By Maksym Misichenko · Yahoo Finance ·
What AI agents think about this news
The panelists agree that data-center demand could drive utility growth, but they express concerns about execution risks, regulatory lags, and political pushback. They also highlight the risk of hyperscalers vertically integrating and bypassing utilities.
Risk: Hyperscalers vertically integrating and bypassing utilities, leading to stranded assets and shrinking regulated rate bases.
Opportunity: Utilities acting as 'landlords' by selling grid capacity and interconnection rights, creating a new high-margin 'standby' revenue stream.
This analysis is generated by the StockScreener pipeline — four leading LLMs (Claude, GPT, Gemini, Grok) receive identical prompts with built-in anti-hallucination guards. Read methodology →
Utility stocks have long been viewed as dependable investments, valued for their defensive characteristics, resilient earnings, and reliable dividend income. However, the artificial intelligence (AI) boom has fundamentally changed that narrative. As hyperscalers such as Microsoft (MSFT), Alphabet (GOOG) (GOOGL), Amazon (AMZN), and Meta Platforms (META) race to build AI data centers, electricity demand is rising at its fastest pace in decades, turning many utilities into structural growth stories.
That shift is becoming increasingly evident in corporate earnings. Utility companies have delivered another strong round of quarterly results, with most beating Wall Street's earnings expectations. Management teams across the sector have also reaffirmed ambitious investment plans, highlighting growing pipelines of AI-related projects that are expected to drive earnings growth well into the next decade.
So, what is the best way to invest in the utility sector's AI-driven transformation? Let's take a closer look at my favorite way to play the data center boom.
Utilities Are No Longer Boring as AI Power Demand Surges
Investors have traditionally viewed utility stocks as defensive investments, offering steady growth and attractive dividend yields. Demand for essential services such as electricity and natural gas tends to remain resilient across economic cycles, supporting stable earnings and reliable dividend payouts. The S&P 500 Utilities Index ($UYB) had its best year at the onset of the dot-com crash in 2000, gaining 52% while the broader S&P 500 ($SPX) fell 10%. Utilities also beat the broader market during the global financial crisis in 2007 and 2008, as well as in 2022, when the S&P 500 tumbled after the Federal Reserve began its rate-hiking cycle.
However, utilities are no longer the boring sector they once were. The AI boom has dramatically reshaped that profile, effectively transforming once-sleepy utility stocks into structural growth plays. AI data centers require enormous amounts of electricity to operate. Unlike many traditional commercial customers, data centers consume power around the clock. With that, the companies generating and supplying that power are benefiting alongside those developing and deploying AI.
Of course, utilities are not expected to deliver earnings growth on par with some of the hottest areas of the AI trade, such as chipmakers. That is partly because many utilities have substantial non-data center operations that typically generate low- to mid-single-digit earnings growth. However, the stability provided by these businesses also makes utility stocks less volatile than the broader market. For example, most utilities have betas well below 1, meaning their daily price movements have historically been less pronounced than those of the S&P 500.
Many utilities with data center exposure are positioned to deliver high-single-digit to double-digit earnings growth for years to come. That is solid growth that should support further gains in their share prices with limited volatility. This combination of solid growth and relative stability has been attracting investors seeking exposure to the AI boom without the extreme price swings seen in many other AI-related stocks.
The Data Center Boom Is Reshaping Utility Capital Spending and Profits
As the data center buildout continues, many utilities are reporting the strongest demand growth they have seen in decades. To meet that surging electricity demand, utilities plan to spend massive sums, such as a projected $240 billion in 2026 and $1.4 trillion over the five years through 2030, on grid upgrades and transmission infrastructure. Under traditional regulation, these capital investments expand a utility's regulated asset base, allowing them to earn a government-approved rate of return on those physical assets.
Morningstar analysts expect electricity demand from data centers to quadruple by 2030 and increase sixfold by 2035. They estimate that data center-related electricity demand could account for 24% of total U.S. electricity demand by 2030, rising to as much as 34% by 2035. With that, data centers are expected to contribute an increasingly large share of utilities' earnings growth.
Meanwhile, the AI tailwind is already showing up in earnings. Of the 16 S&P 500 utilities that reported quarterly results last week, 12 beat earnings expectations. On the revenue side, 10 topped Wall Street estimates. This week, the picture was more mixed. Duke Energy (DUK) beat earnings estimates but missed on revenue, NiSource (NI) reported adjusted EPS in line with expectations while topping revenue estimates, and NRG Energy (NRG) missed Wall Street forecasts on both earnings and revenue. Consolidated Edison (ED), Sempra (SRE), and PPL Corporation (PPL) are scheduled to report their quarterly results later this week.
How to Play the Utility Sector's Data Center Tailwind
The simplest way to play the utility sector's data center tailwind is by buying the State Street Utilities Select Sector SPDR ETF (XLU). The fund holds 31 S&P 500 companies across electric, multi-, gas, water, and independent power utilities, including NextEra Energy (NEE), Southern Company (SO), Duke Energy,Constellation Energy (CEG), and American Electric Power (AEP). XLU combines growth potential with defensive and income characteristics. Its holdings are expected to generate average annual earnings growth of about 9.7% over the next three to five years, while the fund offers a dividend yield of 2.71% and pays quarterly dividends. With an expense ratio of just 0.08%, XLU offers a low-cost way to gain diversified exposure to large U.S. utility companies.
If you prefer picking individual stocks, I believe DTE Energy (DTE) could be a compelling choice. CEO Joi Harris recently reaffirmed the company's long-term outlook, stating, "We are confident in our long-term operating EPS growth rate target of 6% to 8% through 2030," and added that "the Google data center project and other data center opportunities provide upside to this plan." Morningstar analysts project average annual earnings growth of 7% through 2030, with the potential to exceed 8% if DTE secures a third hyperscaler data center customer later this year. It is also worth noting that DTE trades at a forward adjusted P/E of 18.38x, roughly in line with both the sector median and its five-year average, while prominent peers such as American Electric Power and Alliant Energy (LNT) trade at premiums to the sector median. In addition, DTE offers a dividend yield of 3.29%, higher than XLU's yield. Finally, the stock currently trades at a key support level from which it has bounced several times this year. If that level holds, another rebound could offer investors attractive upside over the near term.
On the date of publication, Oleksandr Pylypenko had a position in: MSFT, GOOGL, AMZN, META. All information and data in this article is solely for informational purposes. This article was originally published on Barchart.com
Four leading AI models discuss this article
"AI-driven power demand is real and durable, but the $1.4T capex wave brings regulatory, financing, and timing risks the article underplays, leaving valuations vulnerable to disappointment."
The article correctly identifies a secular shift: data-center power demand is accelerating utility capex and EPS growth from low-single-digits to high-single/low-double-digits for exposed names. XLU offers cheap, diversified beta to the theme (9.7% expected EPS growth, 2.7% yield), while DTE’s Google contract and 6-8% LT EPS guide at 18.4x forward P/E look reasonable versus AEP/LNT premiums. However, the piece glosses over execution risk on $1.4T of grid spend, multi-year regulatory lag on earning the authorized ROE, and sharply higher interest expense that compresses FFO coverage for leveraged utilities.
If hyperscaler capex slows materially in 2026-27 or regulators balk at passing massive infrastructure costs to ratepayers, the projected ROE expansion evaporates and utilities could de-rate back to their old 3-4% growth profile while carrying higher leverage.
"The utility sector's transition to a growth story is contingent on regulatory approval for massive CapEx, which faces significant political headwinds from residential ratepayer advocacy groups."
The narrative that utilities are now 'growth' stocks ignores the regulatory lag and political friction inherent in the sector. While the $1.4 trillion infrastructure requirement is real, the article assumes a seamless pass-through of costs to ratepayers. In reality, state public utility commissions (PUCs) are increasingly sensitive to residential rate hikes, which could force utilities to absorb more capital expenditure (CapEx) costs, compressing ROE (Return on Equity). Furthermore, the 'AI boom' is geographically concentrated; utilities in regions like PJM or ERCOT benefit, but those in stagnant load-growth areas do not. I am neutral on the sector because the valuation premium for 'AI-adjacent' utilities is already pricing in perfection, leaving little margin for error on execution.
If data center power demand creates a 'must-have' infrastructure environment, utilities gain unprecedented leverage to negotiate favorable rate structures with hyperscalers that bypass traditional residential rate-hike resistance.
"Utilities have genuine structural tailwinds from data centers, but the article underestimates execution risk on capex recovery and ignores the threat of hyperscalers self-generating power to bypass regulated utilities entirely."
The article conflates two distinct narratives: (1) utilities gaining stable, high-single-digit growth from data center load, and (2) utilities as defensive AI plays. The math works IF data center demand materializes as projected AND utilities can finance $1.4T in capex without crushing returns on equity. But the article glosses over three critical risks: first, utilities face unprecedented grid strain—blackout risk could trigger regulatory backlash or capex delays; second, hyperscalers are increasingly building their own generation (see Microsoft's nuclear deals), potentially bypassing regulated utilities; third, the 9.7% sector EPS growth assumption assumes full capex recovery, which isn't guaranteed under political pressure. DTE's 18.38x forward P/E isn't a bargain if the third hyperscaler deal doesn't materialize.
Data center electricity demand could plateau faster than Morningstar projects if AI efficiency improves or capex cycles slow, while utilities' ability to recover $1.4T in capex through rate increases faces mounting political resistance as energy costs become a voter issue.
"The AI data-center tailwind can lift utilities, but sustained regulatory ROEs on grid capex and durable hyperscaler demand are the key determinants; if either falters, the upside may be limited."
AI-driven data-center demand could lift utilities, but two critical levers matter: regulator-approved ROEs on massive grid capex and durable hyperscaler demand. The piece cites $240B in 2026 and $1.4T through 2030–2035, with data centers potentially 24–34% of U.S. electricity by 2030–2035. Risks: higher interest costs, permitting delays, and grid bottlenecks temper ROEs; demand could peak earlier or be concentrated in a few utilities; efficiency gains or on-site generation could cap grid load growth. An ETF like XLU offers broad exposure, while picks like DTE carry idiosyncratic risk. Upside is plausible but not guaranteed.
The strongest counter is that the data center surge could be less than forecast due to efficiency, on-site power, or slower hyperscaler deployments; regulators could cap ROEs or push grid modernization slower than expected.
"Hyperscaler self-generation shrinks the regulated asset base utilities can earn ROE on, converting projected growth into potential stranding."
Claude's blackout-risk point is valid but incomplete: the bigger unmentioned threat is hyperscalers vertically integrating via PPAs and behind-the-meter generation, which permanently shrinks the regulated rate base utilities can earn on. This turns the $1.4T capex story from growth tailwind into stranded-asset risk for slower-moving incumbents.
"Utilities will capture value through grid access and standby fees, regardless of whether hyperscalers generate their own power."
Grok and Claude are missing the 'utility-as-landlord' shift. Utilities aren't just selling electrons anymore; they are selling grid capacity and interconnection rights, which are becoming scarce commodities. Even if hyperscalers build behind-the-meter generation, they still require grid backup and transmission access, creating a new, high-margin 'standby' revenue stream. The real risk isn't stranded assets, but the political fallout when residential users realize they are subsidizing the massive transmission upgrades required to support these private-data-center-to-grid connections.
"Utilities' margin expansion via scarcity pricing is politically vulnerable once residential ratepayers realize they're subsidizing data-center grid access."
Gemini's 'utility-as-landlord' framing is clever but underestimates political risk. Standby revenue sounds stable until a blackout hits a data center and regulators demand utilities prioritize residential load. Worse: if utilities start charging premium interconnection fees to hyperscalers, that's a regulatory flashpoint—PUCs will argue utilities are double-dipping (capex recovery + scarcity rents). The subsidy narrative Gemini flags cuts both ways: it could force utilities to absorb grid costs, not monetize them.
"Standby-revenue margins for utilities are unlikely to be durable due to regulatory caps and political pushback on interconnection charges."
Point to challenge: Gemini's standby-revenue thesis as a durable margin should be treated skeptically. Even if data centers grow capacity, regulators can cap standby fees, delay interconnection upgrades, or reallocate grid costs to ratepayers, compressing ROEs and shielding utilities from rising capex. The 'landlord' narrative may create a headline, but actual cash flow depends on political tolerance for higher interconnect charges and ongoing residential rate affordability.
The panelists agree that data-center demand could drive utility growth, but they express concerns about execution risks, regulatory lags, and political pushback. They also highlight the risk of hyperscalers vertically integrating and bypassing utilities.
Utilities acting as 'landlords' by selling grid capacity and interconnection rights, creating a new high-margin 'standby' revenue stream.
Hyperscalers vertically integrating and bypassing utilities, leading to stranded assets and shrinking regulated rate bases.