Will AI End the Utility Monopoly?

Leonard Hyman & William Tilles

Leonard S. Hyman is an economist and financial analyst specializing in the energy sector. He headed utility equity research at a major brokerage house and…

More Info

Premium Content

• The massive financial resources and cost-of-capital advantages of AI-focused tech giants enable them to build and operate power plants independently.
• Utilities face significant challenges as tech companies leverage their superior credit ratings and financial strength to finance power projects.
• The rise of independent power consumers could end up destabilizing the utility-regulatory model.

AI data centers, due to their insatiable demand for electricity and more importantly, due to the vast financial resources of their owners, threaten to upend the operations, regulation and financial basis of the electric utility networks. That is not hyperbole.  A new administration that distrusts regulation and is heavily influenced by a Silicon Valley that advocates breaking things, may speed the process.

Let’s step back. Five years ago we wrote a paper which said that electricity sales growth would resume, that the electricity industry was not prepared, and that the incumbent electricity industry might not benefit from the rise in demand — somebody else might.

Next, back to basics. Regulation of public utilities involves an implied two step process: 1) the city or state grants the utility a monopoly or franchise and 2) in return the utility corporation agrees to service levels, capital controls, and other measures to prevent customer abuse. The prospective AI power plant developer, as a free agent has no legal obligation to the utility or the utility’s customers. And we think it is almost inevitable that the data centers and utilities will end up in a big legal squabble. We expect the  AI folks, as large power generators, will try to shift costs over to the utility and its customers, in what we call cross-subsidization. The AI-led power plants represent an enormous incremental base load addition that has to be physically accommodated up and down the transmission and distribution system. Someone will have to pay for all these utility upgrades. Should small residential and commercial customers pay higher utility bills simply because an AI facility is located in their service area and requires extremely expensive system upgrades?

Related: Indian Oil Probes Allegations of Albemarle Bribes to State Firm’s Officials

The utility-regulatory model is a scarcity model (which presumes investment capital is scarce or expensive). Regulators allocate resources and expenses among competing customer classes—all of whom want cheap electricity. The AI/data center trend upends these traditional relationships between government, corporations, and customers for several reasons: 1) their margins are high enough that, unlike typical industrial customers, they are relatively price insensitive, 2) they have the size and sophistication to own and operate multiple generating stations, 3) and they have large balance sheets, with cost of capital advantages, and virtually unlimited access to capital. And it is this last point gives them an enormous competitive advantage. Tech giants can finance their new generating plants, assuming 100% debt financing, for less than half of what it would cost an investor owned utility. Competitive disadvantages of that magnitude are literally insurmountable.

The disparities are mind-boggling. The top five AI players have sales of $1.76 trillion, net profit of $0.384 trillion, cash balance of $ 0.433 trillion, and free cash flow of $0.345 trillion. That is three times the sales, six times the profit and twenty times the cash balance of the entire electric industry. (No point making a free cash flow comparison because the electric industry does not generate any free cash flow.

Assuming a new power plant costs $3 billion, about 20% of total construction expense is typically dedicated to capital costs—assuming a utility’s cost structure or $600 million. Financing at half the utility’s cost of capital would save $300 million in prospective financing costs. Utilities are at an enormous cost of capital disadvantage here relative to large technology companies like Microsoft which, for example, could finance a large power generating station with 100% debt at a cost of 5% (their 25 year corporate bonds due 2050 yield 5.1%). And from a credit quality perspective, their balance sheet is pristine and their corporate bonds are rated AAA, the highest possible rating. Utilities typically carry a far weaker BBB fixed income credit rating, though still solidly investment grade. The question then becomes, why don’t the technology companies finance these power assets on their own balance sheet for literally less than one half the cost? In order to believe in a perpetuation of the utility finance status quo here, one has to believe that capital cost differentials of the magnitude indicated here will make no difference to large prospective power purchasers, which we frankly find hard to believe.

And if these power plants do get built by technology companies, and not utilities, what happens to utility growth prospects? Hard to construct this as a positive. A related question is whether this triggers a new wave of mergers and acquisitions with technology companies actually purchasing entire utilities. Maybe there are interesting synergies to be realized on both sides of the meter. It is also possible that the data centers end up overbuilding power generating plants perhaps hoping their new plants find lucrative power markets even if that means competing with local utilities. This could be another source of financial exposure for regulated utilities. Having a large parallel network of generating stations owned and operated by sophisticated technology corporations has to be viewed as some type of competitive threat.

In a funny way this continues a long term industry trend. The first power plant builders to break away from the regulated utility model under President Jimmy Carter were called IPPs, independent power producers. They were independent of the obligation to serve a particular service area, like most other utilities of the time. Now with data centers poised to procure power generating plants independently of the local utility, the utility industry now faces the prospect of “independent power consumers”, who will also produce for self consumption. It’s not like they won’t have a relationship with the local utility. They’ll still be part of the grid. But the potentially most lucrative part of this utility-data center relationship, the ownership and operation of the power generating station, is likely to be lost to the incumbent utility for the reasons cited above. The utility industry has never been in this position before. They have always been the elephant or the 800 pound gorilla in the room so to speak, dictating policy to regulators and politicians alike. Now much bigger and more formidable elephants and gorillas from the world of technology—companies thirty times their size—are in the room as well. Microsoft for example has a market capitalization of $3.3 trillion vs Southern Company-a very large utility– with a market cap of only about $91 billion. Today’s US electric utilities are puny compared to large technology firms, with far more leverage, and that has placed them at a significant competitive disadvantage where it matters most, relative borrowing costs. The best the US utility industry can hope for is that they get ignored by the technology corporations, the way a shark ignores a sardine while hoping for bigger prey.

If our present utility industry survives more or less in its present form it is likely to be because Microsoft, Apple, Meta, or whomever is currently scrambling to buy power plants chooses to ignore the regulated utility’s other non-generating assets. And we think this is unlikely. The technology industry has developed all manner of intrusive consumer products under the rubric of “smart” devices. Do we really think they wouldn’t continue to tinker at least with the thermostat, a device that could be in almost every room in a home or office? The near term competitive threat to utility growth, which will likely shake the markets, has to do with who actually builds and profits from new power generating assets. But, longer term, we don’t think the regulatory process itself can mediate between technology companies and utilities—there is no bright line here— once it is no longer clear where the public’s interest lies. As long as electricity prices remain affordable, the PSC’s will likely lapse into the position of referees and just let the big dogs fight. Then if power prices get out of hand, it’ll may be too late for the PSCs to do much.

By Leonard Hyman and William Tilles for Oilprice.com

More Top Reads From Oilprice.com

• Fusion Energy Breakthrough: Virginia to Host Landmark Power Plant
• Will Trump Policies Slow Methane Emissions Cuts in the Permian?
• Russia Expands Global Nuclear Footprint Despite Western Pushback

Download The Free Oilprice App Today