US Building Twice as Much Gas-Fired Power as China to Fuel AI Boom
A surge in data center construction has driven a 50% increase in planned US natural gas power plants in just six months. The unprecedented buildout is straining local grids and raising questions about future electricity costs for homeowners.
By Derya Kaplan
How this story has developed
This report is part of a developing story — read the earlier chapters below.
- Global Data Center Power Consumption Reaches 2% of World Electricity and 6% of US Grid
- Is the AI Boom a Hidden Tax on Your Utility Bill and the Climate?
- Brookfield and NextEra Plan $100 Billion AI Campus at Former DOE Site With Dedicated Power
- Is the AI-Driven Power Demand Forcing a Quiet Retreat from US Climate Goals?
- US Data Center Power Demand Forecast to Consume 20% of US Electricity by 2035
- Is the AI Power Crisis Forcing the US to Finally Embrace Nuclear Energy as the Only 'Green' Solution?
- US Building Twice as Much Gas-Fired Power as China to Fuel AI Boom (this article)
- Tech & Utility Sector
- Prioritizes rapid, reliable energy deployment to support AI infrastructure.
- Consumer & Community Advocates
- Focuses on protecting homeowners from socialized infrastructure costs and local disruptions.
- Environmental Analysts
- Warns against the long-term climate and financial risks of a massive fossil fuel buildout.
The United States is currently undergoing an energy infrastructure shift that few anticipated even a year ago, driven almost entirely by the rapid expansion of artificial intelligence. According to a new analysis released by Global Energy Monitor, the US is now building twice as much natural gas-fired power capacity as China. This represents a stark reversal of historical trends. For decades, China's rapid economic growth meant it consistently outpaced the US in adding new fossil-fuel generation. However, the sudden frenzy to build massive data centers has completely upended that dynamic, pushing the US to the forefront of global natural gas development and sparking new debates over local grid impacts. The scale of the proposed buildout is staggering, with direct implications for regional power grids. Between January and June of 2026, the total capacity of US gas power projects in development ballooned by 50 percent, rising from 252 gigawatts to 378 gigawatts.[1]
To put that massive figure into perspective for a residential consumer, a single gigawatt of electricity is generally enough to power roughly 750,000 average homes. If all of these proposed facilities are ultimately constructed, the existing gas-fired power fleet in the United States would grow by roughly two-thirds. The estimated capital cost for this nationwide expansion exceeds $647 billion, representing one of the largest concentrated investments in energy infrastructure in modern history. The mechanism driving this surge is the unique power requirement of generative artificial intelligence. Unlike traditional cloud computing, training and running large language models requires specialized processors that consume immense amounts of electricity and generate significant heat. Tech companies are demanding reliable, always-on power that current renewable infrastructure cannot always guarantee around the clock, prompting a massive pivot back toward fossil fuels to ensure uninterrupted operations.[1][2]
As a result of these intense energy demands, developers are turning back to natural gas as a rapid, dispatchable energy source that can be deployed faster than nuclear and more reliably than wind or solar. Of the 52 gigawatts of gas-fired capacity currently under physical construction in the US, approximately 16.9 gigawatts is intended directly for data centers. In contrast, China currently has only about 24 gigawatts of natural gas power under construction nationwide. For local real estate markets and homeowners, this macro trend translates into immediate neighborhood-level impacts. Texas has emerged as the epicenter of this expansion, accounting for nearly a third of the planned US buildout with over 80 gigawatts in development. Residents in these high-growth zones are increasingly navigating the realities of new industrial neighbors.[1]
The primary concern for residential ratepayers is the potential impact on monthly utility bills. When a utility company builds new transmission lines, substations, or power plants to serve a massive new commercial client, those infrastructure costs are often socialized across the entire ratepayer base, leading to higher baseline costs for everyday families. Jenny Martos, a project manager at Global Energy Monitor, noted that the projects successfully clearing regulatory hurdles are currently paying top dollar for specialized turbines and heavy equipment. This intense competition for electrical infrastructure is pushing up baseline construction costs, which can eventually trickle down to the electricity prices paid by homeowners who are already grappling with inflation in other sectors of the housing and home maintenance market.[1][2]
The primary concern for residential ratepayers is the potential impact on monthly utility bills.
Grid reliability is another major factor for local communities facing this unprecedented buildout. In regions like the PJM Interconnection—which coordinates the movement of wholesale electricity across 13 states and the District of Columbia—utilities have already proposed mechanisms to curtail data center power usage first during periods of peak grid stress. This curtailment strategy is designed to protect residential neighborhoods from rolling blackouts during extreme weather events, such as severe summer heatwaves or winter freezes, when both homes and data centers are pulling maximum power simultaneously. Ensuring that residential heating and cooling systems remain operational while massive server farms draw gigawatts of power has become a central challenge for grid operators and local zoning boards across the country.[1][2]
Despite the massive pipeline of announced projects, significant uncertainty remains about how much of this capacity will actually materialize. The Global Energy Monitor report highlights that more than half of the gas projects tied to data centers do not yet have a named turbine manufacturer, suggesting they are still in the highly speculative phase of development. Securing the physical hardware required to build a modern gas plant is proving to be a bottleneck, as the global supply chain struggles to keep pace with the sudden surge in demand from American tech companies. This equipment shortage could delay many of the proposed plants, providing a temporary reprieve for local grids but extending the timeline of infrastructure uncertainty for nearby communities.[1]
Local opposition is also proving to be a formidable hurdle for developers hoping to break ground quickly. Homeowners and community groups are increasingly pushing back against the zoning and permitting of new gas plants and data centers in their backyards. These residential coalitions frequently cite concerns over construction noise, massive water usage for server cooling in drought-prone areas, and localized emissions from the gas plants. Ultimately, the tension between the speed required by AI developers and the deliberate pace of local permitting, grid capacity planning, and community consent will dictate the final outcome. For now, the US housing market and its underlying infrastructure are being reshaped by a digital boom that requires very physical, very large-scale power solutions.[1][2]
Looking ahead, prospective homebuyers may need to factor grid stability and local data center development into their purchasing decisions. Just as buyers evaluate school districts and property taxes, the proximity of energy-intensive industrial facilities and the resilience of the local power grid are becoming critical variables in real estate. Real estate agents in tech-heavy corridors are already fielding questions about utility rates and the potential for new high-voltage transmission lines. As the artificial intelligence industry continues its rapid expansion, the intersection of global technology trends and local homeownership will only deepen, making energy infrastructure a permanent fixture in the conversation about residential property values.[2]
Key points
- The US is currently building 52 gigawatts of natural gas power, more than double China's 24 gigawatts.
- Nearly 17 gigawatts of the US capacity under construction is intended directly for AI data centers.
- The total pipeline of US gas projects in development surged 50% in the first half of 2026, reaching 378 gigawatts.
- Homeowners face potential impacts including higher utility bills, new transmission lines, and local grid strain.
- More than half of the proposed projects lack a named turbine manufacturer, indicating significant uncertainty.
Why this matters
For homeowners and buyers, the sudden pivot toward massive gas-fired power plants to feed AI data centers signals a coming shift in local infrastructure. As utilities scramble to meet unprecedented commercial demand, residential ratepayers may face new transmission lines in their communities, increased competition for grid capacity, and potential shifts in monthly electricity costs.
Key terms
- Gigawatt (GW)
- A unit of power equal to one billion watts, typically enough to provide electricity to about 750,000 average homes.
- Dispatchable Energy
- Sources of electricity, like natural gas plants, that can be turned on or off quickly to meet fluctuating demand, unlike weather-dependent renewables.
- PJM Interconnection
- A regional transmission organization that coordinates the movement of wholesale electricity in all or parts of 13 US states and Washington, D.C.
- Ratepayer
- A residential or commercial customer who pays for utility services, often bearing the socialized costs of grid upgrades.
Frequently asked
Will the new data centers increase my monthly electric bill?
It is possible. When utilities build new power plants or transmission lines to support commercial clients, those infrastructure costs are often spread across all customers, which can lead to higher residential rates.
Why are they building natural gas plants instead of using solar or wind?
AI data centers require massive amounts of uninterrupted, 24/7 power. Developers argue that current renewable energy and battery storage technology cannot yet guarantee that level of constant baseline power.
How much power does a data center actually use?
A single large-scale AI data center can consume hundreds of megawatts of electricity, rivaling the power demands of a small city or hundreds of thousands of individual homes.
Sources
[1]Global Energy MonitorEnvironmental AnalystsBetting big on data centers, U.S. now leads world for new gas power development
Read on Global Energy Monitor →
[2]WiredConsumer & Community AdvocatesData Centers Are Driving an Alarming Gas Power Expansion in the US
Read on Wired →
Comments
Every angle. Every day.
Get home stories with full source coverage and perspective breakdowns delivered to your inbox.