This article is for informational purposes only and is not financial advice. TheGatBull may earn a commission from some links at no cost to you — see our disclosure and full disclaimer.
The whole market is buying one AI trade: the chip. But here is the dull, physical question almost nobody asks — what actually carries the electricity into an AI data center? The answer is a giant steel box called a power transformer, and right now America cannot get them fast enough. The companies that can build the biggest ones, quickly, are mostly in Korea. This is not financial advice.
This is Episode 1 of our series on the AI data center power supply chain — a close-up on the link Korea is strongest in: transmission and transformers.
The short answer: AI’s bottleneck is the transformer, not the chip
You can buy all the GPUs you want, but a data center is useless without power — and the power has to be carried and voltage-matched by large transformers. America’s grid is decades old, data-center demand is exploding, and the giant transformers that knit the two together are in chronic short supply. Lead times for large US power transformers have reportedly stretched from about two years to as long as five. The handful of companies that can make ultra-high-voltage units at scale includes Korea’s “Big 3”: Hyosung Heavy Industries (KRX: 298040), HD Hyundai Electric (KRX: 267260), and LS Electric (KRX: 010120). Roughly speaking, they are the transformer business of a GE Vernova or a Hitachi Energy — carved out and sold on its own.
Why now — AI hits a wall at electricity before it hits one at silicon
The hidden twist in the 2026 AI story is simple: the queue isn’t only at the chip fab, it’s on the power grid. Why can’t America just build its own? Ultra-high-voltage transformers are a slow manufacturing business — decades of know-how in electrical steel and winding, plants that take years to stand up. The US emptied out that capacity long ago. Hitachi Energy has pledged over a billion dollars to expand US transformer output and Siemens is adding capacity too, but new lines realistically come online only after 2027. Korea is filling the gap in the meantime, and its order books show it: the Big 3’s combined backlog reportedly pushed past ₩32 trillion (roughly $23 billion) in early 2026, with some orders booked out to 2031.
🎩 Under the Gat — The easiest way Wall Street plays “AI power” is buying one share of GE Vernova for about $1,100. But the company that physically fills the shortage — the plant that actually presses out the 765 kV units America can’t make — sits in Changwon and Ulsan. Transformers sound like a boring smokestack business to outsiders. That smokestack now runs AI.
A second engine: Korea’s “5 Poles, 3 Zones” plan and data-center decentralization
So far this is an export story. But in 2026 a domestic demand engine started turning too — the Lee Jae-myung government’s “5 Poles, 3 Zones” (5극 3특) balanced-growth strategy. The plan carves the country into five mega-regions (the capital area, southeast, Daegu-Gyeongbuk, central, and the Honam region) and three special self-governing provinces (Jeju, Gangwon, Jeolbuk), each with its own growth engine — backed by a second wave of public-institution relocation, a “southern semiconductor belt” (Gwangju, Gumi, Busan), and a sharp funding increase (the balanced-development special account reportedly rising from ₩3.8 trillion to ₩10.6 trillion, with a large share of a ₩150 trillion national growth fund earmarked for these regions).
Layered on top is a push to decentralize data centers. Because data centers have clustered around Seoul and overloaded the capital-region grid, the government now screens new greater-Seoul projects for grid impact (under the distributed-energy law) and dangles incentives — such as steep facility-fee discounts — to steer them toward non-capital regions, closer to where nuclear and other plants actually sit. More than 50 new domestic data centers, totaling over 1,000 MW, are projected by 2026.
Why does this matter for transformers? Putting data centers, relocated agencies and chip clusters in the provinces means wiring fresh power to those regions — transmission lines, transformers, switchgear — and moving electricity from non-capital power plants to load also takes ultra-high-voltage transformers and HVDC links. In other words, the Big 3 get a second demand curve at home, on top of US exports. And like the Value-Up reform, this is a policy the government frames as a survival strategy, so the direction is unlikely to flip with the political winds.
The caveat, fairly stated: regional relocation has been slow in practice, held back by cost and the concentration of talent around Seoul. The policy direction is clear; the pace at which it turns into actual groundbreaking and grid connections may not be. Treat the domestic engine as structural upside, not next quarter’s earnings.
🎩 Under the Gat — If the US is the Big 3’s now-profit, “5 Poles, 3 Zones” is the next one. Foreigners watch only the American orders; on the ground, the government is sketching a plan to re-wire the entire country. The one thing to track: when that policy budget turns into actual transmission and transformer tenders. Watch the order, not the announcement.
Meet the Big 3 — who does what
They look alike but the exposure differs.
- Hyosung Heavy Industries (KRX: 298040) — the ultra-high-voltage king. It reportedly supplies about half of the 765 kV transformers on US transmission grids and has held the No. 1 spot in that segment since the early 2010s. Its order backlog passed ₩15 trillion — a first for a Korean power-equipment maker — and over half of a record first-quarter order haul came from the US. The purest ultra-high-voltage bet.
- HD Hyundai Electric (KRX: 267260) — the US-capacity story. Backlog around ₩11.6 trillion (up roughly 17% from year-end), with a new $200 million Alabama plant set to lift ultra-high-voltage capacity by about 50%. Capacity additions equal forward revenue visibility — and it’s the stock that ran hardest this year (see the risk section).
- LS Electric (KRX: 010120) — profitability and direct data-center deals. First-quarter operating profit reportedly jumped about 45% year on year, and it has signed deals straight with data-center customers (switchgear with AWS, distribution gear with Bloom Energy), broadening from AC into DC power equipment. The name most directly wired into the data center itself.
The anchor for global readers: GE Vernova
| Korea’s Big 3 (Hyosung · HD Hyundai Electric) | GE Vernova (US anchor) | |
|---|---|---|
| Business | Ultra-high-voltage transformers & distribution “hardware” | Gas turbines, grid, services — a full power line-up |
| AI exposure | Direct leverage to the transformer shortage | Generation (gas) plus grid |
| Backlog | ~₩32T combined, some booked to 2031 | ~$163B (equipment + services) |
| Valuation | Cyclical capital goods on a growth multiple, after a big run | Trading near 52-week highs — a premium |
| Decisive difference | Leveraged to the shortage — bigger upside and downside | Revenue diversification cushions a single-theme shock |
Not a perfect parallel — GE Vernova is diversified across gas, grid and services, so its cycle swings less; the Korean trio is concentrated in transformer hardware, so it swings harder when the shortage eases. For a direct transformer comparison, Japan’s Hitachi Energy and Germany’s Siemens Energy are closer peers.
Both sides, fairly — bulls vs. skeptics
The bull case. The AI build-out and the aging-grid replacement cycle are both structural and arrived at once. Transformers can’t be scaled overnight, so the shortage runs long. Korea holds the capacity America vacated, and multi-year backlogs give unusual earnings visibility — now turning into margin, as the strong first-quarter profits showed.
The skeptic case. A good business and a good stock are different things. This is cyclical capital equipment, and today’s fat margins carry a “scarcity premium” that normalizes when the shortage eases — which is exactly what Hitachi, Siemens and GE are spending billions to do. Valuations have already raced ahead: HD Hyundai Electric rose well over 200% this year, pushing past some brokers’ targets, and tariffs cut both ways.
The local current: retail’s “national power stock” vs. foreigners taking profit
The most Korean tension shows up in the order flow. The donghak gaemi (동학개미), Korea’s army of retail investors, rediscovered the boring transformer as the new “national stock” of the AI era — much of HD Hyundai Electric’s huge rally was retail-driven. On the other side, foreign institutions have been booking profits into the spikes. The same stock is a story of “we’re still early” (retail) versus “the valuation is stretched” (institutions).
🎩 Under the Gat — The shortage is real and so is Korea’s edge — when America waits five years, Korea presses the units out, ppalli-ppalli. But separate two things. First, how long the backlog converts into margin, not just headlines. Second, are you buying the business or the chart? Stepping in after a 200%+ run is a completely different game from stepping in before it.
The bottom line: AI’s hard limit isn’t only silicon — it’s the unglamorous steel that moves electricity, and Korea makes a striking share of America’s biggest units. Now a domestic balance-growth push is quietly adding a second demand curve at home. The opportunity is real; just remember it’s cyclical hardware wearing a growth valuation, so when and at what price matter as much as which name. A view, not advice.
— Mr. Gat 🐂
This is not financial advice. Past performance does not guarantee future results, and all figures should be checked against primary sources as of your trade date.
Frequently Asked Questions
Why transformers, and not power generation like nuclear or gas?
Generation makes the electricity; transformers move it to the data center and step the voltage up or down. America’s most immediate bottleneck is grid equipment, not generation, and large power transformers have the longest lead times — reportedly two to five years. That is why this link matters first. This is not financial advice.
Which of the Big 3 is the purest ultra-high-voltage transformer play?
Hyosung Heavy Industries is the name most cited for US 765 kV exposure; HD Hyundai Electric is the capacity-expansion story (its new Alabama plant); and LS Electric leads on profitability and direct data-center deals. They are the same theme with different risk profiles. Not a recommendation to buy any specific stock.
Is there a domestic Korean tailwind, or is this only a US export story?
Both. The Lee Jae-myung government’s “5 Poles, 3 Zones” balanced-growth strategy and a push to move data centers out of the congested greater-Seoul grid create fresh demand for regional transmission lines and transformers. But regional relocation has been slow, so treat it as structural upside that materializes gradually rather than an immediate earnings driver.
Why buy Korea instead of a US transformer maker?
There is essentially no large US-listed pure-play transformer maker. Grid exposure runs through GE Vernova and distribution through Eaton, while the ultra-high-voltage transformers themselves are supplied mainly by Korea, Japan and Europe.
What is the biggest risk?
This is fundamentally cyclical capital equipment wearing a growth-stock valuation. Key risks are shortage normalization as rivals add capacity, an already-large share-price run-up, US tariffs, and order volatility. Always verify figures against primary sources as of your trade date.
This article is for informational purposes only and is not financial advice. TheGatBull may earn a commission from some links at no cost to you — see our disclosure and full disclaimer.