Skip to main content
  • Home
  • Tech
  • AI Boom Deepens U.S. Power-Grid Bottlenecks as Aging Infrastructure and Equipment Shortages Spur Push for Greater Efficiency

AI Boom Deepens U.S. Power-Grid Bottlenecks as Aging Infrastructure and Equipment Shortages Spur Push for Greater Efficiency

Picture

Member for

1 year 2 months
Real name
Oliver Griffin
Bio
[email protected]

Oliver Griffin is a policy and tech reporter at The Economy, focusing on the intersection of artificial intelligence, government regulation, and macroeconomic strategy. Based in Dublin, Oliver has reported extensively on European Union policy shifts and their ripple effects across global markets. Prior to joining The Economy, he covered technology policy for an international think tank, producing research cited by major institutions, including the OECD and IMF. Oliver studied political economy at Trinity College Dublin and later completed a master’s in data journalism at Columbia University. His reporting blends field interviews with rigorous statistical analysis, offering readers a nuanced understanding of how policy decisions shape industries and everyday lives. Beyond his newsroom work, Oliver contributes op-eds on ethics in AI and has been a guest commentator on BBC World and CNBC Europe.

Modified

AI data centers drive surging electricity demand, lifting U.S. power consumption and wholesale prices
“Aging transmission networks and transformer shortages” deepen bottlenecks in U.S. power infrastructure
U.S. industry and government mobilize to address the crisis by improving the efficiency of existing generating facilities

U.S. electricity consumption is forecast to set successive records this year and next. Rapid growth in commercial electricity demand, led by artificial intelligence (AI) data centers, is intensifying upward pressure on power consumption and prices. Against this backdrop, market observers warn that the United States faces a severe shortage of the critical power infrastructure needed to accommodate this demand growth. Aging equipment, supply-chain bottlenecks and grid-interconnection delays are widening the lag between the construction of new generating facilities and the delivery of electricity to the grid.

U.S. Electricity Consumption Surges

According to the Short-Term Energy Outlook (STEO) released by the U.S. Energy Information Administration (EIA) on October 6, with all dates hereafter given in local time, U.S. electricity consumption is projected to rise from last year’s record 4.195 trillion kilowatt-hours (kWh) to 4.288 trillion kWh this year and 4.356 trillion kWh next year. Commercial electricity sales, which include data centers, are forecast to reach 1.549 trillion kWh this year, up 3.8% from last year’s record. That is more than twice the projected growth rate for residential sales of 1.7%. The EIA expects commercial demand to expand by a further 2.8% next year.

Wholesale electricity prices are also expected to rise alongside demand. The EIA estimates that extreme weather will push average U.S. wholesale electricity prices to $52 per megawatt-hour (MWh) this year, an increase of approximately 11% from last year. Heavy snowfall and cold snaps in winter, together with summer heat well above seasonal norms, are seen as contributing to the increase. Regionally, wholesale electricity prices are projected to surge 41% in the PJM service territory, which supplies electricity to 13 eastern states, including Virginia, a major data center hub.

Power-Grid Crisis Comes into Focus

Concerns are mounting that the U.S. power grid will be unable to withstand this demand. Calvin Butler, chief executive officer (CEO) of Exelon, the largest U.S. electric utility by customer count, likened the state of the country’s power grid to “a car on the verge of breaking down” at the Fortune Brainstorm AI conference in San Francisco late last year. “It is like driving along when the check-engine light comes on and deciding to keep going a little longer instead of taking the car to a repair shop,” Butler said. “Then nobody pays attention until it finally breaks down.” He added: “On the hottest day or the coldest day, there will come a moment when the electricity supply falls short. That is when people will actually suffer, so we need to fix it now.”

Underlying these concerns are structural imbalances in U.S. power infrastructure. In its National Transmission Needs Study released this year, the U.S. Department of Energy (DOE) concluded that “surging electricity demand, led by data centers and manufacturing, is exacerbating transmission-capacity constraints and congestion.” Indeed, the expansion of U.S. high-voltage transmission infrastructure has lagged demand growth. An analysis of Federal Energy Regulatory Commission (FERC) data by grid consultancy Grid Strategies found that new high-voltage transmission-line construction in the United States fell from an annual average of approximately 1,700 miles in the first half of the 2010s to 645 miles in the second half. In 2024, construction of new extra-high-voltage transmission lines rated at 345 kilovolts (kV) or above likewise amounted to only a few hundred miles.

Aging Infrastructure Poses a Major Challenge

Delays in building new transmission networks place additional strain on existing power infrastructure that is already aging. According to the DOE, more than 70% of U.S. transmission lines and power transformers have been in service for over 25 years. Large power transformers (LPTs), which are essential for long-distance transmission, typically have a design life of approximately 40 years, yet the average age of units operating in North America was already estimated at 38–40 years in 2014. Conditions are similar at the distribution level. In 2024, the DOE and the National Renewable Energy Laboratory (NREL) found that approximately 55% of distribution transformers operating in the United States were at least 33 years old. As large new sources of electricity demand, including data centers and electric-vehicle charging facilities, proliferate, heavier loads on existing equipment could accelerate deterioration and failures.

Replacing aging equipment within a short period is also difficult because supply chains for critical electrical equipment are struggling to keep pace with demand. Shortages of high-capacity transformers are particularly acute. According to energy consultancy Wood Mackenzie, U.S. demand for generator step-up transformers (GSUs) surged 274% between 2019 and last year, while demand for substation power transformers rose 116% over the same period. However, large power transformers are custom-built to meet the voltage and capacity requirements of individual power plants and substations, and supplies of essential manufacturing inputs, including grain-oriented electrical steel (GOES) and copper, are constrained. As a result, lead times for large transformers used in substations and power plants have reportedly lengthened to three to four years.

Table 1. Structural Constraints on the U.S. Power Grid

AreaChallenge
Delays in transmission expansionSlower construction of new high-voltage transmission lines exacerbates capacity constraints and grid congestion
Aging equipmentA substantial share of transmission lines and transformers is approaching its design life, increasing the risk of failures and performance deterioration as new electricity demand grows
Equipment supply-chain bottlenecksCustom manufacturing requirements for large transformers and shortages of essential materials prolong lead times
Grid-interconnection backlogDelays in transmission upgrades and system-impact studies increase the volume of generation and energy-storage capacity awaiting connection
Sources: U.S. Department of Energy, Federal Energy Regulatory Commission, National Renewable Energy Laboratory, Lawrence Berkeley National Laboratory, Wood Mackenzie

New Generating Facilities Face Sharp Delays in Entering Service

These supply-chain disruptions are also delaying the actual commissioning of generating facilities. According to Lawrence Berkeley National Laboratory (LBNL), approximately 8,200 generation and energy-storage projects in the United States had applied for transmission-grid connections and remained in the queue at the end of last year, representing a combined capacity of 2,061 gigawatts (GW). Generation accounted for 1,312 GW, while batteries and other energy-storage facilities accounted for 749 GW. That total exceeds the capacity of the entire existing U.S. generation fleet. The backlog reflects the requirement for new power plants to undergo system-impact studies and grid-reinforcement work before connecting to the transmission network. Grid operators assess whether existing transmission lines and substations can accommodate the additional power from new facilities and, where necessary, require transmission-line additions, substation expansion and transformer replacements.

The time between a project’s interconnection application and the start of commercial operations is also increasing rapidly. According to LBNL, projects that entered commercial service in the early 2000s had a median interval of less than two years between application and operation, whereas projects completed in 2025 took more than five years. The median time from interconnection application to agreement alone exceeded three years. At the end of last year, generation and energy-storage facilities with draft or executed interconnection agreements that had yet to enter commercial operation totaled 549 GW.

Efforts to Upgrade Existing Infrastructure

As the U.S. power-grid crisis becomes increasingly apparent, efforts to alleviate energy bottlenecks are continuing across the domestic market. Google, for example, announced on October 6 that it had signed a 20-year power purchase agreement (PPA) with Constellation Energy to upgrade 11 reactors at six U.S. nuclear power plants, adding 890 megawatts (MW) of generating capacity. The plan is to expand electricity supply by improving existing facilities rather than building new nuclear plants. Constellation will replace or upgrade turbines, steam generators and digital control systems to improve the reactors’ thermal and electricity-generation efficiency. Google also formally launched the energy coalition Utilize in March alongside Tesla and other major companies. The coalition aims to increase utilization of the U.S. electricity system, simultaneously easing the pressures of rising electricity consumption and power bills. Its plan is to bring companies, public agencies and policymakers together to advance policies and technologies that improve the operation of existing infrastructure.

The government is also pursuing technological innovation to extend the operating lives of existing generating facilities and improve generation efficiency. On October 5, the DOE’s Hydropower and Hydrokinetic Office (H2O) announced that it had selected 14 projects for the second round of technical support through the Hydropower Testing Network (HyTN). The program draws on 21 testing facilities and more than 65 specialized testing and analytical capabilities, helping companies validate the performance and safety of new technologies at facilities operated by national laboratories and other institutions. Distributed hydropower technology company Emrgy, for example, will test a low-head turbine capable of generating electricity from existing artificial waterways such as irrigation canals, while modular hydropower startup GenH will validate hydropower technology that can be rapidly installed at existing non-powered dams. RCAM Technologies, a developer of 3D concrete-printing technology for renewable energy applications, is developing pumped-storage energy technology that uses 3D-printed concrete spheres installed on the seafloor.

Picture

Member for

1 year 2 months
Real name
Oliver Griffin
Bio
[email protected]

Oliver Griffin is a policy and tech reporter at The Economy, focusing on the intersection of artificial intelligence, government regulation, and macroeconomic strategy. Based in Dublin, Oliver has reported extensively on European Union policy shifts and their ripple effects across global markets. Prior to joining The Economy, he covered technology policy for an international think tank, producing research cited by major institutions, including the OECD and IMF. Oliver studied political economy at Trinity College Dublin and later completed a master’s in data journalism at Columbia University. His reporting blends field interviews with rigorous statistical analysis, offering readers a nuanced understanding of how policy decisions shape industries and everyday lives. Beyond his newsroom work, Oliver contributes op-eds on ethics in AI and has been a guest commentator on BBC World and CNBC Europe.