3 Energy Stocks to Buy as AI Power Demand Surges—and 2 to Avoid

U.S. Gov’t To Shock Gold Investors On April 15 (From Paradigm Press)

The biggest names in energy and technology are all in the same room this week—and the conversation isn't about oil prices. It's about electricity. And that distinction matters.

Often dubbed the "Super Bowl of energy," CERAWeek is the world’s premier annual energy conference, where the big players convene in Houston, Texas, to discuss global energy markets, geopolitics, and technology.

This year, speakers from Amazon Web Services (NASDAQ: AMZN), Alphabet's Google (NASDAQ: GOOGL), Microsoft (NASDAQ: MSFT), NVIDIA (NASDAQ: NVDA), and Meta (NASDAQ: META) are sharing the stage with legacy energy producers, and the dominant theme is power demand.

Altimetry Research’s Joel Litman and Rob Spivey are focused on one major theme emerging from the conference: the United States is not energy-independent when it comes to electricity, and the AI-driven buildout could take five to ten years. That creates a specific, investable opportunity—and a few traps worth avoiding.

U.S. Electricity Demand Is Outpacing the Grid

For roughly 15 years, electricity usage in the United States barely moved, even as GDP grew. But that changed around 2022.

Even before the latest geopolitical concerns in the Middle East, power demand was already rising. Reindustrialization, data center proliferation, and the rise of AI computing have sent power consumption sharply higher. Data center electricity demand alone could account for as much as 10% of total U.S. usage, and the infrastructure to support it simply doesn't exist yet.

That's the tension at CERAWeek this year. Energy producers and AI hyperscalers are negotiating who builds what—and who pays for it. Residential electricity rates still exceed commercial rates on a per-kilowatt-hour basis, a dynamic that could become politically explosive heading into November's elections. Companies that need power may increasingly be forced to source it at market prices or off-grid entirely, which only accelerates total demand.

3 Stocks Positioned to Profit From the AI Power Buildout

1. MasTec: The Builder Behind the Buildout

MasTec (NYSE: MTZ) is the engineering, procurement, and construction firm that physically builds power plants, lays fiber-optic cable, and constructs data centers. The company's client list reads like a who's who of the energy-AI convergence: Kinder Morgan (NYSE: KMI), Duke Energy (NYSE: DUK), AT&T (NYSE: T), IBM (NYSE: IBM), and Microsoft.

What makes the case compelling is what standard financial reporting misses.

According to Altimetry, MasTec is roughly twice as profitable as reported metrics suggest. The company carries an approximately $19 billion backlog—a figure that gives it years of revenue visibility.

Management guided for $17 billion in 2026 revenue, representing 19% growth, and adjusted earnings per share (EPS) of $8.40. The record $18.96 billion 18-month backlog gives that guidance unusual credibility.

The market, however, is pricing MasTec for a normal economic cycle. It's not pricing a multi-year infrastructure supercycle. That disconnect is the opportunity.

Altimetry's research on "doubles that double again" found that in the middle of a bull market, stocks that have already doubled carry a roughly 50% chance of doubling again, and uniform accounting filters push that probability closer to 60%.

2. Regal Rexnord: Solving the Power Problem Inside the Data Center

Regal Rexnord (NYSE: RRX) tells a different story. This legacy industrial company—historically known for motors, machine parts, and HVAC components—has moved up the value chain into data center power management, and the market hasn't fully caught on.

The key product is the E-Pod, a modular, plug-and-play power management system roughly the size of a shipping container. It steps down and manages the electrical load coming into a data center so high-value chips from NVIDIA and Micron (NASDAQ: MU) don't fry.

In Q4 2025, the company secured orders worth approximately $735 million for multiple E-Pod projects. The broader data center business could reach $1 billion in revenue over the next two years, up from roughly $120 million today.

Regal Rexnord's return on assets has climbed by about a third over the past few years as it shifted toward higher-margin solutions. But reported metrics don't capture the transformation. Recent stock volatility—driven partly by geopolitical jitters and recurring "AI spending is over" scares—may actually offer a more attractive entry point.

The distinction Altimetry draws is worth repeating: the current AI investment cycle is nothing like the dot-com bubble. In 1998 and 1999, capital flowed to companies with no revenue, let alone profits. Today, the spending is coming from massively cash-rich hyperscalers with demand they can't yet fulfill. Microsoft's Satya Nadella has said publicly that Azure would generate more revenue if the company simply had more power and more data centers.

3. EQT: The Natural Gas Bridge That Funds the Future

EQT (NYSE: EQT) is the largest natural gas exploration and production company in the United States, and Altimetry calls it the essential near-term cog in the AI power story.

The logic is straightforward: while nuclear and renewables generate long-term excitement, natural gas is the only viable baseload power source that can be deployed at scale in the next five years.

Solar doesn't run at night. Wind can't operate when it's too calm or too windy. Battery storage extends capacity for two to four hours, far short of overnight demand. If the United States needs to rapidly build new power plants for data centers, those plants will run on natural gas.

EQT holds nine years of reserves without drilling a single new well and 12 years of proven reserves if it ramps up. The company's vertical integration makes it one of the country's lowest-cost gas producers at $2 per MMBtu.

Management guided for 2026 adjusted EBITDA of about $6.5 billion and free cash flow of $3.5 billion. The company is also completely unhedged for 2026—a deliberate bet by management that natural gas prices will move higher.

The dual catalyst here is domestic power demand and LNG exports. Geopolitical disruption in the Middle East is reinforcing the case for U.S. energy exports, giving EQT upside on both sides of the ledger. Stock volatility reflects short-term geopolitical skittishness, not a fundamental problem. It could even represent a buying opportunity.

2 AI Power Plays That Look More Like Hype Than Opportunity

1. CoreWeave: The WeWork of AI?

Now for the names to avoid. The first is CoreWeave (NASDAQ: CRWV), and the comparison Altimetry draws is blunt: CoreWeave is the WeWork of the AI boom.

The pitch sounds compelling on the surface: CoreWeave builds and operates data centers for AI workloads. But Altimetry argues the company is functionally a data center REIT with a slicker brand, and claims it has never generated a dime of actual profitability. The company posted a negative 22.74% profit margin and a negative 50.27% return on equity.

Yet the market is pricing CoreWeave for return on assets north of 25%, roughly five times what comparable data center operators typically achieve. The company carries a $29.8 billion debt load with a 0.46 current ratio and 16.5% short interest. Even as revenue surges, capital expenditures are expected to more than double in 2026, constraining any path to near-term profitability.

Altimery believes that while metrics may try to suggest profitability, the underlying economics tell a different story. CoreWeave's economic profit has been negative since it went public, and Altimetry doesn't see a reason for that to change soon.

2. Oklo: A Cool Idea Still Years Away From Reality

Altimetry's critique isn't about nuclear energy broadly—it's about Oklo (NYSE: OKLO) specifically.

The small modular reactor company captured investor imagination with a partnership with Meta and backing from Sam Altman. But it is behind at least two competitors (NuScale (NYSE: SMR) and BWXT (NYSE: BWXT) on the technology curve. And, more importantly, the business model is misunderstood. Oklo doesn't plan to sell reactors. It plans to build them and lease the power, making it fundamentally a leasing business with cost-of-capital-level returns.

The math doesn't work at current pricing. New-build nuclear power costs roughly $200 to $250 per megawatt hour. Hyperscalers are currently contracting power in the mid-hundreds of dollars per megawatt hour.

The market, meanwhile, is pricing Oklo for $400 to $500 million in earnings when the company is currently losing $100 million per year. Oklo has around $1.2 billion in cash and marketable securities, which provides runway, but a cash cushion doesn't change the economics of a leasing model that may never reach the return profile investors are pricing in.

If the SMR thesis does play out, Altimetry suggests watching BWXT. The company already manufactures key components for the U.S. Navy's nuclear reactors and has decades of proven technology, is generating revenue today, and carries less speculative premium.

Where Power Meets Profit

The through-line across all five names is the same: the AI power buildout is real, it's massive, and it's early.

But not every company riding the narrative deserves investor capital. The companies with proven demand, deep backlogs, and underappreciated profitability—MasTec, Regal Rexnord, and EQT—look positioned to capture years of growth. The ones trading on hype and venture capital packaging—CoreWeave and Oklo—could leave investors holding expensive lessons.

The real signal from CERAWeek isn't about any single stock. It's that the convergence of energy and AI is now the defining investment theme of this cycle, and the companies that physically build, power, and fuel that infrastructure may be the smartest way to play it.

Read this article online ›

Related Video: