Can Power Infrastructure Companies Like GEV and ETN Sustain Premium Multiples as AI Capex Normalizes?
The AI infrastructure buildout has minted a new class of market darlings: power and electrical equipment companies. GE Vernova has tripled in a year. Vertiv has more than tripled. Eaton, Amphenol, and Quanta Services have all delivered triple-digit returns. The thesis is straightforward — hyperscalers are spending $200 billion-plus annually on data centers, and every megawatt of compute needs transformers, switchgear, cooling systems, and grid connections. But with AI capex growth showing early signs of moderation and these stocks trading at historically elevated multiples, the question is whether the premium is sustainable or already priced in.
Why This Theme Matters Now
Microsoft, Meta, Google, and Amazon collectively guided to over $200 billion in 2026 capital expenditure, much of it directed at AI data center capacity. But the tone has shifted: Microsoft paused or delayed several data center projects in early 2026, and Meta signaled a more disciplined approach to spending cadence. The power infrastructure supply chain — transformers, switchgear, cooling, and grid interconnection — still faces multi-year backlogs, but investors are starting to ask whether peak growth rates are behind us. The stocks in this space now trade at 25–60x forward earnings, valuations more typical of high-growth software than industrial companies.
The Companies: Can Premium Multiples Hold?
We examined five companies across the AI power infrastructure supply chain — from grid-scale equipment to connectors to construction services — to assess which have the earnings trajectory to justify their valuations and which look most vulnerable to a re-rating.
1. GE Vernova (GEV) — The Grid Bottleneck Play
GE Vernova is the largest pure-play power equipment company globally, spanning gas turbines, grid solutions, wind, and electrification. Its grid segment is the most direct beneficiary of data center interconnection demand.
GEV's revenue reached $38.1 billion TTM, but the real story is margin expansion: EBITDA margins climbed from near-breakeven at the 2024 spin-off to 9.7% as the wind segment losses narrowed and gas/electrification margins expanded. Q4 2025 net income of $3.7 billion included one-time items, but even adjusted, the trajectory is sharply positive. The challenge: at 59x forward P/E, the stock prices in several years of flawless execution.
| Metric | Value |
|---|---|
| Market Cap | $230.0B |
| Revenue (TTM) | $38.1B |
| Revenue Growth | 9% YoY |
| EBITDA Margin | 9.7% |
| P/E (fwd) | 59.4x |
| 1Y Price Return | +184% |
GEV has the strongest structural tailwind — grid infrastructure is a genuine bottleneck — but its valuation leaves almost no room for disappointment.
2. Eaton Corporation (ETN) — The Steady Compounder at a Premium Price
Eaton is a diversified power management company whose Electrical Americas segment has become the go-to supplier for data center power distribution, UPS systems, and switchgear.
Eaton delivered $27.4 billion in TTM revenue with 10% growth and 22.6% EBITDA margins — steady and predictable, which is exactly why the stock commands a premium. Q4 2025 EPS of $2.91 beat expectations. Management has consistently guided to mid-to-high single-digit organic growth, with data center orders providing multi-year visibility. At 26.7x forward P/E, ETN is the most reasonably valued name in this group, though still well above its historical 18–20x range.
| Metric | Value |
|---|---|
| Market Cap | $138.0B |
| Revenue (TTM) | $27.4B |
| Revenue Growth | 10% YoY |
| EBITDA Margin | 22.6% |
| P/E (fwd) | 26.7x |
| 1Y Price Return | +22% |
ETN offers the best risk-adjusted exposure: diversified enough to absorb a data center slowdown, with margins and execution that justify a modest premium.
3. Amphenol Corporation (APH) — The AI Connector King
Amphenol is the world's second-largest connector maker, with IT datacom now representing 36% of sales. Its high-speed interconnects are critical components inside AI server racks and networking equipment.
APH has been the standout performer on fundamentals: 52% revenue growth in 2025, record 27.5% operating margins, and a book-to-bill ratio of 1.31x in Q4. IT datacom revenue has been growing at 100%+ rates as AI cluster buildouts accelerate demand for high-speed optical and copper interconnects. The CommScope acquisition adds communications infrastructure exposure. At 30.9x forward P/E, APH is priced for continued hyper-growth — but with $8.4 billion in Q4 orders, the backlog supports it.
| Metric | Value |
|---|---|
| Market Cap | $165.4B |
| Revenue (TTM) | $23.1B |
| Revenue Growth | 52% YoY |
| EBITDA Margin | 29.8% |
| P/E (fwd) | 30.9x |
| 1Y Price Return | +114% |
APH has the strongest fundamental momentum and the clearest direct AI exposure. Its valuation is elevated but backed by extraordinary order growth.
4. Vertiv Holdings (VRT) — Pure-Play Data Center Cooling and Power
Vertiv designs and manufactures thermal management, power distribution, and infrastructure management systems specifically for data centers. It is the purest data center infrastructure play in public markets.
VRT grew revenue 28% TTM to $10.2 billion, with Q4 2025 revenue of $2.9 billion and margins expanding rapidly — EBITDA margin hit 21.1%. The AI-driven shift to liquid cooling for GPU-dense racks is a major tailwind, and Vertiv is the market leader. However, at 44.6x forward P/E and 53x forward EV/EBITDA, VRT is priced for sustained 20%+ growth. Any deceleration in hyperscaler ordering would hit this stock hardest given its concentration.
| Metric | Value |
|---|---|
| Market Cap | $102.6B |
| Revenue (TTM) | $10.2B |
| Revenue Growth | 28% YoY |
| EBITDA Margin | 21.1% |
| P/E (fwd) | 44.6x |
| 1Y Price Return | +214% |
VRT offers the most direct AI data center exposure but carries the highest concentration risk if capex normalizes faster than expected.
5. Quanta Services (PWR) — The Picks-and-Shovels Builder
Quanta is the largest specialty infrastructure contractor in the U.S., building the transmission lines, substations, and grid connections that data centers need to operate. It is the physical execution layer of the AI buildout.
PWR generated $28.4 billion in TTM revenue with 20% growth, driven by electric power infrastructure and renewable energy projects. The challenge is margins: at 8.9% EBITDA margin, Quanta operates on thin contractor economics, yet the stock trades at 43.6x forward P/E — a valuation that implies either massive margin expansion or sustained revenue acceleration. Q3 2025 showed strong execution with $7.5 billion in revenue and improving margins, but Q4 operating margins slipped.
| Metric | Value |
|---|---|
| Market Cap | $84.9B |
| Revenue (TTM) | $28.4B |
| Revenue Growth | 20% YoY |
| EBITDA Margin | 8.9% |
| P/E (fwd) | 43.6x |
| 1Y Price Return | +128% |
PWR has real backlog visibility from grid expansion, but its low-margin business model makes the current valuation the hardest to justify in this group.
The Verdict: Ranking the Picks
Eaton (ETN) offers the best risk-reward: diversified revenue streams, 22.6% EBITDA margins, and the most reasonable valuation at 26.7x forward P/E. Amphenol (APH) has the strongest fundamental momentum with 52% revenue growth and record order books, justifying a higher multiple. GE Vernova (GEV) has the most compelling structural thesis — grid bottlenecks are real and multi-year — but at 59x forward P/E, the stock needs flawless execution to grow into its valuation. Vertiv (VRT) and Quanta (PWR) are the most vulnerable to a re-rating: VRT because of its pure-play concentration, and PWR because its thin margins don't support a 44x multiple if growth decelerates.
Risks to Watch
- Hyperscaler capex moderation: Microsoft's data center pauses in early 2026 signal that the spending trajectory is not linear — any broader pullback would compress multiples across the group
- Interest rate sensitivity: These stocks have rerated sharply higher on growth expectations; rising rates would pressure the elevated multiples disproportionately
- Supply chain normalization: Much of the premium reflects supply scarcity (transformer lead times, switchgear backlogs); as capacity catches up, pricing power may fade
What to Monitor
- Hyperscaler quarterly capex guidance and data center lease commitments — the leading indicator for the entire chain
- Transformer and switchgear lead times (currently 2–3 years); any shortening signals demand peaking
- Margin trends at GEV and PWR — the two with the most room for expansion, but also the most execution risk