AI Was Supposed to Cut Emissions. Instead, Fossil Fuel Companies Are Using It to Find More Oil and Gas.

When United Nations Secretary-General António Guterres warned world leaders last week in his UN address about the unchecked power of both Big Oil and artificial intelligence, he treated them as two of the defining challenges facing governments.

The fossil fuel industry is turning to AI to map seabeds, identify new oil reservoirs, assess oil wells’ “abandonment risks”, inspect facilities, and streamline regulatory processes, potentially adding to the emissions of an industry that Guterres said governments must move away from. 

At the same time, the technology companies providing these AI tools are not generally held accountable for the additional fossil fuel production and emissions their products may enable.

“The danger is not technology,” Guterres told the UN General Assembly. “The danger is technology without accountability.”

A DeSmog review of company announcements, industry presentations, and other publicly available information found that, across major fossil fuel companies — Aramco, Chevron, Equinor, ExxonMobil, Shell, TotalEnergies and others — AI is increasingly being integrated into fossil fuel operations, from seismic exploration and drilling to equipment monitoring and field development.

The examples provide a real-world look at the mechanism behind a recent study published in Nature: that AI can increase emissions not only because of the electricity required to run data centers, but because it can make fossil fuel production more productive and profitable.

The study found that AI-driven productivity gains in fossil fuel production accounted for most of the modeled increase in global energy-related carbon dioxide emissions. It estimated that AI could increase global energy-related carbon dioxide emissions by between 1.2 and 4.8 percent, with productivity gains in fossil fuel production accounting for most of that increase. 

In one example of AI gains, Norwegian energy company Equinor re-scanned the Norwegian continental shelf with “new seismic technology and AI,” resulting in 27 new discoveries of oil, Hege Skryseth, the company’s executive vice president, chief technology officer, said during a June presentation for Equinor’s Capital Market Day.

“By reprocessing the data, the seismic image became much sharper, leading to the discoveries,” Skryseth said. “AI was key here, from automated data interpretation to efficient well planning.”

AI saved Equinor $130 million last year, according to a January 2026 press release.

Chevron is also using AI to produce oil that was “previously considered unreachable.” 

In the oil company’s recently launched podcast — in which racing driver Tanner Foust and Chevron Chief Technology and Engineering Officer Ryder Booth chat from a pickup truck tailgate — Booth said that breakthroughs at Anchor, an oil and gas development, represented the “first time unlocking kind of a new horizon around the world.”

“AI, it’s attracted to big data and big opportunities, and the oil and gas industry has big data,” Booth said. Last year, Chevron said it had improved drilling and fracking efficiency in the Permian Basin by over 30 percent.

The value of these productivity gains is significant. A recent McKinsey & Co. report estimated that, for the upstream oil and gas sector, AI can generate $65 billion in profits annually in the short term, which could increase to $230 billion when the technology is deployed at its “full potential.”

AI-driven efficiency gains have already reduced operational costs by up to 18 percent; and industry executives attribute about 5 percent of their current revenue to “AI-driven initiatives,” according to a 2025 report from IBM, which has multiple partnerships with oil and gas companies.

‘Done in seconds and live-streamed’

At Equinor’s enormous Johan Sverdrup oil field in the North Sea, “data flows up from the ground faster than the oil,” the company’s site states, at the rate of “10,000 Netflix movies a second.”

The company is also using AI to optimize the placement of subsea equipment and wells.

“What used to take 2-3 weeks and involved flying hard disks to shore with helicopter for analysis can now be done in seconds and live-streamed,” Equinor said.

ExxonMobil is also embracing enormous datasets. In 2019, ExxonMobil partnered with Microsoft to use AI in its Permian Basin operations, a move Exxon said would “generate billions of dollars in value” and produce up to an additional “50,000 oil-equivalent barrels a day by 2025.”

The oil giant also said that gains in efficiency from advanced technologies, including AI, “is a first step” towards letting systems respond to events without human intervention.

Alpine said the significance of such partnerships is easy to miss when attention focuses on the emissions produced by powering AI systems.

“For some reason, tech companies are exempt from accountability for the tech they are creating,” she told DeSmog.

Alpine and her husband, Will Alpine, another author of the Nature study, previously worked for Microsoft, but left after coming to the conclusion that the company hid how its products were harming the climate. The couple then founded the Enabled Emissions Campaign, a nonprofit that calls attention to tech’s role in fossil fuel production and advocates for better regulations.

She said that now, oil companies like Exxon pumping 50,000 barrels a day is “actually quite a small deal.” 

“The deals have only grown in size, and we realized that that was not part of Microsoft’s accountability whatsoever, even though they are the company creating the technology that is making this happen.”

A recent Reuters article about Exxon’s use of automated drilling in the Permian Basin, which said the company plans to increase production in the region to 2.5 million barrels a day, didn’t mention if Microsoft was still involved with the project; on Exxon’s website, references to Microsoft partnerships still link to the announcement from 2019.

Microsoft did not confirm whether it’s still partnering with Exxon on those operations, but the tech company is the leading cloud provider for the oil and gas industry, according to Kimberlite Research, followed by Amazon and Google. 

Data Centers Are Intertwined with AI

Just two years ago, fossil fuels’ share of the world’s energy supply was projected to decline in coming years. Then, AI data centers began expanding rapidly, and with them, the argument that fossil fuels must scale up, too, for the U.S. to win the “AI arms race.”

At the same time, the fossil fuel and technology industries are becoming increasingly intertwined. AI data centers require enormous amounts of electricity, and natural gas supplies more than 40 percent of the power used by U.S. data centers, according to the International Energy Agency (IEA).

“AI’s advance will depend not only on the design labs of Silicon Valley, but also on the gas fields of the Permian Basin,” Mike Wirth, Chevron’s chairman and CEO, said in 2024.

In the face of power grid constraints, some data centers are turning to natural-gas power plants built onsite, a setup known as “behind-the-meter” generation, like Chevron’s Kilby project with Microsoft. U.S. data centers’ investments in new gas turbines recently surpassed the total investments of every country except the U.S., according to an IEA report. 

Meanwhile, the oil and gas companies are benefiting both from using AI to increase efficiency, and from supplying the additional energy that its data centers demand.

AI and Net Zero Aims

Online, Microsoft promotes AI tools as a way to “accelerate the energy transition,” “reduce emissions,” and “accelerate climate innovation.”

Instead, AI may be having the opposite effect, prolonging our reliance on fossil fuels and delaying a broader transition to cleaner energy.

“In the absence of continuous reinvestment, oil and gas production would fall by approximately 8 percent annually,” the Nature paper says, quoting IEA reports, and “technological progress has repeatedly delayed ‘peak oil’ forecasts by unlocking resources previously considered un-viable.”

Equinor’s website recognizes the growing scarcity of oil reserves, noting that undersea oil and gas discoveries have dramatically declined in recent decades. “Even with planned projects, the (production) trend is downward,” it states. “Gently at first, but more noticeably if we don’t act fast to create new resource opportunities. This will affect value creation across Norway.”

Chevron’s Booth, meanwhile, says there’s still enough oil, “and technology and innovation will unlock that,” though he also admitted we would need all forms of energy to meet the world’s demand.

In spite of concerns about future scarcity, wartime windfalls and beneficial policies have recently brought record profits for major oil and gas companies, news reports show.

After the August Nature article showing that AI helps the fossil fuel industry create more pollution, Johanna Fornberg, a Greenpeace senior research specialist, said in a statement that, while Big Tech continues to promise that AI technologies will benefit humanity and offer climate solutions, “what is hidden from that story is the aggressive support tech companies provide to fossil fuel companies to continue producing oil and gas that harms communities and the climate.”

Microsoft says it plans to “become carbon negative by 2030,” though the company’s emissions increased by 25 percent last year, driven in part by expanding data centers, according to its 2026 Environmental Sustainability Report.

Microsoft’s 2025 Responsible AI Transparency Report also said that its enterprise contracts “incorporate our AI Services Code of Conduct, which requires our customers to implement responsible practices (such as human oversight and access controls) and prohibits using our AI services in ways that inflict harm on individuals, organizations, or society, or affects individuals in any way that is otherwise prohibited by law.”

Alpine said very little action is being taken to regulate the use of AI’s climate impacts. Microsoft announced in 2022 that it would only work in certain capacities with companies with net zero targets, she said, but “they don’t follow any standards for net zero.”

That principle, which is listed on Microsoft’s website, says the company may provide “technical or engineering resources to develop or co-develop specialized subsurface exploration and extraction services” or provide products at no cost to energy customers with goals to reach net zero Scope 1 and 2 carbon emissions by 2050.

Exxon and Chevron have both walked back their plans to reach net zero by 2050, saying that necessary advancements in technology and policy that are beyond their control haven’t materialized. Exxon now says it’s on track to achieve net zero across its Permian Basin operations by 2035, while Chevron “continues to have the aspiration” to achieve net zero, but without a timeline.

Microsoft did not respond to a question about whether the changes to Exxon’s and Chevron’s net zero goals has affected the companies’ agreements. Chevron, Equinor, and Exxon also did not respond to requests for comment.

Equinor says it intends to become net zero by 2050 “as part of its commitment to do zero harm to people, the environment, and material assets.”