ERIE, Pa. — The “modern-day gold rush” is here in Appalachia, according to oil and gas industry leaders. But it’s not gold they are referring to, it’s artificial intelligence.
Rural areas are particularly “compelling” for AI data center development as they provide lower startup and operational costs, Babst Callard officials said at the 2025 Shale Insight Conference, held in Erie, Pennsylvania.
AI was the main topic of discussion on Sept. 18 at the annual conference where industry leaders from the Marcellus Shale Coalition discussed how Pennsylvania, Ohio and West Virginia are the ideal locations for AI data centers due to the states’ abundant supplies of natural gas reserves.
The conference featured information on developing AI data centers as well as studies on how the emerging technology could make the oil and gas industry safer and more efficient.
Why AI?
Natural gas is necessary to power AI data centers, according to industry leaders who spoke at the conference. These data centers require massive amounts of energy to operate 24/7, and, because of this, many will be accompanied by their own natural gas power plants.
Power demands for individual AI data centers can exceed 100 megawatts or more, consuming as much energy annually as 100,000 households, reports the International Energy Agency.
“In the United States, power consumption by data centers is on course to account for almost half of the growth in electricity demand between now and 2030,” according to a recent International Energy Agency report, consuming more energy than all manufacturing industries in the U.S. combined, including aluminum, steel, cement and chemicals.
The Trump administration has put a particular focus on the development of AI data centers.
President Donald Trump signed an executive order in January, “Removing Barriers to American Leadership in AI,” followed by policy guidelines in July called “Winning the AI Race: America’s AI Action Plan.”
The policy outlines the Trump administration’s goals for AI data centers, including expediting permits and removing AI regulations that hinder its development.
Discussions at the Shale Insight Conference revolved around Ohio, Pennsylvania and West Virginia being prime spots for AI data center buildout.
Trump announced a $90 billion investment in AI data centers across Pennsylvania at Sen. Dave McCormick’s Pennsylvania Energy and Innovation Summit in Pittsburgh on July 15.
According to Trump, AI data centers will be accompanied by power plants of all kinds, from natural gas, nuclear and “clean, beautiful coal.”
Central Ohio is also seeing significant growth in AI data centers. Amazon announced in December 2024 that it would invest an additional $10 billion to expand its data center infrastructure.
Columbus is a particular hotspot; the city is home to five Cologix data centers, and data centers from mega corporations Google, Amazon and Meta (Facebook) in New Albany. According to Data Center Maps, Ohio has 191 data centers — the fifth state with the most data centers.
Babst Calland, a law firm that represents energy companies, noted that data centers are the backbone of businesses dealing with finance, healthcare and e-commerce.
Soon, AI data centers could also provide the oil and gas industry with a particular advantage.
AI in the oil and gas industry
The conference featured information on recent and ongoing studies, including two that utilize AI in oil and gas operations, from West Virginia, Ohio and Pennsylvania universities.
One study, titled “Next Generation AI Tools for Enhancing Flow Assurance and Leak Detection in Shale Pipeline Networks”, looks at how AI could be used to detect leaks and corrosion in pipelines — a common issue in the United States.
Ali Sajedian, a Ph.D. candidate at West Virginia University, a petroleum engineer and author of the study, developed the AI software known as MidStreamAI using data collected from a pipeline in the Middle East.
Sajedian’s program analyzes pipeline data to determine leaks or corroding pipelines in the present moment or predict when they might happen in the future.
“For example, you have some corrosion, based on the data, you know that the leakage will happen after six months,” Sajedian said.
He adds that MidStreamAI can also detect problems in pipeline separators and valves, and analyze geological data. According to Sajedian, the AI software is still in its early stages, but he has had several talks with oil and gas companies that have expressed interest.
AI and CCS
Other research, like “Expanding the scope of CCSU: Evaluating the feasibility of CO2 Geo-sequestration in Tight and Unconventional Formations,” looked at how AI could be used to determine suitable rock formations for carbon capture and sequestration, a technique that consists of capturing carbon dioxide emissions and storing them in rock formations.
In order for CCS to be effective, petroleum engineers need to find the “sweet spot” within rock formations, said Prince Henry Sampson Eduam, a Ph.D. student at Ohio State University and co-author of the study.
Carbon cannot be injected into rock formations with low-permeability and low-porosity, as it is hard to inject into these rocks, but cannot be injected into some formations with high-permeability and high-porosity either, as the CO2 could leak, he said.
“How exactly do we find that balance between optimizing injection and security at the same time? That is where this research (comes in),” Eduam said.
Eduam used rocks from the Marcellus shale formation to conduct experiments that mimicked injecting carbon into the formation. He found that the high organic content in the Marcellus shale causes kerogen in the rock — solid organic matter made up of dead organisms — to swell when it comes into contact with CO2, trapping the gas in place.
This data was then used to train his AI software PetroAI, which, when completed, will give oil and gas companies a “predictive analysis” of the best uses for a specific geological formation.
“The AI, based on its learning models, is going to tell you about (rock formation) information and its characteristics and how best it’s going to help whatever you have chosen, whether to sequester CO2 or whether to be hydrogen storage,” Eduam said.
The AI software will also give companies a cost analysis and the success rate of operations. But Eduam notes that in order for PetroAI to be used widely, he needs to collect data on more rock formations to train the software.
“We need to get a lot of rocks as possible, and then run the experiments, use the data to train the AI, so that we can have large data sets, and the AI can keep learning to make it perfect,” Eduam said. He is currently looking for funding to continue his research.
(Liz Partsch can be reached at epartsch@farmanddairy.com or 330-337-3419.)
