[[{“value”:”
The U.S. energy landscape is at a crossroads. Aging coal and natural gas power plants, once slated for retirement, are being propped up with temporary fixes to meet surging electricity demand, particularly from the booming AI data center sector. These “band-aid” solutions—ranging from delayed retirements to innovative turbine workarounds—are reshaping the power grid. But at what cost to reliability, efficiency, and the environment? In this deep dive, we explore the retirement extensions of coal and natural gas plants, the turbine shortage crisis, creative solutions like smaller jet turbines, and how these dynamics are influencing AI data center builds across the U.S.
The Retirement Rollercoaster: Coal and Natural Gas Plants Clinging to Life
Coal and natural gas plants have been the backbone of U.S. electricity for decades, but many are reaching the end of their operational lives. Economic pressures, stricter environmental regulations, and competition from cheaper renewables and natural gas have driven utilities to plan retirements. However, the explosive growth in electricity demand—largely fueled by AI data centers, electrification, and industrial expansion—has forced utilities to rethink these plans.
According to the U.S. Energy Information Administration (EIA), 12.3 gigawatts (GW) of power plant capacity is scheduled for retirement in 2025, a 65% increase over 2024’s 7.5 GW. Coal plants account for 66% of these planned retirements (8.1 GW), while natural gas makes up 21% (2.6 GW). Notable coal plants include the 1,800-MW Intermountain Power Project in Utah and the 1,331-MW J H Campbell in Michigan. Natural gas retirements include older, less efficient simple-cycle turbines like the 859-MW V H Braunig units in Texas.
Yet, many plants slated for retirement are getting last-minute reprieves. The EIA notes that coal retirements dropped to 4.0 GW in 2024, the lowest in a decade, as utilities delayed closures to meet demand. For example, in the Kansas City area, a data center and an electric-vehicle battery factory prompted the local utility to postpone a coal plant’s retirement. In Virginia, Dominion Energy has warned of reliability risks if coal and gas plants are phased out too quickly, citing 18 load relief warnings in 2022 alone.
Natural gas plants are also seeing extensions, particularly in regions with strained grids. EQT Corp., a major natural gas producer, estimates that coal plant retirements and data center demand could boost U.S. natural gas demand by up to 10 billion cubic feet per day by 2030, with over half coming from the Southeast and PJM Interconnection markets. These extensions are often driven by the lack of immediate alternatives—renewables like solar and wind are intermittent, and new nuclear or gas plants take years to build.
The sweet spot for investment will be the replacement of older natural gas and petroleum plants with natural gas.
Table 1: Planned Power Plant Retirements in 2025 (U.S.)
|
Fuel Type
|
Planned Retirements (GW)
|
Share of Total (%)
|
Notable Plants (MW)
|
|---|---|---|---|
|
Coal
|
8.1
|
66
|
Intermountain (1,800), J H Campbell (1,331)
|
|
Natural Gas
|
2.6
|
21
|
V H Braunig (859), Eddystone (760)
|
|
Petroleum
|
1.6
|
13
|
Herbert A Wagner (828)
|
|
Total
|
12.3
|
100
|
Turbine Shortages: A Bottleneck for Power Expansion
The surge in electricity demand has exposed a critical bottleneck: a shortage of natural gas turbines, the workhorses of modern power plants. Turbine deliveries now face lead times of several years, potentially delaying new gas-fired plants beyond 2030. This scarcity is driven by the convergence of demand for all turbine sizes, from large combined-cycle units to smaller models, as developers race to secure equipment for data centers and grid expansion.
Rich Voorberg, president of Siemens Energy North America, describes the current frenzy for gas turbines as the biggest since the Enron collapse over two decades ago. Even smaller turbines are backlogged, pushing developers to explore unconventional solutions. The limited supply chain for turbines, coupled with environmental permitting delays, complicates efforts to replace aging plants or build new ones.
Workarounds: Smaller Jet Turbines to the Rescue?
To bypass turbine shortages, developers are turning to smaller, off-the-shelf solutions like aeroderivative gas turbines, often derived from jet engine technology. These compact turbines, with capacities as low as 5 megawatts (MW), offer flexibility and faster deployment compared to traditional large-scale turbines. For instance, the Stargate AI data center in Texas, a joint venture by OpenAI, Oracle, and Softbank, is using a dozen small gas turbines to power its 900-acre facility, which requires enough electricity for 300,000 homes.
Companies like Mainspring are capitalizing on this trend, producing units so small that 100 are needed to generate 25 MW. These turbines allow developers to “build in chunks,” scaling power incrementally as data centers expand. Once grid connections are established, these smaller turbines can serve as backup power or provide grid-stabilizing services.
However, these workarounds come with trade-offs. Small, single-cycle gas turbines are less efficient (35–42% thermal efficiency) than combined-cycle plants (up to 62%) and produce higher carbon emissions per unit of electricity. Their reliance on natural gas also raises concerns about long-term climate impacts, especially as AI data centers drive up fossil fuel use. BloombergNEF warns that data center demand is slowing the decline in U.S. carbon emissions by leaning heavily on gas and coal.
Impact on AI Data Center Builds
The AI boom is a game-changer for the energy sector. Data centers are projected to consume 9% of U.S. electricity by 2030, up from 4% in 2022, with AI alone adding 323 terawatt-hours of demand—seven times New York City’s annual consumption. Aging plants and turbine workarounds are critical stopgaps, but they create challenges for data center developers:
-
Reliability Risks: Extended coal and gas plants are prone to outages due to their age (many coal plants average 52 years old). Smaller jet turbines, while flexible, may not match the reliability of large-scale systems, especially for 24/7 data center operations.
-
Environmental Pushback: The carbon footprint of gas-fired turbines, particularly less efficient single-cycle models, conflicts with tech companies’ sustainability pledges. For example, Meta has opted for nuclear power in Illinois, while Google partners with renewables developers.
-
Grid Constraints: Long grid connection queues (4–8 years for new transmission lines) and equipment shortages (e.g., transformers) mean data centers often rely on on-site generation, like jet turbines, which increases costs and complexity.
-
Cost Pressures: Delayed retirements and turbine shortages are driving up wholesale power prices, especially in high-demand regions like Texas’ ERCOT market. This could raise operational costs for data centers, potentially passed on to consumers.
Where in the U.S. Are AI Data Centers Thriving?
The best locations for AI data centers balance access to power, grid reliability, land availability, and proximity to tech hubs. Based on current trends, here are the top regions:
-
Texas (ERCOT): Texas leads with abundant natural gas, a deregulated market, and projects like Stargate in Abilene. However, grid reliability issues and turbine shortages pose risks. Solar capacity is growing (8.9 GW added in 2024), but gas remains dominant.
-
Virginia (“Data Center Alley”): Northern Virginia hosts the world’s largest data center hub, but grid strain led to a temporary pause on new connections in 2022. Dominion Energy is expanding renewables and gas to cope, making it a hotspot despite challenges.
-
Southeast (PJM Interconnection): The PJM market, covering Pennsylvania, Maryland, and others, benefits from Marcellus Shale gas and projects like the Homer City redevelopment, where a retired coal plant is becoming a 4.5-GW gas-powered data center campus.
-
Louisiana: Meta’s Richland Parish data center, powered by Entergy’s new gas turbines, highlights Louisiana’s appeal due to low-cost gas and state incentives. However, high electricity bills and grid unreliability are concerns.
-
Arizona and Nevada: These states offer abundant solar potential and land, but water scarcity and grid limitations make them less ideal for massive AI facilities unless paired with on-site generation.
The Road Ahead: Band-Aids or Breakthroughs?
Aging coal and natural gas plants, propped up by delayed retirements and smaller jet turbines, are a temporary fix for the U.S.’s power crunch. They’re enabling the AI data center boom but at the cost of efficiency, emissions, and long-term grid stability. Developers must navigate turbine shortages, grid delays, and environmental pressures while securing reliable power.
The best path forward lies in a diversified energy strategy. Tech giants are already exploring small modular nuclear reactors (SMRs), with 1.5 GW expected by 2030, and renewables paired with battery storage. Innovations like carbon capture and hydrogen-ready turbines could mitigate emissions from gas plants. Meanwhile, utilities must modernize grids and streamline permitting to reduce reliance on band-aid solutions.
For AI data centers, Texas, Virginia, and the Southeast offer the strongest opportunities, but success hinges on balancing immediate power needs with sustainable growth. As the energy sector grapples with this unprecedented demand, one thing is clear: the future of AI depends on the resilience of America’s power infrastructure.
Sources:
-
U.S. Energy Information Administration (EIA)
-
Bloomberg
-
AP News
-
MIT Technology Review
-
Reuters
-
Energy News Beat
-
International Energy Agency (IEA)
-
Wells Fargo, CNBC
-
BNP Paribas
-
Marketplace
The post How Aging Coal and Natural Gas Plants Are Kept Going with Band-Aids appeared first on Energy News Beat.
“}]]
Energy News Beat