📌 Market analysis for informational purposes · This article is an analysis based on publicly available data and industry reports. It does not recommend buying or selling any particular security, and investment decisions are your own responsibility.
Energy × AI · Strategic Brief

The 5-Year AI Power Supply War
Generation·Transmission + US·Korea Beneficiaries

2026~2030: in just five years, the power demand of new AI data centers must be filled. Which generation sources can actually be switched on, and where is the real bottleneck in the transmission grid? A ranking of the top 5 energy sources, an infrastructure checklist, and a tiering of US and Korean listed beneficiaries.

📅 2026.05.13
Power · Grid · Equipment
⏱️ About 18 min
🌐 🇺🇸 USA · 🇰🇷 Korea
⭐ Executive Summary · Bottom Line First

One-Line Conclusion · Top 5 Energy Sources · Top 10 Beneficiaries

🎯 Conclusion: The generation sources that can realistically be switched on within five years are essentially just three: gas + restarting existing nuclear + solar·ESS. SMRs and new large-scale nuclear are a post-2030 story. The real near-term bottleneck is transformers, the transmission grid, and permitting, and there Korea's three power-equipment makers have already locked in five years of order backlog.

⚡ Ranking of Energy Sources by 5-Year Near-Term Feasibility

Natural Gas (LNG) Generation Available immediately · Absolute mainstay for 5 years
CCGT (combined-cycle) and aeroderivative gas turbines are the only large-scale baseload that can be switched on within 2~4 years. Behind-the-meter gas microgrids are already operating and breaking ground at the 1 GW scale at Stargate and Meta. Gas fills 60~70% of the 5-year gap. New 2025~2026 trend: with the Big 3 gas-turbine slots sold out, marine four-stroke medium-speed engines (Wärtsilä·HD Hyundai Heavy Industries HiMSEN·Hanwha Engine) are entering en masse as data center generators — see Part 03b for details.
Restarting & Uprating Existing Nuclear 24/7 + round-the-clock clean
New nuclear is impossible within five years. Instead, restarting shuttered reactors (TMI/Crane → Microsoft, Palisades, etc.) and direct-sale PPAs from operating reactors take effect immediately. Vistra-Meta 2,600 MW and Talen-AWS 960 MW are the flagships. Small volume, but the strongest margins and pricing power.
Solar + Large-Scale ESS Lowest LCOE · intermittency
PV+BESS is the cheapest generation on an LCOE basis (firm LCOE of solar+storage at $50~80/MWh per IRENA·NREL). GW-scale construction is possible within 12~18 months. But 24-hour full supply requires 4~8 hours of ESS + gas backup. A supplementary baseload to gas.
Enhanced Geothermal (EGS) 24/7 clean · small scale
Fervo Energy is supplying Google with 115 MW in Nevada (online 2026), and Ormat signed a 150 MW PPA with Google. EGS has reached TRL 8 and entered commercialization. 1~2 GW scale by 2030 — small, but with a 24/7 clean premium that gives it a PPA price advantage.
SMR (Small Modular Reactor) From the tail end of the 5 years
NuScale's 77 MWe NRC standard design approval (May 2025), TerraPower's Natrium groundbreaking in Wyoming (2024), and PPAs underway with X-energy-Amazon-Dominion and others. Realistically, first commercial operation is around 2030~2032. Barely registers within the 5-year window, but it is the seed for the post-2030 surge.

💎 Top 10 Beneficiaries (by strength of 5-year visibility)

* Composite of 5-year revenue/order visibility + barriers to entry + pricing power. Analysis opinion, not a recommendation. 30 stocks are tiered in the body.

01
GE Vernova NYSE: GEV
Global #1 in gas turbines. Q1 2026 backlog of 100 GW (up from 80 GW at end-2025, +20 GW in a single quarter). The CEO has publicly stated that 2030 slots will be sold out by the end of 2026. Q1 data center electrification orders alone hit $2.4B — more than all of 2025.
02
Eaton NYSE: ETN
Switchgear·UPS·transformers + a new DC 800V architecture. Its data center backlog is at the level of "about 11 years' worth of what was built in 2025." A $1.2B capacity expansion is underway. Power-equipment revenue per AI rack is 3~5x that of a standard DC.
03
Constellation Energy NASDAQ: CEG
#1 nuclear fleet operator in the US. Restarting TMI (Crane Clean Energy Center) under a 20-year Microsoft PPA — a $1B federal loan from the Trump administration is confirmed, targeting startup in 2027~2028. 24/7 clean baseload = the strongest pricing power.
04
Quanta Services NYSE: PWR
North America's largest transmission & distribution EPC. A record Q1 2026 backlog of $48.5B. The company that moves transformers and plants transmission towers — as more power plants come online, its workload automatically grows. Revenue visibility of 4~5 years.
05
HD Hyundai Electric KRX: 267260
Ultra-high-voltage transformers bound for the US. Q1 2026 operating profit of KRW 258.3 billion (a quarterly record). Backlog of KRW 10~11 trillion — five years of work secured. An unusual profit environment in which US tariffs are borne by the purchaser.
06
Hyosung Heavy Industries KRX: 298040
Transformers·circuit breakers. Q1 2026 new orders of KRW 4.17 trillion (a quarterly record), backlog of KRW 11.9 trillion. Double momentum from AI data centers + replacement of the aging US grid.
07
Vistra NYSE: VST
Top 2 US IPP. A triple portfolio of gas + nuclear + ESS. A 2,600+ MW nuclear PPA with Meta. As a company that also runs gas, it is the IPP generating cash most in real time within the 5-year window.
08
LS ELECTRIC KRX: 010120
Transformers·distribution·automation. Q1 2026 revenue of KRW 1.38 trillion and operating profit of KRW 126.6 billion (both quarterly records). Backlog of KRW 5 trillion. Of the three Korean firms, the most total-solution play, reaching all the way to automation and DCIM.
09
Doosan Enerbility KRX: 034020
SMRs (cooperation with NuScale·X-energy) + a new 380 MW gas turbine. After winning its first two gas turbine orders from a US Big Tech firm in H1 2026, it added five more within two months. Gas in the near term, SMRs in the long term — double momentum.
10
Talen Energy NASDAQ: TLN
Owns the Susquehanna nuclear plant. A 960 MW data center campus PPA with AWS. The single stock where the "direct nuclear sale model" shows up most cleanly.

How Much Demand Must Be Filled Within Five Years?

Let's start with the near-term numbers. The IEA (International Energy Agency) projects that global data center electricity consumption will double from 485 TWh in 2025 to roughly 950 TWh in 2030. Goldman Sachs forecasts +165% growth in 2030 versus 2023, and in some scenarios as much as +220%. Under any scenario, data center power alone will need an additional +400 TWh+ within five years.

2025 Global DC Power
485 TWh
Baseline
2030 Global DC Power (IEA)
~950 TWh
▲ +96%
Increment (5 years)
+465 TWh
≒ 80% of Korea's total electricity consumption (about 580 TWh)
GE Vernova Q1 2026 Backlog
100 GW
▲ +20 GW per quarter
US Transformer Lead Time
128~144 weeks
▲ About 3~4 years
Korea DC Applied Capacity
906→7,343 MW
▲ +8x over 4 years
Why "five years" is decisive: AI model training and inference traffic is surging in 2025~2027, yet new large-scale nuclear plants, SMRs, and new transmission lines cannot be switched on within five years. Therefore, during these five years, whoever owns generation and transmission assets that are already running or can be switched on within 2~4 years holds the pricing power — which translates directly into share price.

📈 Near-Term Trend: What Happened Over the Past Few Months

One thing this trend tells us: Big Tech no longer waits for "the public grid to take care of it." They are doing one of three things: buying power plants directly, building them on their own sites, or locking in 20-year direct-sale PPAs. And all three paths mean someone has to machine the gas turbines, wind the transformers, and plant the transmission towers.

Why Only These Three (Gas·Existing Nuclear·Solar+ESS) Within Five Years

When evaluating generation sources within the 5-year window, what matters more than LCOE (levelized cost) is "can GW-scale new capacity from this source actually be switched on starting from this point?" The comparison below uses the criterion of "if construction begins in January 2026, when does it come online?"

⏱️ Lead Time from Groundbreaking to Commercial Operation by Source

SourceMinimum Lead TimeRealistic Lead TimePossible Within 5 Years?Notes
Modular gas (aeroderivative)12~18 months18~30 monthsAdopted by Stargate·Meta. But turbine slots near sell-out by 2030
Large CCGT (combined-cycle)36~48 months48~60 months✓ (barely)Groundbreak 2026 → online 2030
Solar + ESS12~24 months18~30 monthsThe fastest GW-scale option once transformers and interconnection are unblocked
Restarting existing nuclear24~48 months36~60 monthsTMI/Crane 2027~2028, Palisades, etc.
Nuclear uprate (MUR/EPU)24~36 months24~48 months+5~10% to existing units. Nuclear fuel·turbine replacement
Onshore wind24~36 months36~60 monthsPermitting·transmission interconnection are the real variables
Enhanced geothermal (EGS)24~36 months30~48 monthsCapped at 1~2 GW by 2030. Site-dependent
Offshore wind48~72 months72~96 monthsBarely registers within 5 years
SMR60~84 months72~120 months✗ (tail)First startup possible 2030~2032. The very tail of the 5-year window
New large-scale nuclear96~144 months120~180 monthsA next-cycle (post-2032) story

* Lead times are estimates based on EIA·DOE·IEA·industry averages. Subject to ±30% variation depending on site·permitting·transmission interconnection conditions.

🏁 In the End, 5-Year New Supply Is These Four Combinations

Key insight: 5-year available generation = three pillars, "gas + existing nuclear + solar·ESS." All three must pass through the same bottleneck of transformers·transmission·switchgear + EPC labor. In other words, whether you bet on a generation source or on transformers, the destination is the same. That is why Korea's three power-equipment makers, Eaton, and Quanta are all posting record quarterly results at the same time.

① Gas Generation — The Absolute Mainstay Filling 60~70% of the 5-Year Gap

Why is gas #1? The answer is simple: there is no way to switch on 24-hour, GW-scale capacity within five years with any other source. In an era when Big Tech is building single campuses of 1.2 GW (Stargate Abilene) or 1 GW (Meta El Paso), "drawing power from the surrounding grid" no longer works.

📍 By Form: Which Gas Generation Actually Gets Deployed

🏭 The Infrastructure Required (to switch on one gas plant)

📊 Gas Turbine Big 3 + One Korean Firm: Backlogs

GE Vernova
100 GW · 2030 slots near sell-out
Siemens Energy
Full through 2030, expanding capacity to 70~80 units/yr
Mitsubishi(MHI)
Competing for M501JAC data center slots
Doosan(KR)
First 7 units ordered by US Big Tech (2026)
What Doosan Enerbility signifies: with the global gas-turbine Big 3 unable to free up slots through 2030, demand for a 4th supplier has opened up seriously for the first time. Doosan sold its first 380 MW H-class unit to US Big Tech and won five more within two months — the first case of a long-blocked Korean power-generation EPC riding directly on the US data center cycle.

Why Marine Engines Went to Data Centers — A New 2025~2026 Category

"If there are no gas turbines, use ship engines." This is a proposition that has been validated in earnest in the US data center market since the second half of 2025. Four-stroke medium-speed gas engines (5~20 MW reciprocating gas engines typically used as marine propulsion) have begun entering en masse as generators for Big Tech's new campuses. What differs from before is that they go in not as mere backup but as primary generation or behind-the-meter baseload.

🧭 Why This Trend Emerged Now — 6 Reasons

  1. Gas turbine Big 3 sold out: new H/J-class lead times from GE Vernova·Siemens Energy·Mitsubishi are 5~7 years. Order in 2026 → online 2031~2033. AI campuses need power in 2026~2028, but gas turbines can't meet that.
  2. Modular·fast startup: four-stroke gas engines are small at 5~20 MW each, with grid synchronization in about 30 seconds. A 100~1,000 MW campus is stacked with 5~50 units in parallel. Robust against AI workload load swings.
  3. Partial-load efficiency: gas turbine efficiency drops sharply below 75% load. Reciprocating engines stay relatively flat across the full 25~100% range. Favorable for workloads with erratic loads, like AI inference.
  4. Low water use: large CCGTs need large volumes of cooling water (a constraint on data center siting). Gas engines use radiator cooling — a good fit for water-scarce regions like Texas and Arizona.
  5. Natural gas + future fuel options: the Wärtsilä 34SG/50SG support not only natural gas but also biogas, synthetic methane, and hydrogen blending. A hedge against both future LNG inflation and decarbonization pressure.
  6. Capacity synergy with the shipbuilding cycle: global four-stroke medium-speed engine manufacturing is effectively concentrated in Wärtsilä·MAN ES and Korea's three firms (HD Hyundai Heavy Industries·Hanwha Engine·STX Engine). When shipbuilding orders slow, this capacity can be redirected to power generation. For Korean makers, it is an opportunity to take on US data center work without building new capacity.

📦 Verified US Data Center Order Cases

Wärtsilä — 1.6 GW+ in US data center orders (2025.7~2026.4)
Listed: HEL: WRT1V
Key orders during the 8 months in which Wärtsilä entered the US data center market in earnest:
  • 2025.7 — Announces its first engine order for a US data center.
  • 2025.11507 MW: 27 units of the Wärtsilä 50SG. Natural-gas operation, convertible to sustainable fuel in the future. US East Coast data center.
  • 2026.1 — Wins an additional major US power plant project (directly attributable to data center demand).
  • 2026.4.16412 MW: 40 units of the Wärtsilä 34SG, for a hyperscale data center campus in Ohio. The data center debut of the 34SG model. 5.6~9.8 MW per unit. Supports natural gas + biogas, synthetic methane, and hydrogen blending.
  • 2026.4.23790 MW: 42 units of the Wärtsilä 50SG, in Texas (emerging as a next-generation Data Center Alley). 49.9% simple-cycle efficiency, 30-second grid synchronization.
Combined US data center orders of 1.6 GW+. Delivery proceeds in stages from late 2026 to 2027.
HD Hyundai Heavy Industries — First US Entry, 684 MW (2026.4.22)
KRX: 329180
A contract with US energy infrastructure developer Aperion Energy Group (AEG) to supply data center power generation equipment based on HiMSEN 20 MW-class power-generation engines. Total capacity of 684 MW, contract size of KRW 627.1 billion (about $423.7M). The largest power-generation engine contract HD Hyundai Heavy Industries has ever signed, and the first case of a Korean power-generation engine entering the US data center market. The company is in further negotiations — per Ajupress reporting, it cited a cumulative US data center potential order scale of $4.6B.

Competitive edge: HD Hyundai Heavy Industries is the only Korean company to own medium-speed engine IP in-house (able to produce in-house with no license fees) → price competitiveness + immediate use of capacity. The stock rose +45% over the two weeks after the announcement, with market cap surpassing KRW 70 trillion.
Hanwha Engine — Restarts Four-Stroke Medium-Speed Engine Line After 10 Years
KRX: 082740
Hanwha Engine (formerly HSD Engine) is resuming production of four-stroke medium-speed engines after roughly 10 years, having halted it in 2016 due to worsening profitability, and is investing toward operation of production facilities in 2027. The group synergy structure is key — a vertical group structure is taking shape in which Hanwha Energy (a BTM generation operator) signs long-term PPAs with US hyperscalers, and Hanwha Engine procures the medium-speed engines for the BTM plants. Korea Investment & Securities projects Hanwha Engine's revenue will challenge KRW 3 trillion in 2027, with operating profit going from KRW 248.0 billion in 2026 → KRW 454.0 billion in 2027 → KRW 498.0 billion in 2028. Entering as a 2nd mover.
ProEnergy — Aircraft Engine Retrofit (US)
Private
Strictly speaking this is an aeroderivative variant rather than a marine engine, but it belongs to the same "power-generation retrofit of existing industrial engines" category. It converts GE CF6-80C2 engines from Boeing 767 aircraft into generators, launching the trailer-mounted PE6000 module (48 MW). Cumulative delivery of 1 GW+ to two data center customers. With fast deployment, mobility, and use of the secondhand aircraft engine market, its lead time is far shorter than that of a new gas turbine.

🇰🇷 Comparison of Korea's Three Four-Stroke Medium-Speed Engine Makers

CompanyTickerFeaturesUS DC Orders5-Year Visibility
HD Hyundai Heavy IndustriesKRX 329180HiMSEN 20 MW-class, the only Korean firm to own medium-speed engine IP in-house684 MW (2026.4) + further talksStrongest
Hanwha EngineKRX 082740Formerly HSD Engine. Four-stroke medium-speed engine line restarting in 2027Hanwha Energy BTM vertical structure taking shapeStrong from 2027~
STX EngineKRX listedMedium-speed engines + diversification into military·power generationRelated expectations, specific order timing undisclosedAnticipated
Key point: the landing of marine engines in data centers is not a one-off but a permanent category forcibly created by the sell-out of the gas-turbine Big 3 slots. Wärtsilä won 1.6 GW in the US alone in just 8 months, and Korea's three firms have entered in earnest too. This is a change that requires redrawing the gas generation market share in Part 03 above — roughly 10~20% of new gas generation GW in the 5-year window is likely to be taken by this four-stroke engine category. Betting solely on the gas-turbine Big 3 is no longer full exposure to gas.

② Restarting & Uprating Existing Nuclear — The Most Expensive but Cleanest GW

"Let's build new nuclear" is not a 5-year solution. But nuclear that is already built is a 5-year solution. The US has 94 operating units, several closed units that can be restarted, and room for capacity uprates (MUR/EPU) on existing units. As Big Tech signs PPAs directly and underwrites the generators' loss risk, units once shut down for "insufficient profitability" are coming back to life.

📌 The 3 Nuclear Cards Possible Within Five Years

🏗️ Infrastructure Side (what restarting/uprating nuclear requires)

Limits and risks: the US pool of closed-reactor restart candidates is actually fewer than 10 units. Even if all are switched on, it's 5~10 GW — a single-digit % of the +400 TWh demand. It's a card that is "small in volume but with the strongest margins and pricing power." However, a single political risk (NRC·anti-nuclear sentiment) or a single safety incident would set the entire momentum back.

③ Solar + ESS — The Fastest and Cheapest, but Not 24-Hour

The solar + battery (BESS) combination is in fact the cheapest on an LCOE basis. Per IRENA·NREL data, in some US regions firm LCOE (including storage and transmission costs) has come down to the $50~80/MWh level. And the plant itself can be built at GW scale within 12~18 months. The problem is just one thing — when the sun isn't out, it doesn't generate.

🔋 The BESS Cost Curve Made This Possible

📦 How It's Deployed (at actual campuses)

🏭 Core Infrastructure Required

What the 2026.03 LG Energy Solution–Tesla $4.3B contract means: since the US IRA took effect, domestic US production of BESS cells has become a major obligation. With tariffs imposed on Chinese cells, Korea's LGES Michigan line has effectively become a Tesla-dedicated line. LG Energy Solution is the cleanest KR beneficiary of the 2026~2030 surge in US data center ESS.

④ Enhanced Geothermal (EGS) · ⑤ SMR — Small but Decisive Options

④ Enhanced Geothermal (EGS) — A New Category Created by Fervo·Ormat·Eavor

"Geothermal" used to be possible only in regions with natural heat sources in the crust (Iceland·New Zealand). Enhanced geothermal (EGS) borrows shale gas drilling technology to create artificial fractures at depths of 3~4 km and circulate water, extracting geothermal heat anywhere. Fervo Energy commercialized this first, reaching TRL 8 (ready for commercial adoption).

EGS, being 24/7 clean baseload, is faster than SMRs and more stable than solar·wind. It will reach the 1~2 GW level globally within five years — small, but a decisive card filling the "final 1~2%" of Big Tech's clean-power portfolio.

Directly listed beneficiaries are limited: Fervo is private. Ormat Technologies (NYSE: ORA) is the best US geothermal pick. It owns drilling, heat exchangers, and turbines all in-house. Drilling-service firms like SLB·Halliburton also benefit indirectly as EGS spreads.

⑤ SMR (Small Modular Reactor) — From the Last Year of the 5-Year Window

SMRs are modular reactors in 50~300 MWe units. The concept is "made in a factory, transported by truck, and assembled on site," but in reality, the fastest scenario has first commercial operation in 2030. That is, only in the last year of the 5-year window (2026~2030) might the very first unit barely come online.

Realistic assessment: the first commercial unit comes online in 2030~2032. SMR actually switched on within the 5-year window (2026~2030) is nearly 0 GW, or one or two units at best. SMRs are not "solving the 5-year problem" but "a seed for the post-2030 surge." That said, the companies that planted their flags early collect PPAs, licenses, and component reservations over the five years, so it gets reflected in share prices fairly early.

More Clogged Than Generation Is Transmission·Transformers·EPC

This is the single most important chapter of the article. More clogged than building power plants is the transmission grid·transformers·switchgear that carry that power to the data center. This is not mere rhetoric but a measured fact.

📐 Near-Term Lead Time Matrix (as of 2026.Q2)

Equipment/ProcessLead TimeMajor SuppliersResolution Outlook
Large transformers (standard)128 weeks (about 2.5 years)Hitachi Energy, GE Vernova, Siemens Energy, HD Hyundai Electric, Hyosung Heavy Industries, LS Electric2028+
Generator-side transformers (GSU)144 weeks (about 3 years)Same as above2029+
HVDC converters·subsea cables3~6 yearsHitachi Energy, GE Vernova, Siemens Energy, Prysmian, Nexans, LS Cable & System·LS Marine Solution~2030
Gas turbines (H/J-class)Slots sold out ~2030GE Vernova, Siemens Energy, Mitsubishi, Doosan (new entrant)2030+
Transmission line permitting (US PJM)5~10 yearsLocal·federal FERC·environmental regulationStructural
High-voltage circuit breakers·GIS40~80 weeksHitachi Energy, ABB, Siemens, HD Hyundai ElectricImproving
Switchgear·LV/MV distribution30~60 weeksEaton, Schneider, ABB, LS ElectricImproving
Grain-oriented electrical steel for transformers~2 yearsPOSCO·Nippon Steel·AK Steel (now Cleveland-Cliffs)Expanding

🇰🇷 Why Korea's Three Power-Equipment Makers Are Doing Insanely Well

This isn't simple luck. Two cycles erupted at once: (i) a surge in new generators·transformers for AI data centers, and (ii) the replacement cycle for aging US grid transformers (average transformer lifespan of 30~40 years — the 1980s~1990s installations are now due for replacement). The two overlap, putting global transformer demand at 1.5~2x capacity.

🇺🇸 Core Players on the US Transformer·EPC Side

Why transformers are the real bottleneck: even if you build 100 more power plants, you can't send electricity to data centers without transformers and transmission lines. NREL projects that the 2026 transformer shortage could delay about 25% of global renewable projects. "Who makes the transformers" may be a simpler and stronger bet on 5-year visibility than "which generation source is #1."

🇺🇸 US Listed Beneficiaries — Sorted Into 3 Tiers

Within each tier, stocks are ordered by a composite assessment of 5-year visibility·barriers to entry·pricing power. All stocks are analysis opinions, not buy recommendations.

🏆 Tier 1 — Strongest 5-Year Revenue Visibility·Pricing Power
GE Vernova NYSE: GEV
#1 in gas turbines
Global #1 in gas turbines. Q1 2026 backlog of 100 GW (80 GW at end-2025). Data center electrification orders alone at Q1 $2.4B (more than all of 2025). A full lineup spanning wind·HVDC·transformers. Exposed to both generation and transmission.
Constellation Energy NASDAQ: CEG
#1 in nuclear
#1 nuclear fleet operator in the US. A 20-year Microsoft PPA + TMI/Crane restart + a confirmed $1B Trump federal loan. Startup 2027~2028. Just about the only company that can supply GW-scale clean new capacity within five years.
Eaton NYSE: ETN
Integrated power equipment
Switchgear·UPS·transformers·a new DC 800V architecture. Its data center backlog is on the scale of "about 11 years' worth of what was built in 2025" (the company's phrasing). $1.2B capacity expansion. Power-equipment revenue per AI rack is 3~5x that of a standard DC. An NVIDIA Reference Design partner.
Quanta Services NYSE: PWR
#1 in T&D EPC
North America's largest T&D EPC. A record Q1 2026 backlog of $48.5B (QoQ +20%+). A downstream player that gets work in proportion to whatever anyone builds in plants·transmission lines — on a stock basis, the most risk-diversified grid bet.
⚡ Tier 2 — Direct Generation·IPP Exposure
Vistra NYSE: VST
Top 2 IPP
A triple portfolio of gas + nuclear + ESS. A 2,600+ MW nuclear PPA with Meta. As a gas IPP, it generates cash most in real time within the 5-year window.
Talen Energy NASDAQ: TLN
Direct nuclear-sale PPA
Owns the Susquehanna nuclear plant. A 960 MW PPA for an AWS data center campus. The stock where the "direct nuclear sale model" shows up most cleanly.
NRG Energy NYSE: NRG
Gas IPP expansion
Acquisition of about 13 GW of gas generation assets from LS Power expected to close in 2026. Abundant new data center PPA negotiations in the Texas·East markets.
Dominion Energy NYSE: D
Virginia DC stronghold
The monopoly utility for Northern Virginia (Data Center Alley). 26% of DC power runs on Dominion's grid. With a 4~7 year wait for new connections, its pricing power is strong. Exposed to both transmission and generation.
🛠️ Tier 3 — Infrastructure·Materials·BESS·Geothermal
Vertiv NYSE: VRT
DC cooling·power
Integrated data center cooling·power. With AI chip heat surging, demand for immersion cooling·D2C cooling is exploding. An NVIDIA Reference Design partner.
First Solar NASDAQ: FSLR
US-made PV
#1 in US CdTe thin-film solar. A beneficiary of IRA US-content bonuses·tariffs. The most natural module supplier for Big Tech's solar PPAs.
Fluence Energy NASDAQ: FLNC
BESS integration
#1 in grid-scale BESS system integration. The more solar+ESS PPAs near data centers grow, the more system-integration work it gets. That said, competition with Tesla Megapack/CATL is intense.
Tesla NASDAQ: TSLA
Megapack / Megablock
Megapack 3 (5 MWh modules) → Megablock (a 20 MWh integrated package, launching 2026). $430M supplied to xAI Colossus. A $4.3B LFP cell supply contract with LG Energy Solution. Becoming the standard for "AI DC ESS."
Ormat Technologies NYSE: ORA
Best geothermal pick
Owns drilling·heat exchangers·turbines in-house. Signed a 150 MW PPA with Google. Geothermal is a small market, but being 24/7 clean it commands a PPA price advantage with Big Tech. The cleanest direct listed geothermal exposure.
Cameco NYSE: CCJ
Uranium
#1 uranium miner in Canada. A joint venture with US Westinghouse (2023). The more nuclear restarts·new PPAs grow, the more nuclear-fuel prices are supported. Good 5-year visibility.
MYR Group NASDAQ: MYRG
Mid-cap T&D EPC
Smaller than Quanta but a key player in transmission-line construction·substation EPC. A direct downstream beneficiary of the new PJM·ERCOT cycle.

🇰🇷 Korean Listed Beneficiaries — Sorted Into 3 Tiers

Korean stocks are exposed to two markets at once: (i) US grid transformer·generation exports, and (ii) domestic Korean data center transmission·generation. Both have strong 5-year visibility.

🏆 Tier 1 — Simultaneous US·Korea Beneficiary + Record Quarterly Results
HD Hyundai Electric KRX: 267260
Transformers·generators
A core exporter of ultra-high-voltage transformers to the US. Q1 2026 revenue of KRW 1.04 trillion, operating profit of KRW 258.3 billion (a quarterly record), backlog of KRW 10~11 trillion. An unusual seller's market in which US purchasers even bear the tariffs. Capacity expansion + US line expansion underway.
Hyosung Heavy Industries KRX: 298040
Transformers·circuit breakers·STATCOM
Q1 2026 revenue of KRW 1.36 trillion, operating profit of KRW 152.3 billion (+48.7%), new orders of KRW 4.17 trillion (a quarterly record). Backlog of KRW 11.9 trillion. Diversified across the US grid + exports to Saudi Arabia·Europe.
LS ELECTRIC KRX: 010120
Power·automation·DCIM
Q1 2026 revenue of KRW 1.38 trillion, operating profit of KRW 126.6 billion (both quarterly records, YoY +33%/+45%). Backlog of KRW 5 trillion+. A total solution: transformers·distribution + data center power management (DCIM·UPS·even EV chargers).
Doosan Enerbility KRX: 034020
Gas turbines + SMR
First 380 MW gas turbine order from US Big Tech in 2026 + five more within two months = direct entry into the gas field. A global core supplier of NuScale·X-energy SMR components. Gas in the near term, SMRs in the long term — double momentum at 5 years+ and beyond.
HD Hyundai Heavy Industries KRX: 329180
HiMSEN engine, first US DC entry
Core shipbuilding business + a 2026.4 contract with Aperion of the US to supply 684 MW of HiMSEN power-generation engines (KRW 627.1 billion) — the first landing of a Korean power-generation engine in a US data center. The only Korean firm to own medium-speed engine IP in-house = no license fees. Further negotiations underway. The stock rose +45% over the two weeks after the announcement, with market cap surpassing KRW 70 trillion. A direct beneficiary of the gas-turbine Big 3 slot sell-out.
Hanwha Engine KRX: 082740
Four-stroke medium-speed engine + Hanwha Energy PPA
Formerly HSD Engine. Restarting the four-stroke medium-speed engine line halted in 2016 after 10 years, targeting full operation in 2027. Entry into US hyperscaler PPAs is taking shape via the group's vertical structure of Hanwha Energy (a BTM generation operator) + Hanwha Engine (engine supply). Korea Investment & Securities projects 2027 revenue of KRW 3 trillion and operating profit of KRW 454.0 billion. A 2nd mover.
⚡ Tier 2 — Transmission·Cable·Heavy Electrical
LS Cable & System·LS Marine Solution KRX: 229640 (LSM) etc.
HVDC·subsea cable
One of the global big four in HVDC subsea cables (Prysmian·Nexans·NKT·LS). Large orders underway from the UK's SSEN·US East offshore wind, etc. As AI DCs form power clusters, orders for nearby transmission cables link directly.
Taihan Cable & Solution KRX: 001440
Power cable
#2 in ultra-high-voltage cable. Expanding exports to the US·Middle East. A co-beneficiary of the transformer super-cycle.
Jeryong Electric KRX listed
US distribution transformers
A large share of direct distribution-transformer exports to the US. The most explosive KR volatility stock of the 2024~2025 super-cycle — good 5-year visibility, but heavy volatility·valuation burden.
Sanil Electric KRX listed
Special transformers
Strong in special transformers for data centers·renewables. Expanding US export share after its IPO (2024).
Iljin Electric KRX listed
Transmission lines·wires
Transmission lines·transformers + some power equipment. Exports to the US·Southeast Asia.
🔋 Tier 3 — Batteries·Generation Support·Materials
LG Energy Solution KRX: 373220
BESS LFP cells
A $4.3B LFP cell supply contract with Tesla for Megapack 3 (2026.03). The Michigan line is effectively dedicated to Tesla ESS. US IRA bonuses + Chinese cell tariffs = Korean cells become the de facto default for the US DC ESS market.
Samsung SDI KRX: 006400
ESS·EV
ESS cells·systems. A US Indiana line + European ESS orders. A lower ESS share than LGES, but pricing power recovering.
Hanwha Solutions KRX: 009830
US-made PV
Hanwha Q CELLS' US Georgia plant — a direct beneficiary of IRA bonuses for US-made modules. Korea's largest exposure to module supply for Big Tech solar PPAs.
POSCO Holdings KRX: 005490
Grain-oriented electrical steel
One of the global top 2~3 in grain-oriented electrical steel (GOES) for transformer cores. The deepest downstream material in the transformer super-cycle.
Nuclear maintenance stocks like Woojin Entech·Woojin KRX listed
Nuclear maintenance
Exposure to Korean nuclear maintenance·uprates of aging units + a new nuclear cycle. Bigger potential beneficiaries when SMRs ramp up in earnest. Large differences in business mix·financial health by stock.
Korea's own data center market: Korea's DC power-application capacity goes from 906 MW (2023) → 7,343 MW (2027) = +8x over four years. About 75% of operating DCs and about 68% of new ones are concentrated in the capital region — under the 11th Basic Plan for Electricity Supply and Demand, reinforcement of the capital-region transmission grid·West Coast HVDC, etc., are the core of the 5-year infrastructure work. KEPCO·KEPCO KPS·KEPCO Industrial Development and other KEPCO subsidiaries + LS Cable & System·Taihan Cable & Solution·HD Hyundai Electric all benefit directly.

Five Watch-List Items (Just Two Breaking Would Shake the Scenario)

AI training-efficiency leap
HIGH

If a DeepSeek-style training-efficiency breakthrough happens once more, power consumption per GPU could fall 30~50%. If the demand curve flattens, gas turbine slots could empty before the backlog fills.

Interest rates·cost of capital
MID

If the US 10Y staying at 4.5%+ becomes prolonged, the cost of capital for IPPs·utilities rises. PPA pricing power shifts slightly toward Big Tech. But Big Tech capex is covered by its own cash flow, so a major retreat is unlikely.

Nuclear accident/politics
TAIL

A single major accident in the US or Europe within five years would set back both restart and SMR momentum. A tail risk: low probability but large impact.

Transformer supply normalization
MID

If global transformer capacity normalizes between 2028~2030, the pricing power of the KR three·Hitachi gradually weakens. But it barely loosens within the 5-year window.

Tariff·trade policy
HIGH

The current structure has purchasers bearing US transformer tariffs — if Trump administration policy changes, it directly affects the margins of the Korean three. But as long as US capacity is short, it's only volatility in the near term; the structure holds.

The asymmetry of risk: the probability that all five above break negatively is low, but even one breaking would heavily affect particular stocks. That is why diversifying across 1 generation stock·1 transmission stock·1 BESS stock is safer than concentrating in a single name. Transmission·T&D EPC serves the broadest insurance role.

The 5-Year Asymmetry — "Those Who Make the Electricity, Those Who Carry It"

Betting money on AI data centers themselves is already largely priced in. NVIDIA·Microsoft·Meta·Amazon·Oracle have all set new highs. But the companies that can actually switch on the electricity they need to run are far fewer. Gas turbines are the Big 3, transformers are 5~6 firms globally, nuclear operation is 3~4 firms, and BESS cells are within 5 firms. This narrow group of companies takes 30~40% of Big Tech's $5 trillion in capex over the five years.

To restate the priorities of the 5-year window:

One-line summary: the real 5-year gate of the AI era is neither GPUs nor HBM but transformers and gas turbines. The companies that make those two are already posting record quarterly results, and their slots are selling out through 2030. Make generation a 3-way #1 priority of LNG·existing nuclear·solar+ESS, and transmission·transformers·EPC a separate #1 — diversifying across these two axes is the most robust positioning in the 5-year window.