A quiet energy revolution is unfolding in Appalachia, where natural gas from the Marcellus Shale is being turned into power for artificial intelligence data centers. Developers are converting old power plant sites into massive gas fired stations and building new plants that are directly wired to sprawling data center campuses, all aimed at tapping into the surging demand for electricity to run generative AI.
In April, Homer City, Pennsylvania, announced an ambitious project to resurrect its retired coal plant. Instead of coal, the facility will burn gas to generate 4.5 gigawatts of electricity. That power will feed a 3,200 acre data center campus, igniting what could be the largest power for AI deal yet. The gas supply will flow from the Marcellus, roughly 700 million cubic feet per day. The decision sidesteps political blockades that have prevented new interstate pipelines from expanding takeaway capacity from the basin.
That pipeline logjam has capped output at under 40 billion cubic feet per day. Yet analysts estimate the Marcellus could support up to 60 billion cubic feet per day if given enough export infrastructure. By using gas close to home, the region keeps more value local, avoids tolls, and enables producers to capture more profit.
The energy companies at the heart of this move are not small operators. EQT and the newly formed Expand Energy, born from the merger of Chesapeake Energy and Southwestern Energy, now both boast triple investment grade credit ratings. That financial standing is a key advantage. Data center operators are unlikely to partner with producers who lack rock solid creditworthiness. AI data centers represent massive investments, with one gigawatt of capacity potentially powering a facility using one million Nvidia H100 chips, hardware that could cost tens of billions of dollars.
EQT’s leadership sees more than just opportunity. Their gas is certified net zero on Scopes 1 and 2, making it attractive to hyperscalers sensitive to emissions. Moreover, they control both production and midstream — upstream drilling and processing of gas. This offers a simplified one stop solution for data centers that want reliable, clean, and direct energy delivery.
What’s Driving the Market and What Lies Ahead
Momentum is building fast. Expand Energy reported that one of its wells recently came online generating 87 million cubic feet per day in just the first month, a staggering figure for a mature basin. Producers need outlets for this surplus gas, and power for AI projects offer a lucrative answer.
Pipeline operators like Williams and Energy Transfer are discussing direct supply deals to power plants, but cannot supply the gas itself. That makes producers with midstream control, like EQT, uniquely positioned to lead. EQT even bought Olympics Energy for $1.8 billion, reinforcing its infrastructure and boosting its ability to meet big power demands.
But this race raises important questions: how fast can these power plants be built, and who will pay? Data center developers will likely opt for fixed price or index plus deals to manage their risks. They want energy deals tied to investment grade producers for stability. Large utilities are also exploring behind the meter arrangements, self contained setups that bypass traditional power grids and remain unaffected by public grid outages or regulatory hurdles.
Speed matters. Investors and developers heard a signal with OpenAI’s “Stargate” announcement, plans to invest $500 billion in AI infrastructure over the next four years. That sparked an immediate scramble among gas producers and utilities to put power structures in place quickly. Logistically, building a 4.5 GW gas fired plant and connecting it to a data center takes years. So developers are racing now to secure gas and turbine arrangements ahead of demand peaks.
The economics appear strong. Without long haul pipeline fees, producers capture more of the revenue stream. Data center operators secure utility scale power with reliable delivery and have local access to rich Marcellus supplies. It is a regional win win in a way more confined than traditional gas export models.
Still, there are potential roadblocks. Local opposition is rising in some Texas regions where gas powered data center plants are planned. Environmental advocates point to emissions of nitrogen oxide, ammonia, particulate matter, and hazardous air pollutants. Bright lights, industrial noise, and constant operations also stir concerns in rural communities accustomed to quiet. Planners may face local permitting delays or pushback from landowners.
Another key issue is balancing emissions with sustainability goals. While gas is cleaner than coal, questions remain regarding methane leaks and lifecycle carbon impact. EQT and Expand Energy have pledged aggressive emissions reductions. One plant in New York aims to reach net zero by 2035. But AI data center energy needs are growing faster than renewables or electrification can meet. Unless alternatives like nuclear, geothermal, or long duration storage catch up, gas power may remain essential.
What happens next depends on investment timing and scale. Gas fired plants and data campuses will require multiple billion dollar commitments each. Rival projects in western Pennsylvania and Texas are already under discussion, with detailed proposals emerging for three to five gigawatt scale facilities. Developers, utilities, and large tech companies will be watching costs, carbon profiles, and credit risks.
For the Marcellus basin, AI powered demand could redefine its future. Expansion could drive more well drilling and midstream growth. It could sustain local economies previously tied to LNG exports or traditional power markets, economies now supported by data center revenue streams instead. But success hinges on market signals holding up: data center operators must commit, gas producers must maintain green credentials, and permitting must move forward smoothly.
Ultimately, this may represent a new chapter for Appalachian energy. The region is becoming both source and power provider for the AI age. That dual role elevates its strategic importance in a global energy landscape where data demand is surging. If managed well, with attention to emissions, community concerns, and economic partnerships, Marcellus and its producers could lead the nation’s next energy revolution.
