Understanding the electric transformer shortage

Start with the puzzle. In 2018 a utility could send a purchase order for ten transformers or two hundred and have them in 12 to 14 weeks, and Mills says nobody worried about it. Today the single-phase pole-top units, the gray cans on utility poles, are normalizing but may still run beyond a year, and pad mounted units, single-phase and three-phase alike, run as long as two years. Every other COVID-era supply chain unkinked. This one did not, which means the COVID explanation was never the right one. Mills draws the distinction sharply: this was not the microchip problem, where carmakers physically could not obtain a necessary input. It was the toilet paper problem, a demand behavior problem. Because deliveries had been dependable for years, utilities did not forecast or pre-order. As lead times crept out, he thinks that kicked out more and more orders, which pushed the queue further out and confirmed the fear that caused it.

Behavior alone does not sustain a four-year shortage, though, and underneath it the demand curve genuinely bent. Mills went back through the history and found transformer volumes tracked the housing market closely, with the last comparable episode in 2006 and 2007 when US housing starts approached two million a year. Then housing crashed, transformer demand crashed with it, and manufacturers either left the business or moved resources elsewhere. His gut estimate is that total US capacity in 2021 was about what it had been in 2007. Meanwhile the demand side acquired drivers that did not exist 15 years earlier: federal infrastructure spending, electrification and EVs, renewables landing on the distribution system, and storm response, with natural disaster severity stepping up from 2017 onward, plus a housing market that was healthy if not at 2006 levels. Four or five variables arrived together against a manufacturing base sized for the previous era.

So why has supply not caught up by now, when everyone agrees the demand is real? Partly because the industry was already flat out. Mills describes plants running 24 hours a day, five days a week at 90 to 92% utilization, so the next incremental unit was the expensive one. Buying the weekend means hiring, and the plants sit in rural communities with little slack labor in a national market with very low unemployment, where staffing a new line means hundreds of people. A wholly new line costs on the order of a hundred million dollars, then waits on equipment lead times, so it is one to two years before production starts and longer before it ramps. And the core of the machine depends on a specialized input: grain-oriented electrical steel, which has exactly one US producer, Cleveland-Cliffs, already at full output. Foreign steel exists and competitors use it, at higher cost because of tariffs whose purpose Mills says he does not know.

Then the part that turns a hard problem into a frozen one. The Department of Energy has proposed a higher efficiency standard for distribution transformers. As initially written, Mills says it would make grain-oriented steel impractical and effectively require amorphous steel, which loses less energy when the transformer is sitting unloaded. Set aside that there is not enough amorphous steel in the world to build every transformer with it: 95 to 98% of American production currently runs on grain-oriented cores, and switching means new equipment, a changed process, a retrained workforce and, he expects, a drop in output before any learning curve arrives. That leaves the expansion decision undecidable. If he commits capital today, he does not know whether to buy grain-oriented or amorphous core equipment. An efficiency rule intended to save energy is, in the near term, a reason not to expand the very supply that is short. Mills expects a compromise rather than the original proposal, notes legislation from Senators Brown and Cruz that would hold current standards for a period, and thinks the question will be settled within one to three months.