The largest machines in American life are ones almost nobody has seen. Large power transformers — steel boxes the size of a small house, filled with oil and copper coil — sit behind chain-link fences at the edges of cities, stepping voltage up and down so that electricity generated hundreds of miles away arrives at a socket at exactly the pressure a phone charger expects. There are tens of thousands of them. Most were installed when Lyndon Johnson or Richard Nixon was president. They were built to last forty years. The average unit is now pushing well past it.
This would be a manageable problem if a transformer were a thing you could order. It is not. Each large unit is engineered for its exact position on the grid — its voltage, its load, its physical footprint — and then built largely by hand, by workers winding copper with a patience that resists automation. An order placed today is quoted at three to five years. A decade ago it was under one.
A supply chain with no slack
The arithmetic is uncomfortable. Domestic factories can produce a few hundred large power transformers a year. The grid needs thousands over the next decade merely to replace what is wearing out — before counting data centres, electrified transport, or the re-industrialisation both parties now campaign on. The gap is filled from abroad, mostly from a handful of plants in Asia and Europe whose order books are just as full.
The margin for failure, meanwhile, is thin in ways the public rarely sees. When a big unit fails, utilities borrow spares from one another through informal sharing agreements — a system one engineer described to me as "a neighbourhood tool shed, except the tool weighs four hundred tonnes and there are three of them for the whole neighbourhood."
We are running civilisation on machines nobody remembers how to make quickly.
A federal grid-security adviser
What failure actually looks like
Grid planners are careful with their scenarios, because the honest ones sound like fiction. Lose a transformer to age and the neighbours carry the load while a spare is trucked in. Lose a dozen to a storm, a solar event, or sabotage, and the sharing arrangements exhaust themselves within days. The replacement queue is then not measured in weeks. It is measured in years — years in which some region of the country runs on rationed, rotating power.
The fixes are known and unglamorous: standardise designs so units are interchangeable; stockpile strategic spares the way the nation stockpiles oil; pay to train the winders and engineers a factory takes a decade to develop. Bills to those ends have been drafted, praised, and left to expire in committee more than once. The machines, meanwhile, keep aging at the rate machines do — one day at a time, indifferent to the legislative calendar.
The grid is the rare piece of infrastructure whose success is measured by its invisibility. The people who maintain it would like it to stay boring. Their worry, expressed quietly and consistently, is that the country has mistaken boring for safe.
