The most consequential industrial chokepoint in the North American grid is not a chip foundry or a battery gigafactory. It is a steel mill in Butler, Pennsylvania, that makes the silicon-iron sheet that goes inside the core of every power and distribution transformer on the continent. There is exactly one domestic producer of grain-oriented electrical steel (GOES) — Cleveland-Cliffs’ Butler Works — and its output, together with imports from a small set of mills in Japan, Korea, Germany, China and India, sets the manufacturing pace for every transformer the grid needs to keep growing.
The U.S. Department of Energy made this concrete in its July 2024 Large Power Transformer Resilience Report: GOES is roughly 25% of large-transformer production cost, current large-power-transformer (LPT) lead times have stretched to 80 to 210 weeks (about 1.5 to 4 years), and the majority of GOES used in U.S. LPT manufacture is not produced domestically. The companion NIAC report of June 2024 treated the transformer shortage as a national-security concern.
Below is how that supply chain actually works — and what the policy response is changing, and on what timeline.
What GOES is, and why nothing replaces it cleanly
Grain-oriented electrical steel is a silicon-iron alloy, typically around 3% silicon, that is cold-rolled and decarburised through a sequence of annealing steps that align its crystal grains in the rolling direction. That alignment gives GOES two properties that the core of an efficient transformer cannot practically do without: very low hysteresis loss in the rolling direction, and high magnetic permeability under AC excitation.
The alternative most often discussed is amorphous metal ribbon. It carries lower no-load loss at low excitation and is a real option for distribution transformers, especially in countries that have mandated it. But amorphous ribbon is more fragile, has a lower saturation flux density (about 1.56 T versus 2.03 T for GOES), and is itself produced by a handful of mills worldwide. The substitution is not free, and at large power-transformer scale GOES still dominates.
This is why the DOE’s final 2024 distribution-transformer efficiency rule explicitly allowed up to 75% of distribution-transformer cores to use GOES, walking back an earlier proposed rule that would have forced a rapid shift to amorphous metal. Industry comments — including from Cleveland-Cliffs and representatives of the Butler Works workforce — argued that a forced shift would strand the only domestic GOES producer and worsen the very supply problem the rule was meant to ease.
The producer map
The geographic concentration is the heart of the issue. The world’s GOES output sits at fewer than ten primary producers, and the share landing in North America is smaller still.
| Producer (parent) | Country | Notes |
|---|---|---|
| Cleveland-Cliffs (Butler Works, PA) | United States | Sole U.S. producer of GOES; only domestic high-perm grade (TRAN-COR®). |
| Nippon Steel | Japan | Long-established producer; large export volumes. |
| JFE Steel | Japan | Major exporter to Asia and North America. |
| POSCO | South Korea | Large integrated producer. |
| thyssenkrupp Electrical Steel | Germany | European supply leader. |
| Baowu / multiple Chinese producers | China | Largest national output, primarily for domestic use. |
| AK Bilstein / Tata Steel (limited) | India / EU | Smaller-volume regional supply. |
(Compiled from public corporate filings and the DOE’s 2024 LPT Resilience Report; specific mill mix shifts over time.)
The headline that the United States has one domestic producer is load-bearing on the rest of the supply story. Every Section 232 trade action, every IRA credit, every utility resilience programme has to reckon with that fact before it touches anything else.
How GOES sets the lead time
Inside a power transformer, the cost stack looks roughly like this:
| Line item | Approximate share of large power transformer cost |
|---|---|
| Grain-oriented electrical steel (core) | ~25% |
| Copper or aluminium windings | ~20% |
| Insulation system (oil, paper, board) | ~10% |
| Tank, structural steel, fittings | ~10% |
| Bushings, tap changer, accessories | ~10% |
| Labour, testing, transport | ~15% |
| Engineering, overhead, margin | ~10% |
(Cost shares per the DOE’s July 2024 LPT Resilience Report and the 2020 U.S. Department of Commerce industry survey it cites; individual unit mix varies.)
Because GOES is the largest single bill-of-materials line, supply discontinuities in GOES propagate through the entire production schedule with no easy buffer. Several effects then compound. Mills allocate GOES to their largest and most strategic customers first. Transformer manufacturers begin locking in steel forward on contract, which is rational at the firm level but tightens the spot market. Spec-changing buyers (a utility that has to add on-load tap changers or change a winding configuration) lose their slot and re-enter the queue at the back. The result is the 80-to-210-week range the DOE now treats as the operative figure for large transformers.
This is also why distribution transformers — historically a few months — have stretched to a year or more for many lines. They draw from the same steel mill, and they are now competing for it with utility-scale orders.
The policy response and its timeline
Three coordinated threads of policy are in motion.
The Section 45X Advanced Manufacturing Production Credit, established by the Inflation Reduction Act, pays a credit per kilogram of domestically produced electrical steel and per VA of domestically produced transformer. It is the direct industrial-policy incentive to expand both ends of the supply chain simultaneously. Bills such as the CIRCUIT Act have proposed extending the credit’s scope and rate further to specifically target transformer manufacturing, though as of mid-2026 these remain proposals rather than enacted law.
The DOE’s final efficiency rule for distribution transformers (2024) settled on the 75% GOES allowance described above, reducing the regulatory pressure that would otherwise have forced rapid substrate substitution.
Capacity is being added on the supplier side. Cleveland-Cliffs announced a $150 million transformer-production plant in Weirton, West Virginia — co-invested with the West Virginia Economic Development Authority and slated to start production in early 2026 — explicitly to consume more of the GOES it already makes at Butler Works. The company has indicated plans to increase domestic GOES tonnage by roughly 30–40%. New transformer capacity has been announced by several OEMs in Mexico, the U.S. Southeast, Tennessee and Quebec.
None of this changes 2026 procurement. New mill capacity routinely takes three to five years from commitment to qualified, certified output. New transformer lines need engineering, hiring and the same supply chain everyone else is using. The realistic expectation is that conditions begin to ease in the late 2020s; the merchant queue does not.
What this means for the next 24–36 months
There is a procurement implication that surfaces in nearly every project review right now. The merchant market is not a free market in any meaningful sense of the term; it is a queue. A position in that queue is set by when the order was placed and how strategic the buyer is. Spec changes, late additions and specification ambiguities are all expensive in calendar time.
Two practical responses tend to work.
The first is to freeze the spec early and order against it, accepting that a small amount of engineering pessimism is cheaper than re-entering the queue. The second is to source outside the merchant queue altogether — from a manufacturer that runs its own source factory and a vertically integrated production line. The relevant queue then is that manufacturer’s own production schedule, not the broader market.
Where Entogo fits
Entogo manufactures transformers, prefabricated substations and switchgear in its own source factory with a vertically integrated production and supply chain. Its standard European-standard (IEC/CE) catalogue transformers ship in an average of 12 weeks, and within 36 weeks even when a product requires new UL or other North-American certification — set against the DOE’s 80-to-210-week figure for the merchant LPT market. The companion piece on transformer lead times in North America treats the procurement decision in more detail.
For projects on the critical path through 2026 and 2027 — data centers, substation rebuilds, renewables interconnections, utility fleet replacement — the GOES chokepoint is the single most useful thing to understand about the North American transformer market. It explains why merchant lead times look the way they do, why policy is moving in the direction it is, and why sourcing strategy is now a project-management variable, not just a procurement preference.
