The Case for Boring Utilities

The Setup

For the past 20 years, U.S. electricity demand was essentially flat. The grid grew at less than 1% annually. Utilities were rate-regulated bond proxies: predictable, yield-oriented, and perpetually boring. The investment thesis was simple: collect your 3% to 4% dividend, sleep well, and accept that you would never beat the S&P 500 in a good year.

That era is over. The artificial intelligence infrastructure buildout has created the first real load growth cycle in the American power sector in a generation. Data centers consumed roughly 4% of total U.S. electricity in 2024. By 2030, multiple independent forecasts project that figure will climb to 9% to 12%, with total data center power demand reaching 108 to 134 GW, nearly tripling current levels.

The hyperscalers are not being subtle about what they need. Total AI infrastructure spending from Amazon, Google, Meta, and Microsoft approaches $700 billion in 2026 alone, roughly double what they spent in 2025. In March 2026, seven of the largest AI companies signed a White House Ratepayer Protection Pledge committing to build, procure, or fund the new power generation their data centers will consume, and to pay for all grid upgrades without passing costs to residential customers.

They need power. They need it now. And they need it at a scale that exceeds anything the grid has been asked to deliver in decades. The companies positioned to deliver that power are the ones this thesis is about.

The Power Demand Nobody Is Ready For

The numbers are worth understanding because they are what makes this cycle structurally different from anything in utility history.

AI-optimized servers are the primary driver. According to Gartner, AI server electricity usage is projected to rise nearly fivefold, from 93 TWh in 2025 to 432 TWh in 2030. By that point, AI workloads will represent 44% of total data center power consumption, up from 21% today.

For context, a large nuclear power plant generates roughly 1 GW. The projected increase in data center demand between now and 2030 is equivalent to building 60 to 100 new nuclear reactors. That is not an infrastructure challenge. That is an infrastructure emergency, and it is the kind of emergency that creates pricing power for the companies that already own generation capacity.

The Department of Energy has characterized this as a doubling or tripling of data center electricity consumption by 2028. Lawrence Berkeley National Laboratory estimates data centers could consume between 325 TWh and 580 TWh annually by 2030. Even the low end of that range represents a fundamental transformation of who is buying electricity in America and how much they are willing to pay for it.

The buyer quality matters as much as the volume. Amazon, Google, Microsoft, and Meta are among the most creditworthy counterparties on Earth. They are signing 15 to 20 year power purchase agreements at fixed prices, with commitments to pay whether or not they consume the electricity. These are not speculative customers. They are the best credit quality a utility has ever seen in a customer base, and they are buying power at a scale that dwarfs residential demand growth.

Three Names, Three Angles

Each of these three companies is positioned differently within the same structural tailwind, giving investors a choice of risk profiles and growth characteristics.

NextEra is the largest electric power and energy infrastructure company in North America and the largest producer of wind and solar energy on the continent. Its regulated utility subsidiary, Florida Power & Light, serves over 6 million customer accounts in one of the most constructive regulatory environments in the country and contributes roughly 70% of consolidated operating earnings.

The AI play at NextEra is scale. In March 2026, President Trump approved the company's plans to develop up to 10 GW of natural gas generation in Texas and Pennsylvania as part of a U.S.-Japan trade deal. The projects are designed specifically to serve data centers and advanced manufacturing. CEO John Ketchum has outlined a "15 by 35" strategy: 15 GW of new generation for data center hubs by 2035, a target he has said he expects to double to 30 GW. NextEra currently has roughly 30 energy hub sites in its development inventory and is targeting 40 by year-end.

NextEra also has an expanded partnership with Google Cloud to develop gigawatt-scale data center campuses with associated generation, building on an existing relationship covering more than 3.5 GW of power in operation or under contract. The company's contracted backlog stands at 30 GW, with 95% committed under long-term agreements.

Southern Company is the nuclear story. With Vogtle Units 3 and 4 now fully operational, Southern is the only U.S. utility with significant recent experience in large-scale nuclear deployment. These are the first newly built commercial nuclear units in the United States in approximately three decades. The four-unit Vogtle plant is now the largest generator of clean energy in the country, producing more than 30 million MWh annually with an average capacity factor of 94% over the past decade.

The Vogtle saga was painful. It took roughly 15 years and cost over $36 billion, running years behind schedule and billions over budget. For investors who endured that period, the payoff is arriving at exactly the right moment. Having this operational now, ahead of what could be a golden age for nuclear power demand, is the kind of timing advantage that cannot be replicated by competitors.

Southern's $81 billion five-year capital plan (2026 to 2030) is one of the largest in the industry. Discussions have already begun regarding a potential Vogtle Unit 5 or small modular reactors. The company guides for 5% to 7% annualized EPS growth, with management suggesting upside potential. Mizuho Securities upgraded the stock to Outperform in February 2026, citing its unique exposure to the AI power theme.

Constellation is the pure-play nuclear growth story. After acquiring Calpine, it became the largest private-sector power producer in the world, with approximately 55 GW of capacity, roughly two-thirds of it carbon-free. It operates the largest nuclear fleet in the United States: 21 reactors across 12 sites.

The headline catalyst is the Three Mile Island restart. Constellation is investing $1.6 billion, backed by a $1 billion DOE loan, to restart TMI Unit 1 (renamed the Crane Clean Energy Center) under a 20-year power purchase agreement with Microsoft. The 835 MW reactor is expected to come back online in 2027 or 2028 and will supply power exclusively for Microsoft's data center operations. It would be the first retired U.S. nuclear plant brought back to life for a single commercial customer.

Constellation also signed a 20-year PPA with Meta for the entire 1.1 GW output of its Clinton, Illinois, nuclear plant, beginning June 2027. These deals illustrate something important about the nuclear value proposition: hyperscalers are willing to sign multi-decade contracts at premium prices for carbon-free, 24/7 baseload power. No other generation source can guarantee that combination at scale.

The stock has pulled back 34% from its October 2025 peak of $413, in part because 2026 EPS guidance came in below consensus and no new hyperscaler contract was announced at the time. But the company targets 20% compounded base EPS growth through 2029, and 147 million MWh of clean nuclear generation remains uncontracted for 2030 and beyond. That is a massive pipeline of potential deals sitting in inventory.

The Contracts Nobody Is Talking About

The structural shift in this cycle is not just that demand is rising. It is that the nature of utility customer relationships is fundamentally changing.

Traditionally, utilities serve a broad base of residential, commercial, and industrial customers through regulated tariffs. The customer base is diffuse, the credit quality is mixed, and individual customer concentration is low. The AI power cycle introduces a new category: hyperscale customers with investment-grade credit ratings, willing to sign 15 to 20 year fixed-price power purchase agreements, and willing to commit to paying for contracted power whether or not they consume it.

Microsoft has surpassed Amazon as the largest corporate buyer of clean power, with 34.7 GW contracted as of late 2025. U.S. data centers have collectively contracted more than 80 GW of clean energy to date. These are not speculative grid connections. They are signed, funded, long-term agreements with the most creditworthy counterparties in the world.

The hyperscalers are also moving beyond traditional PPAs into direct ownership of generation assets and co-location arrangements where data centers are built adjacent to or directly connected to power plants. Google acquired clean energy developer Intersect Power outright. Amazon signed a 17-year PPA with Talen Energy for 1.92 GW from the Susquehanna nuclear plant. Meta signed agreements with three nuclear providers for its Ohio and Louisiana data center campuses, including a facility that will be the world's first data center requiring over 1 GW of power when construction completes.

For utilities positioned on the right side of this shift, these contracts provide something the sector has never had: decade-scale revenue visibility backed by the strongest credit quality in the corporate universe.

Why the Market Is Still Sleeping

Utility investors are conditioned by decades of precedent to treat the sector as rate-sensitive bond proxies. When the 10-year Treasury yield rises, utility stocks fall, because institutional investors treat them as interchangeable with fixed income. That framing made sense in an era of flat electricity demand and purely regulated returns.

It is the wrong framing for this cycle. The data center power buildout is not a rate-sensitive story. It is a volume story. Hyperscalers are not choosing between bonds and utility stocks. They are choosing between having enough power to run their AI infrastructure and not having enough. The demand is structural, not cyclical. It does not reverse if the Fed raises rates by 50 basis points.

The market is also anchored to historical utility multiples. A 20x to 28x P/E range feels elevated for a utility. But these companies are no longer growing at 2% to 4% annually. NextEra's revenue is projected to grow nearly 16% in 2026. Constellation is targeting 20% compounded EPS growth through 2029. Southern is guiding 5% to 7% with upside potential. These are growth rates that, in any other sector, would command substantially higher multiples.

The clearest sign that the market is still catching up: as of early 2026, NextEra and Constellation Energy had both reached valuations comparable to some Big Tech companies on an enterprise value basis. A Benzinga headline captured it perfectly: utilities hitting Big Tech valuations was described as "2026's weirdest trade." It is only weird if you have not looked at the power demand curve.

The Income Angle

For income-focused investors, the utility power thesis offers something rare: dividend coverage that is safer than at any point in the last decade, supported by a growth tailwind that should sustain payout increases for years.

NextEra yields approximately 2.5% with a payout ratio around 69%, leaving ample room for continued increases. Southern Company offers a higher yield in the range of 3.5%, backed by decades of consecutive dividend increases and now supplemented by the Vogtle revenue contribution hitting the rate base for the first time. Constellation is the lowest yielder at roughly 0.6%, but it is also the highest-growth name, with the EPS trajectory suggesting significant dividend growth potential as contracts convert to revenue.

The key for income investors is that the hyperscaler contracts are structurally different from traditional utility load. A 20-year PPA with Microsoft at a fixed price provides a revenue floor that residential customer growth never could. Dividend coverage ratios built on this kind of contract quality are fundamentally more reliable than coverage built on weather-dependent residential volumes and regulatory rate cases.

What Could Go Wrong

No investment thesis is complete without the bear case, and utilities have real risks that deserve transparent treatment.

Regulatory backlash on residential rates. As utilities invest tens of billions to serve data center customers, residential ratepayers may push back if they perceive their bills are rising to subsidize Big Tech. The White House Ratepayer Protection Pledge is designed to address this, but enforcement is voluntary and political pressure could complicate rate cases, especially in election years.

Interest rate risk. Utilities carry significant debt. Southern Company's debt-to-equity ratio is approximately 2.1x, a legacy of the Vogtle construction period. If the Fed is forced to raise rates aggressively to combat tariff-driven inflation, the cost of financing these massive capex plans rises, compressing the spread between authorized returns and borrowing costs.

Nuclear execution risk. Constellation's Three Mile Island restart is subject to NRC approval and could face delays. Southern's Vogtle experience is a cautionary tale about nuclear project timelines: the original budget was roughly $14 billion and the final cost exceeded $36 billion. Any new nuclear construction (SMRs, potential Vogtle Unit 5) carries execution risk until proven otherwise.

Data center demand may be overstated. Some analysts argue that utility forecasts of 90+ GW of additional data center load by 2030 are inflated by speculative interconnection requests. Grid Strategies has noted that actual demand may be closer to 65 GW, still transformative but meaningfully lower than the highest projections. If AI models become more efficient and require less compute per unit of intelligence, the demand curve could flatten.

Valuation overshoot. Constellation trades at a significant premium to the utility sector average. If the expected hyperscaler contracts do not materialize on schedule, or if EPS guidance continues to disappoint consensus, the stock could give back more of its 2025 gains. The 34% pullback from the October peak is a reminder that the market is willing to punish misses quickly.

The Thesis

The investment case rests on three pillars.

First, this is the first real volume growth cycle in a generation. U.S. electricity demand was flat for 20 years. It is now accelerating at a pace that multiple independent forecasters describe as unprecedented, driven by AI data center buildouts that will not reverse regardless of macroeconomic conditions. The companies that own generation capacity and grid infrastructure are the direct beneficiaries of this demand wave.

Second, hyperscaler demand is price-insensitive and credit-strong. Amazon, Google, Microsoft, and Meta are signing 15 to 20 year contracts with commitments to pay for power whether or not they use it. This is the highest credit quality customer base in utility history, providing revenue visibility and cash flow predictability that regulated residential tariffs cannot match. Over 80 GW of clean energy has been contracted by data centers to date.

Third, dividend coverage is safer than at any point in the last decade. The combination of rising rate base investment, hyperscaler contract revenues, and generation capacity that is already built (Vogtle, NextEra's 30 GW backlog, Constellation's nuclear fleet) means the cash flows supporting current dividends are backed by harder assets and longer-duration contracts than ever before.

NextEra offers scale and diversification: the largest infrastructure builder in the sector, with a clear path to monetizing the data center power wave through its hub strategy. Southern offers the nuclear moat: the only utility in America with proven experience building new reactors, positioned at the center of the nuclear renaissance. And Constellation offers the growth premium: the largest nuclear fleet, the most uncontracted capacity available for future deals, and the highest targeted EPS growth rate in the sector.

The risks are real. Rate cases, execution delays, and demand uncertainty are all legitimate concerns. But the balance of evidence, based on the financial data, the contract pipeline, and the structural demand projections, suggests that utilities are being repriced from bond proxies to infrastructure growth platforms. Investors who still view them through the old lens may be looking at one of the most asymmetric opportunities in the market.

At Wealth Engine Pro, the philosophy is straightforward: look at what the data says, not what the narrative has been. For 20 years, the narrative on utilities was "boring." The data now says something different. The companies that own the power that feeds the AI revolution are not boring. They are essential. And the market is only beginning to price that in.