 4.8% of total global energy-related CO2 emissions in 2024. So what is AI doing to help fossil fuel productivity? The main drivers included its ability to increase the productivity of fossil fuel extraction and lower production costs.
Renewable energy gap
Reversing this surge would be a tall order. The research team reports that AI-driven improvements in renewable energy would need to be four to five times larger than those in the fossil fuel sector to break even.
"Under modeled conditions, AI-driven productivity gains enable more emissions than they avoid—reinforcing fossil fuel incumbency rather than displacing it," wrote the study authors.
This happens because AI boosts existing industries rather than changing how the economy works. "AI does not inherently decarbonize or intensify emissions; its productivity gains reinforce the economic and institutional structures of the system in which it operates."
Key priorities
The paper doesn't just highlight the problem; it also suggests a number of solutions.
The researchers suggest five key priorities to tackle the problem. First, policymakers should constrain AI-enabled fossil production by placing direct limits on tech tools that boost fossil fuel extraction. Another suggestion is tracking the broader climate impacts of AI applications, rather than focusing only on data center electricity use.
The study also recommends combining support for green AI with limits on fossil fuels and recognizing both the emissions AI saves and the extra emissions it enables. There should be carbon pricing policies in place, though the authors stress that carbon pricing alone cannot close the gap.
More information: Will Alpine et al, AI-driven productivity gains enable more CO₂ emissions than they avoid in a global energy–economy model, npj Climate Action (2026). DOI: 10.1038/s44168-026-00411-0
© 2026 Science X Network