The AI Power Build-Out's Picks and Shovels: 8 Stocks That Get Paid No Matter Who Wins
The AI Power Build-Out's Picks and Shovels: 8 Stocks That Get Paid No Matter Who Wins
The businesses that get paid either way
Every AI headline is about who builds the smartest model. I'd rather own the companies that get paid no matter which giant wins. Building your own power plant — the mandate now sweeping through data centers — takes fuel, turbines, and wiring, and a surprisingly small group of companies supplies all three. Here are the eight I keep coming back to, roughly in the order the electrons flow: from the gas in the ground to the breaker on the rack.
1. Energy Transfer (ET) — the fuel already in the pipe
Energy Transfer is one of the largest natural gas pipeline operators in the country, running more than 130,000 miles of pipe, with close to 90% of earnings coming from long-term, fee-based contracts. It gets paid to move and store gas regardless of energy prices. Nobody can rebuild a network this size, so when a new data center needs gas, the fastest path is often a pipe Energy Transfer already owns — gas is already flowing to Oracle's data centers, with deals tied to campuses like Fermi and Cloud Burst and new pipe going into the ground in Texas. The payout is the kicker: a 7% dividend, more than four times the S&P 500 average, covered a little over twice by cash flow. Someone who bought around 6 dollars a unit in 2021 has more than tripled their money and collected 7% the whole way. Wall Street sees over 20% upside in 12 months.
2. Eaton (ETN) — the electrical backbone
If Energy Transfer moves the fuel, Eaton moves the electricity once it's made. It builds the switchgear, breakers, and systems that carry raw power to thousands of servers without letting it drop. What makes it a true picks-and-shovels play: it doesn't matter whether the electricity comes from a gas turbine, a fuel cell, or the grid — all of it runs through Eaton's equipment. Data center orders jumped 240% in a single year, the backlog is past 14 billion dollars, and data centers are now more than a fifth of sales. Eaton also co-developed an 800-volt power system with Nvidia for dense AI racks and bought Boyd Thermal to move into liquid cooling — so it sells both power and cooling into the same building. EPS has doubled since 2019, margins climbed from ~14% to nearly 20%, the stock has more than tripled since 2021, and the dividend has risen 16 straight years. Analysts peg over 15% upside.
3. GE Vernova (GEV) — the turbine itself
GE Vernova builds the giant gas turbines these on-site plants require. Spun out of General Electric in 2024, it's one of only a handful of Western companies that can build large-frame gas turbines at scale, and its factories are capped near 10 gigawatts a year. Demand has blown so far past that ceiling that the order book is sold out through 2030 — which hands it enormous pricing power, so it now charges 10–20% more per turbine and buyers pay because they have nowhere else to go. Revenue grows only single digits, but profit margins are set to more than double in two years, and free cash flow swung from burning billions to nearly 4 billion dollars, with guidance to roughly double again. It's also building one of the first Western small reactors, so it's on the nuclear side of the trade too. Since spinning off around 140 dollars in 2024, the stock is up nearly 8x. Analysts still see over 20% upside.
4. Howmet Aerospace (HWM) — the pick behind the pick
A gas turbine is only as good as the blades spinning inside it, and those single-crystal castings survive temperatures hotter than the metal's own melting point. Almost nobody can make them — Howmet is the largest that can, and the only other one at scale is buried inside Berkshire Hathaway. That choke point comes with real pricing power: as jet-engine demand roared back, EPS grew more than 6x since 2021, margins climbed from ~17% to nearly 26%, and the stock rose roughly 10x. Now the same casting expertise makes blades for data center gas turbines, and Howmet expects that industrial-turbine business to double in the next few years. Analysts see about 14%.
5. Baker Hughes (BKR) — the cheap, overlooked turbine maker
Most people file Baker Hughes under oil and gas, but tucked inside is one of the few makers of aeroderivative gas turbines — the lighter, faster-to-install cousins of GE Vernova's heavy frames, exactly what a data center grabs when it can't wait years. In a single quarter it booked around 1 billion dollars of data center power orders, matching an entire prior year, and it's buying Chart Industries to bolt on advanced cooling. Over four years it doubled operating margins from ~6% to nearly 13%, grew free cash flow, and built a record backlog — yet the stock has moved a fraction of the flashier names because Wall Street still calls it an oil-services company. Of the whole group, analysts hand it the most room: over 35%.
6. Cummins (CMI) — the income-paying way in
You know Cummins for truck engines, which is exactly why the market keeps underrating it. Bolted onto that steady, slightly dull business is a power-generation arm building the large natural gas engines a data center drops on site to run its own microgrid. Cummins roughly doubled data center revenue to around 3.5 billion dollars, power-gen sales grew nearly 30% in a single quarter, and that segment just posted a record margin near 30% — remarkable for a heavy industrial. Its data center order book runs through 2028, it trades around 23x forward earnings (a fraction of the pure plays), has raised dividends 15+ years, and the stock has roughly tripled since early 2021. Analysts give it about 20%.
7. Williams Companies (WMB) — the pipeline turning into a power company
Williams runs one of the country's largest gas pipeline systems, but it made a shrewd move: it realized it could earn far more building the power plant at the data center than just shipping the fuel. For Meta, it's putting up a 2 billion dollar, 400-megawatt gas plant right at the campus, fully off-grid, with Meta paying Williams on a long-term contract. That's a higher-margin, longer-lasting business than pushing molecules through a pipe, and Williams has committed more than 5 billion dollars to build a fleet of them. Cash earnings grew from about 4 billion dollars in 2019 to over 7 billion today, and the stock has more than tripled since early 2021. Analysts figure at least 15% more.
8. EQT (EQT) — the cheapest, most direct bet
Every on-site gas plant needs an enormous, steady supply of gas, and EQT is the largest natural gas producer in the country, sitting on the Marcellus Shale in Appalachia and producing more of it at lower cost than anyone else. That scale won it an exclusive deal to supply Homer City, a 4.4-gigawatt AI campus rising on the site of Pennsylvania's former largest coal plant. Here's the twist: while the pipeline names have run up hard, EQT is basically flat on the year and still cheap, because its fortunes rise and fall with gas prices — which is exactly what gives it the most torque. A company losing money a few years ago now throws off nearly 3 billion dollars in free cash flow, and any real lift in gas demand drops almost straight to the bottom line. Analysts see over 34% upside.
How the eight stack up
| Company | Role in the build-out | Standout number | Analyst 12-mo upside |
|---|---|---|---|
| Energy Transfer (ET) | Moves the gas | 7% dividend, ~2x covered | 20%+ |
| Eaton (ETN) | Wires the power | Orders +240%, 14B backlog | 15%+ |
| GE Vernova (GEV) | Builds the turbine | Sold out to 2030 | 20%+ |
| Howmet (HWM) | Casts the blades | Margins ~17%→26% | ~14% |
| Baker Hughes (BKR) | Aeroderivative turbines | ~1B orders in a quarter | 35%+ |
| Cummins (CMI) | On-site gas engines | Segment margin ~30% | ~20% |
| Williams (WMB) | Builds & owns plants | 2B Meta plant, off-grid | 15%+ |
| EQT (EQT) | Produces the gas | ~3B free cash flow | 34%+ |
If you want the pipeline-and-producer end specifically, I went deeper on EQT and Williams alongside NRG here, and on the grid-component angle in this picks-and-shovels breakdown.
FAQ
Q: Why own "picks and shovels" instead of the data center operators themselves? A: Because you don't have to guess which hyperscaler wins. Fuel, turbines, and wiring get bought no matter which campus succeeds, so the demand is spread across the whole build-out rather than concentrated in one bet.
Q: Which of the eight is the cheapest right now? A: On the numbers, Baker Hughes and EQT are the least-noticed and carry the highest analyst upside (35%+ and 34%+), largely because the market still files them under oil and gas rather than AI power.
Q: Isn't this just the nuclear trade in disguise? A: No. Nuclear is the long game — first US small reactors aren't expected until around 2030. This group is built on natural gas, which can be running in about a year, so it captures the demand that can't wait. GE Vernova notably plays both sides.
This is educational, not financial advice. Do your own due diligence.
More in this Category
Sales Exploding, Stock Stuck: The Complete Nvidia Bull vs Bear Case
Sales Exploding, Stock Stuck: The Complete Nvidia Bull vs Bear Case
Nvidia's revenue rocketed from $16B in 2021 to $253B in under five years, yet the stock has trailed AMD and Micron. Here's my read on Jensen Huang's 'parabolic demand' claim, the three bull cases, and the three bear cases.
Nvidia's Valuation: What's a Fair Price to Pay Right Now
Nvidia's Valuation: What's a Fair Price to Pay Right Now
A $5 trillion market cap, a 19.6x price-to-sales ratio, and a 63% one-year net margin. Running a conservative 10-year model (10-25% revenue growth, 35-55% margins), I get a mid fair value of $250 at a 9% required return, and $154 at my personal 15%.
Getting Paid to Hold Nvidia: Understanding the Covered Call
Getting Paid to Hold Nvidia: Understanding the Covered Call
If you're torn between selling Nvidia and holding it, a covered call can be the answer. Selling a Sept 18 $250 call pays about $3.37 per share (roughly 8.8% annualized); a $220 call pays $10.39 (about 27%). Here's how it works and where it bites.
Next Posts
How to Find Stocks the Market Is Practically Giving Away: Hunting Near 52-Week Lows
How to Find Stocks the Market Is Practically Giving Away: Hunting Near 52-Week Lows
With the lowest-returning stock among the S&P 500's top 20 still up 77% while the bottom 300 are negative, I hunt near 52-week lows. Here's my process for separating names like Intuit, Adobe, and PayPal — where revenue and profit are rising but the price fell — from businesses that are actually broken.
Is Ulta Beauty a Broken Retailer, or a Number-One Store on Sale?
Is Ulta Beauty a Broken Retailer, or a Number-One Store on Sale?
Ulta Beauty is down over 23% this year near its 52-week low, yet it trades at 17x free cash flow with double-digit revenue growth and a 40% gross margin. My DCF puts fair value near $610 (versus $477 today), about a 12.5% expected return.
Two Software Giants Under AI Fear: Adobe vs. Salesforce Valuation
Two Software Giants Under AI Fear: Adobe vs. Salesforce Valuation
Adobe and Salesforce are both down about 30% this year, yet trade at just 9.5x and 10.25x free cash flow. Under my midpoint DCF, I get roughly 22% expected return for Adobe and 19.5% for Salesforce.
Previous Posts
Eaton vs BWXT: Two Ways to Own the Power Layer of AI
Eaton vs BWXT: Two Ways to Own the Power Layer of AI
A single year of AI data-center construction opens up roughly $46 billion in electrical-equipment demand. I compare Eaton, the power-management giant, with BWXT, the only profitable nuclear name, as buys on the pullback.
Nvidia, Broadcom, Alphabet, Arista: Four Monopolies of the AI Chip War
Nvidia, Broadcom, Alphabet, Arista: Four Monopolies of the AI Chip War
From Nvidia, whose revenue grew eightfold in three years, to Alphabet, sitting on a $460 billion cloud backlog, I break down the monopoly structure and PEG valuation of the four AI-infrastructure megacaps.
Rambus and Fabrinet: Small-Cap AI Picks-and-Shovels and the Asymmetric Basket
Rambus and Fabrinet: Small-Cap AI Picks-and-Shovels and the Asymmetric Basket
In June the chip index fell about 10% in days, but Micron’s 346% revenue jump showed the AI build-out speeding up, not cooling. I cover the high-risk small caps Rambus and Fabrinet and how to size them small.