The Electrical Paradox: Why Surging AI Power Demand Is Not a Gift to Electric Utilities

Summary

• The growth in AI data centre power demand has created a popular narrative that electric utilities are an obvious beneficiary. In early 2026 the market disagreed.
• Electric utilities face a capex trap: massive infrastructure investment is required, but regulatory constraints limit their ability to recover costs, compressing margins and weakening the risk-reward profile.
• At the time of writing, February 2026, the ImGeld industry ranking placed Electric Utilities in the bottom tier of all rated industries, with a score of 42.05 and a falling rank.
• Not all utilities are the same: gas utilities operate under different capex dynamics and regulatory frameworks, so sector-level generalisations are an analytical error.

The consensus narrative

At any investment conference in 2025 and 2026 the pitch was the same: artificial intelligence is an energy consumer on a new scale. Data centre power consumption is projected to keep rising sharply this decade. Hyperscalers are signing unprecedented power purchase agreements.

The conclusion seems obvious: more demand for electricity means more revenue for electric utilities, so own the sector. Except the market was not pricing it that way.

What the data showed

The ImGeld industry ranking scores industries on a composite of earnings estimate revisions and price action, with no narrative input (the method is described in how ImGeld ranks US industries). It measures whether companies in an industry are generating improving earnings expectations and whether price is confirming them.

In February 2026 Electric Utilities sat in the bottom tier with a score of 42.05, and the rank had fallen from the previous update. That was not a one-day move. It reflected a deterioration in the industry's fundamental profile: large capital needs to serve rising data centre demand, while regulators and customers raised affordability concerns that clouded the risk-reward profile.

In other words, the very catalyst the bulls pointed to, surging AI demand, was weighing on the industry.

The capex trap

Here is where the paradox crystallises. To serve rapidly growing power demand, electric utilities need to invest hundreds of billions in new generation capacity, grid upgrades, transmission lines and interconnection infrastructure. These are capital-intensive, multi-year projects with uncertain returns.

But electric utilities are regulated businesses. They cannot simply raise prices to fund investment. Every rate increase must be approved by state public utility commissions, which face growing pressure from consumers and politicians over affordability. The result is a squeeze: rising capex obligations on one side, constrained pricing power on the other.

The pattern resembles the telecom sector in the early 2000s. The internet boom created seemingly insatiable demand for bandwidth, and telecom operators poured billions into fibre networks. Many destroyed shareholder value in the process. The demand was real. The economics of serving it were brutal. Electric utilities face the same structural problem: the demand is real, and the ability to serve it profitably is in question.

The interest rate headwind

A second factor compresses the industry. Electric utilities are among the most capital-intensive and leveraged businesses in the market and rely heavily on debt to fund infrastructure. In a higher-for-longer rate environment their cost of capital rises, directly reducing returns on new projects. The mechanism is covered in why high interest rates affect the stock market.

Utilities have also traded historically as bond proxies: investors bought them for stable dividends when fixed-income yields were low. With government bonds offering competitive yields, the relative attraction of utility dividends has diminished, and capital has flowed from utility equities into bonds, creating persistent valuation pressure.

Not all utilities are equal: the gas utility distinction

This is where precision matters. While Electric Utilities sat near the bottom of the ranking, Gas Utilities are a fundamentally different industry with distinct dynamics. Gas utilities do not face the same capex cliff driven by data centre demand. Their infrastructure requirements are more predictable, and their regulatory environment operates under different cost structures and demand profiles. Individual gas utilities can carry very different industry ratings, and some can present sound fundamental profiles even while their electric counterparts deteriorate.

Grouping both under the label "utilities" is a fundamental analytical error. An investor who avoided all utilities because of the Electric Utilities' bottom-tier ranking would have excluded a different industry on the strength of a shared word. Separating them is what allows precise industry allocation rather than blunt sector exclusions, which is the same argument made in why 11 sectors are too blunt for rotation.

Who the demand actually benefits

When a quantitative ranking places an industry in the bottom tier and the rank is still falling, it reflects aggregate earnings trajectory and price momentum. Institutional capital is not ignoring the AI power story. It understands it, and has concluded that the economics do not work in favour of electric utility shareholders.

The industries positioned to capture the AI power build-out are the ones selling the infrastructure: turbine manufacturers, electrical equipment makers, grid technology providers and independent power producers with more flexible pricing. They capture the demand growth without being constrained by regulated rate structures.

What follows for industry allocation

• The demand narrative is not a sufficient reason for broad utility exposure. A bottom-tier ranking reflects fundamental weakness, and the capex-regulation squeeze is structural.
• Electric and gas utilities deserve independent analysis. They share a label, not an industry profile.
• Exposure to the AI power theme sits upstream, in equipment, grid technology and independent power, where pricing is not regulated.
• When earnings revisions, price momentum and capital flows all point the same way, narrative alone is a weak basis for a contrarian position. The market understands that more demand is good; when it disagrees, there is usually a reason.

Conclusion

The Electrical Paradox is a reminder that the obvious thesis is often the wrong trade. AI is driving unprecedented power demand, and electric utilities will be called upon to serve it. But the economics of serving it, massive capex, regulated pricing and rising financing costs, turned electric utilities into a bottom-tier industry rather than a growth story at the very moment the narrative was loudest. The ranking's placement was not a mistake. It was the data saying what the narrative obscured. Where the industry sits today is on the All US Industries Momentum Heat Map.

ImGeld publishes research, not investment advice.