Global electricity demand is growing faster than the grids that deliver it, and the bottleneck is no longer power plants—it's the transformers and substations that step voltage up and down. According to the International Energy Agency's Electricity 2026 report, worldwide power demand will rise above 3.5 percent annually on average through the decade's end—more than double the rate of the previous ten years. Meeting that load requires annual grid investment to climb roughly 50 percent by 2030 from about $400 billion today, the IEA says. More than 2,500 gigawatts of generation, storage, and large-load projects are already stuck in connection queues around the world, while planning and constructing new transmission still takes 5 to 15 years. Data centers, by contrast, take 1 to 3.
The United States set a new record for electricity use after twenty years of near-stagnation, with data-center load as the main driver of long-term growth. The Energy Information Administration projects U.S. sales growth near 3.2 percent a year through 2035, with data centers accounting for roughly two-thirds of that increase. The IEA's U.S. figure is slightly lower—close to 2 percent annually through 2030, still more than double the prior decade—and attributes about half of advanced-economy demand growth to data centers. Asia-Pacific is expected to represent nearly three-quarters of global demand growth through 2035, driven by industrialization and urbanization in China, India, and Southeast Asia. Europe's rebound is smaller and policy-driven—electrification, heat pumps, and a push to reduce gas exposure—landing on networks that were maintained, not expanded, for three decades.
The Department of Energy's 2012 Large Power Transformers report placed the average age of installed large power transformers—generally 100 MVA and above—at roughly 40 years, with 70 percent already 25 or older. A later Commerce Department Section 232 review put the in-service average near 38 years. Many of those units were designed with a 25- to 40-year technical life and have been run past it. On the distribution side, NREL estimates 60 to 80 million distribution transformers in service, with about 55 percent more than 33 years old and nearing end of life. Wood Mackenzie puts more than half of the distribution fleet—roughly 40 million units—beyond its expected service life, with failure rates expected to climb sharply after 2030. By 2050, NREL estimates that 60 to 80 percent of in-service distribution units will need replacement, and required distribution-transformer capacity could reach 160 to 260 percent above 2021 levels. These aren't interchangeable parts: DOE and industry counts place the number of distribution configurations above 80,000, while large power transformers are closer to one-of-one—about 1.3 units per design, matched to a specific impedance, voltage ratio, and substation footprint.
The industrial base that used to build these machines has been allowed to shrink. In 2019, the Commerce Department's unit count showed 137 large power transformers produced in the United States and 617 imported—an import share of roughly 82 percent. Capacity utilization was only about 40 percent, implying a theoretical domestic ceiling near 340 units a year if every line had been full. By the mid-2020s, domestic production was meeting roughly 20 percent of U.S. large-power-transformer demand and about half of distribution-transformer demand, according to RMI's 2026 supply-chain review. Grain-oriented electrical steel, the core material, has a single domestic producer—Cleveland-Cliffs' Butler Works. Lead times that were four to six weeks for many distribution units, and 12 to 18 months for large ones, are now two to four years. Generator step-up units are quoting around 144 weeks, with some high-capacity orders stretching to five years. Prices have risen roughly 77 percent for power transformers and as much as 80 to 95 percent for distribution units since 2019, with DOE noting cases of four- to ninefold increases on certain transformers.
The problem isn't abstract. The 2013 Metcalf attack in California disabled 17 transformers with rifle fire; restoration took 27 days with the rest of the system still energized. The August 2026 New York Times account of that attack notes the new arithmetic: average waits for a large power transformer have moved from under a year to about 128 weeks, with some quotes at five years. A multi-substation event now competes for the same factory slots as data centers and end-of-life replacements. Units of that size move by Schnabel rail car, and North America has only a handful. Moore County, North Carolina, lost power to about 45,000 people for nearly a week in December 2022 after a substation attack—that was a local outage with spares and crews available. A regional storm that takes out multiple 230 kV and 500 kV banks, or a physical campaign against several at once, runs into the customization problem: sharing programs exist but cover only a fraction of designs. Cyber risk sits beside the hardware risk. Volt Typhoon, the Chinese state-linked campaign disclosed in 2023, pre-positioned on U.S. electric, water, port, and telecom networks. In July 2026, suspected Iranian-linked actors hit drinking- and wastewater systems in at least a dozen states, including more than 30 systems in Minnesota and a brief operational disruption in Clayton County, Georgia.
DOE's Office of Electricity is the lead civilian office on component supply, with a current program funded at up to $375 million targeting distribution and power transformers, materials, and other grid components. The stated levers are refurbishment and reuse, U.S.-sourced substitute materials, standardization around fewer configurations, and next-generation gear such as solid-state transformers. The published outcome targets are modest: cut imports and lead times by up to 10 percent, and utility spending on essential equipment by as much as 25 percent. In April 2026, the White House issued a Presidential Determination under Section 303 of the Defense Production Act finding that grid infrastructure—transformers, conductors, substations, breakers, power electronics, core steel, and the tools to make them—is essential to national defense, and that domestic capacity is "dangerously limited." Manufacturers have announced on the order of $1.8 to $2 billion in North American expansions since 2023, including Hitachi Energy's billion-dollar program and Siemens Energy capacity targeted for 2027. Burns & McDonnell's assessment is that those additions are unlikely to close the U.S. gap inside five years, with workforce the slower constraint: design and build skills for large units are scarce, and training them takes longer than pouring a foundation.
The report lays out five simultaneous requirements for a credible rebuild: domestic capacity for large power transformers and grain-oriented electrical steel has to be sized to replacement plus growth, not to a 10 percent trim in imports; specifications have to converge—80,000 distribution configurations and near-bespoke large units are "a factory killer"; the spare fleet has to match the designs actually in the ground; physical and cyber baselines have to cover the substations and the water and port systems that fail with them; and allied capacity in Mexico, Europe, Japan, and South Korea remains necessary while U.S. lines ramp. The IEA's arithmetic is blunt: grid spending has to rise by half before 2030, or the connection queues—and the outage risk on an already old fleet—keep growing. The United States can still build these machines. It's not building enough today to replace what's wearing out and serve what's being added.