Yves here. Our expert readers have sometimes criticized articles from OilPrice on electricity and nuclear topics due to them missing or misreading key production issues. This is mainly the result of authors, even ones who are industry analysts, being finance types first and engineering oriented at best second.
This article makes an important point which is more finance-y than engineering-y but engineering-oriented readers are likely to have important insight. But the authors talk around their key notion, which is that the cost estimates for electricity production increases are probably too low.
Pretty much everyone with an operating brain cell can see how electricity prices are rising in the US due to AI demand. That is projected to increase as one can similarly see from data center construction plans and reports of major AI players trying to secure future utility capacity. The authors posit 25% “inflation” increasing capital costs, but don’t unpack the possible drivers, let alone how they arrived at the 25% gueesstimate. General inflation? Price trends in key relevant commodities? High demand for relevant service providers giving them pricing power, as in too much demand chasing particularly construction capacity bidding up prices?
Of course, one can simply fall back on the idea that big projects alway run over budget.
By Leonard S. Hyman, an economist and financial analyst specializing in the energy sector who headed utility equity research at a major brokerage house and William I. Tilles, a senior industry advisor who headed utility equity research at two major brokerage houses and then became a portfolio manager investing in long/short global utility equities. Originally published at OilPrice
- Utility capital spending is set to surge far beyond official projections, driven by higher equipment costs and accelerating electricity demand from AI and electrification.
- Electricity prices are likely to rise materially, with revenue requirements increasing 7–9% annually.
- Financing risks and regulatory uncertainty are growing, as utilities rely heavily on debt and equity issuance to fund expansion.
Americans spend close to half a trillion dollars per year on electricity. The electric industry spends over a quarter trillion dollars a year on a capital program to replace old plants and add new ones to enable it to serve new demand. The industry and its customers have conflicting feelings about big capital spending. On the one hand, the spending should assure better service (good for consumers) and increase the rate base (good for utility investors). On the other hand, the spending will surely raise prices because the new plant and equipment will cost so much more than the old plant (30-40 years old on average) that it replaces or supplements. Until recently, utilities could get away with spending without much fuss because lower fuel prices offset some of the capital cost increases and because the regulators were out to lunch. Now, however, prices have begun to rise, aggravated by AI demand in many areas, which is not good news because energy affordability has now become a political issue.
The Edison Electric Institute (EEI), which represents investor-owned utilities serving the bulk of American consumers, issues five-year capital expenditure (CAPEX) projections annually. Over time, those projections have shown two tendencies: spending growth flattens or declines in the third and fourth years, and CAPEX comes in over the projections. Why? Many companies may provide sketchy or no information about new capital needs four to five years out because they don’t have a handle on them. They may also underestimate inflation or construction costs over the longer term. Today, they might have an additional reason to play down capital spending plans for years four and five: they might not want to face regulators to explain in advance the big numbers. (Regulators are the same kind of people who oppose spending to fix the century-old sewer pipe until it ruptures.) In addition, they may think that they have escaped climate and environmental remediation spending because the Trump administration won’t require it, and they won’t have to replace old coal-fired generating plants either. (The Trump administration, though, will end in year four of the five-year program).
Let’s look at the latest numbers with those thoughts in mind. Those numbers matter because they will affect prices and financing requirements. Consumers, regulators, and investors need dependable numbers. In the five years ended 2024, the EEI estimates that utility industry capital expenditures (including for natural gas, which makes up about 15-20% of the business) was $762 billion. The EEI projects the 2025-2029 capex number at $1,150 billion, a 51% increase. Seems reasonable, doesn’t it? Not if you consider that prices of new plant and equipment will probably be 25% higher in the second period, and that growth of demand for electricity may triple or quadruple in the second period over the first period.
Perhaps we should tackle the numbers differently. Basically, utility capital expenditures consist of two parts: replacement of retired plant and service for incremental demand. If the company keeps to the depreciation schedule, it replaces perhaps 2.5% of the plant each year. So, in 2020-2024, when growth was close to 0.5% per year, the industry’s capital expenditures bought and installed new plants equal to about 3% of the existing plant in each year. In 2025-2029, when growth in demand may reach 2% per year, the industry needs to buy a new plant each year equal to about 4.5% of the existing plant. That analysis points to a 50% increase in spending in real terms and, if 25% for inflation is added on, an 88% increase over the 2020-2024 spending. That would bring the 2025-2029 CAPEX to $1,432 billion. Now, add another component to the calculation. The percentages for 2025-2029 should be applied to the higher base year (2024 plant being 26% higher than the 2019 year end plan). That would raise the spending projection to $2,367 billion.
Consumers will end up paying for additional capital as well as for increases in non-capital costs like higher natural gas prices. Using CAPEX projections near the low end of the range, prevailing returns and depreciation rates, and assuming that non-capital costs will rise by 4% per year (2% for inflation and 2% for volume growth), we estimate that revenue requirements would have to rise 7-9% per year to pay the additional costs. If KWH sales volume grows 2% per year, that means that the price per unit rises 5-7% annually. At an inflation rate of 2%, the real price increases 3-5% per year for the energy output sold by EEI’s members. That might not seem like a big number, but median-income families have barely kept up with inflation in the recent past, so energy pricing looks like an affordability issue, which could pressure regulators to lower returns in order to keep prices down
Financing became an issue in the past, when CAPEX got out of control, bond quality fell, and electricity suppliers had to sell new stock offerings at low prices. If rate base grows 10% per year, as it might over the coming five years, while sales grow only 2% a year, where does the money come from to pay for the expansion? That may not be a big issue. If the industry spends $290 billion per year, its internal funds from depreciation and deferred taxes would furnish about $90 billion per year. Shareholders and creditors would have to fork over the balance. If management wants to raise 60% of new capital from creditors, it will borrow $120 billion. As for the shareholders, retained earnings might run around $25 billion per year, meaning the company would have to sell $55 billion a year of new stock. “Wait, wait”, you say, new stock sales averaged only $11 billion per year in the past five years, and selling all that stock will be dilutive. and crash the industry’s stock prices.” Actually, not so. The new stock offering raises book equity (and earning power) by about 8%, but because the stock price is roughly 170% of book value, it raises the number of shares outstanding by only 5%, so the new offering will, eventually, raise earnings per share by 3%.
Adding it all up for shareholders, investment of retained earnings could raise annual earnings by 3-4%, new stock sold above book value adds another 3%, to produce 6-7% earnings per share growth and current dividends will add another 3-4% to shareholder return, to produce a prospective total return of around 10%, not bad for a supposedly low risk investment, considering that 10 year Treasuries yield only 4%. If the industry stuck to its projected program, however, we calculate that total return could fall to 7%.
Now to the elephant in the room. As spending rises, electricity more and more becomes a financing business that depends on government decisions and regulations. First, there are the many Trump administration decisions on fuel choice and environmental protection that could easily be overturned in the future, exposing the industry to additional expenditures and the risk of stranded assets. Second, there is the increasing likelihood that state and federal regulators will react to complaints about the affordability of electricity and natural gas supplies by curtailing expenditures and lowering allowed returns, more worried about their reelection in November than long-term electricity supplies.
To sum up, we don’t expect the utility industry to stick with its unrealistically low CAPEX projection, but we doubt that, within the near future, it will spend at a level that will produce a modern, reliable electricity system suited for the digital age (assuming that the equipment manufacturers can furnish the equipment needed on a timely basis). As for you, consumers out there, electricity prices will rise. The only question is: how much?
21 comments
Comments are closed.
I don’t think that a lot of people are taking into account that U. S. shale is about to start producing less.
A lot of the LNG the Europeans need comes from those shale fields and while they may need less LNG in the future because of deindustrialization and economic incompetence.
This will still very likely result in higher natural gas prices in the U. S. till they deprioritize the Europeans because of elections.
Of course natural gas prices are somewhat correlated with electricity but I don’t know how strong that correlation is in the United States.
https://energynow.com/2025/08/sliding-us-rig-count-outpaces-efficiency-gains-threatening-onshore-oil-output/?amp
https://www.reuters.com/business/energy/us-share-europes-lng-imports-increased-60-january-2026-01-30/
https://www.consilium.europa.eu/en/infographics/where-does-the-eu-s-gas-come-from/#:~:text=Imports%20from%20Russia%20declined%20from,58%25%20of%20total%20LNG%20imports.
I think you meant to write “a lot of the LNG that the USA is forcing Europe to buy cones from those shale fields”. :-)
Which suggests that when they run out, Europe will need to find another supplier like, say, Russia. Ironically, so may the USA!
European vassals will escape one way or another, either making rapprochement with non-US LNG or pipeline gas suppliers and/or diversifying their supplies into renewables (my money is on gas still being relevant though…).
I have a spreadsheet with each electric bill since January of 2022. I track all the charges. I’m on Ohio AEP. There are 4 things we pay for on the bill. The electric usage based on rate per kilowatt used. Transmission charges, distribution charges, and customer charge, which is always $10.00.
Kilowatts vary by usage, the transmission and distribution charges are proportional to the kilowatts used. We can shop for better rates through alternative suppers, which may be cheaper for kilowatts used, but the rest of the chargers still apply.
If I take my total bill amount and divide by Kw I used, it cost me $.1553 per KwH in Jan 22, to $.2289 per KwH in Jan of this year. Highest was .2306 a month ago. So that’s around a 50 percent increase in 4 years. And we are only getting started.
Can you show how transmission & distribution charges changed in the same period? (in per kWh terms). This can vary a lot with the region and the investments that have been made in each. For what I have read these were the main cause driving up electricity prices in California. And these guys who are running a natural monopoly tend to charge somehow in excess for the investments, for what I have seen in Spain.
Yes, I have it all broken down. I calculate by how much the charges are to how many kilowatts I use.
4 years ago to this exact month, transmission charges were charged at a rate of .0293 on the dollar per my Kw usage. Distribution charges were .0430. They are now .0355 and .0617.
Percentages close with all, but they do fluctuate. It would be interesting to make a chart to see the volatility, but I feel I have already shown I’m nuts doing all this anyway :-)
I actually read my electric meter every night for a year. Just to check…
Transmission was not big deal but distribution raised at more or less the same pace as the total bill (about 45% in the 4 year period). Whether such inflation in distribution had to do with general improvements of the distribution lines or was somehow fuelled by data centres in different ways (via inflation of components for instance, or distributing distribution costs of data centres between all users) is something difficult to find out.
Yes, I agree.
“altermative suppliers” is a full-blown scam. unless OH state regulation is drasticly different from the national state-level norm. the vast majority of people are best off staying with the state-regulated (a) power supplier and (b) state-regulated delivery provider
the only legit option is “real-time pricing”, if offered by AEP. ****but**** real-time pricing is cheaper only if you have a specific pattern of flexible electricity consumption and do not have inelastic peak-summer demands, like high wattage, always-on medical devices.
read the fine print, caveat emptor
I have done it for several years. It has saved me money, and you can even get a better deal if you go gas and electric with the same supplier. You only save on the energy product, not the fees. But its not a rip off.
If I don’t choose a market supplier I default back to AEP of Ohio. They are more expensive, always. Changing is easy to do. Once chosen, shows up on the monthly bill.
Just a few points.
First off, AI/data centres are not the primary reasons for rising electricity demand outside of specific regions – the IEA estimates that between 2024 and 2030 it represents just 10% of rising demand. Overwhelmingly, increasing demand is coming from general economic growth and the electrification of industry and transport (a good thing, as this reduces overall energy demand significantly). In the US, demand has been more or less stagnant for a couple of decades. It seems now to be increasing somewhat due to industrial/domestic electrification along with data centres. When analysing an issue like this, its important not to confuse national/international level impacts with regional grid pressures.
Secondly, what is currently happening is that the unexpected level of increase within some grids is causing rising prices partly due to an increased reliance on those sources of power with higher marginal costs (this happens when overall capacity is above what is anticipated) along with a requirement for much more infrastructure investment, which will inevitably be pushed onto the consumer. Which ‘consumer’ this is (as residential consumers will invariably have different charging structures than large scale industrial users) is a matter of regulation.
A further point is that while increasing infrastructure investment does require additional inputs, but there have been some very significant technical breakthroughs over the past decade or more in grid efficiency – in the long term, this is important and should mitigate costs, although the increased reliance on capital inputs will take time to work through the system.
One the point of gas pricing – in the US and elsewhere, fossil gas has grown very rapidly due to the rise of fracking and LNG exports – but this has primarily been at the cost of coal, the big loser in recent years. To many peoples surprise (including myself), the costs of fracked gas has not increased as much as expected – there are claims of some quite spectacular productivity gains in fracking recently, although its hard to separate the reality from marketing. But it does seem that the US frack industry has become exceptionally good as squeezing out gas with less and less capital inputs. However, there is only so far they can go with this before geology wins.
I don’t know enough about the topic to go deep into the weeds on the article, but what has been clear to anyone following electricity policy and economics over the past few decades is that we are going through a period of enormous and unpredictable change, both in terms of technological inputs and pricing. There are solid grounds for believing that the spectacular drop in solar/battery costs are a genuine game changer, and not just in the very long term. Much depends on demand rise, and AI/data centres are not the main drivers of this internationally (even assuming the AI bubble doesn’t pop).
I just heard today that China is looking at sodium ion batteries. If successful it would help with the electrification of cars in colder climates.
https://m.youtube.com/watch?v=-gBmoYYliWE
But yeah, if there is a technological revolution coming I don’t think anyone would know what to expect.
I’ve also heard some people like Doomberg say that we have effectively infinite oil because of technological progress. Personally I’m very sceptical, but that could be because I did not pay the subscription to hear his whole argument.
Sorry to go ad hominem, but that YT channel is terrible at tech, he just regurgitates press releases. Just have a Think is a good channel on upcoming tech, including sodium batteries. Sodium ion batteries are well on their way to commercialisation, but so far have specialist uses due to having somewhat lower power densities than lithium ion. Potentially, they will be much cheaper and safer than lithium ion, but not quite yet. The potential very low cost makes them very promising for shorter term storage.
There is more than enough oil in the ground to completely doom the planet. The industry has been much better than sceptics expected at squeezing out more and more oil and gas from existing wells – this is the primary reason why peak oil didn’t happen when many of us thought it would. But arguably this will just make the fall-off more dramatic (possibly).
Here is but just a snap shot of sodium ion.
It is already in commercial production by CATL as well as a few others. It is almost at the same energy/weight of LFP. But coming in at well under 1/2 the cost and expected drop of 50% more in a year -2. It doesn’t need heating or cooling so for BESS and cars you can eliminate the HVAC and those losses. Operates at extreme temps without issues and is double the cycle life of LFP at about 20,000 cycles.
And almost zero fire risk.
The link below is one of lots that goes into more detail.
https://cleantechnica.com/2026/01/23/catl-begins-commercial-production-of-sodium-ion-batteries/amp/
Just to add to what you wrote, sodium ion batteries are already available in consumer products today. You can buy products that use them such as this portable power station . You can also buy bare sodium ion cells in sizes like 18650s on Chinese e-commerce sites. The main thing is getting the cost down for it below the cost of lithium iron phosphate batteries, but that is the kind of thing that China excels at (along with, as noted, the general material favourability in terms of costs for the technology).
Your first point does not ring true to me, as an inhabitant of the u.s.:
“Overwhelmingly, increasing demand is coming from general economic growth and the electrification of industry and transport…”
I know Trump has told us about how much economic growth the u.s. has enjoyed. I am not sure what ‘industry’ has been growing other than AI — maybe “financial services”, inferred rent to home owners, actual rents to Corporate and Private Equity home and apartment building owners, costs for Medical Care … Corporate Monopoly rents? Who are the big Corporate electricity users with growing use? I am not sure how much electrification transport has enjoyed. I suppose it might have enjoyed more electrification if Chinese electric vehicles had been admitted to the u.s. market.
I referenced the EIA source you linked to and located the following quote to support your statement:
“Despite the strong increase, data centre electricity demand growth accounts for less than 10% of global electricity demand growth between 2024 and 2030 in the Base Case.” That is fine for the world as a whole, but does not seem right for conditions in the u.s. — which this post appears to be referring to. My congratulations to the world extend to the closing paragraph of your comment, but the u.s. history regarding solar energy seems to me a sad story at best. I still remember Reagan pulling down the solar water heating panels Carter put up on the White House.
Your further point “there have been some very significant technical breakthroughs over the past decade or more in grid efficiency” is very interesting to me, in and of itself. I am ignorant of those breakthroughs and would very much like to learn more about them. I am not too confident those breakthroughs will find their way into the u.s. grid, but any of them applicable to small scale local grids are of great interest to me.
Your one point more discusses gas pricing and the fracking industry, and coal pricing. That is well beyond anything I feel comfortable speculating about.
Your second point loses some of its weight if the “unexpected level of increase within some grids” is primarily the result of AI demands for electricity. I sincerely hope “the spectacular drop in solar/battery costs are a genuine game chang
There is no reason a private transmission owner overloard of a toll road would want to increase capacity for renewables or new sources without public bribery.
I disagree the AI, data centers, bitcoin and chip fabs are NOT driving marginal electricity prices.
There’s nothing hidden about the nearly 50% increase in my electricity bill from PG&E, or PG&E’s support for Mrs Newsome’s Charity.
In some good news my county council is apparently poised to reject a tax break deal and therefore the local data center that went with it. We are already a growing area so why take risks supporting what increasingly appears to be speculative projects that may not be around that long?
I recall but cannot readily fetch links to, some past posts and links here that have discussed the exceptional demands the AI industry has made and projects to make upon the u.s. electric grid. Those exceptional demands, added to the projected costs for building additional generation capability for AI will greatly impact the cost for electricity in the u.s. adding to the monopoly rents due to the u.s.’s privatized electricity Corporations, and the additional demands for heat and cooling in response to the increasingly chaotic weather. As I recall some past posts and links here have described the great uncertainty underlying the future of the u.s. AI build-up, and noted how that uncertainty added to the risk and thereby the cost of the investment money electric utilities needed to build any expansion they made to accommodate AI. This cost further adds to the increase in the price for electricity. I believe past posts and links here have described how the uncertainty in the future of u.s. AI build-up also adds to the potential for building future electric generation capacity which may oversupply the projected future demand if AI crashes, as many commentators here and in the bloggosphere project. I suspect that risk is at best poorly captured in the risk assessed on investment in additional generating capacity to meet the projected AI demand.
Bottom line for me, I am looking at electricity bills much much higher than I had last month. If the doubling of this month over last month is any indication of the future cost to live my much much much less than opulent life I will remain considerably less than a happy camper.
Though cost-plus contracts are increasingly frowned upon (due to inefficiencies), I would not be surprise to find out that they may be continuing to play a major role in the energy sector, including electricity generation and transmission.
Does any one have any idea about this possibility?
Large-scale energy infrastructure is often described as “unaffordable.” But what makes it unaffordable is not steel, labour, or turbines. It is interest.
When projects are financed through private capital markets, they must deliver a required rate of return. Debt carries interest. Equity demands dividends. Risk premiums are embedded in pricing. Over decades, those financing costs can rival the underlying construction cost. A transmission line funded privately is not just copper and towers, it is 30 years of coupon payments.
That financing structure inflates cost projections and causes projects to stall when rates rise. “Unaffordability” is frequently the result of embedded interest obligations, not a shortage of real resources.
There is, however, an alternative: fund public infrastructure with public money rather than private credit.
The federal government is not a private borrower. It is the issuer of the currency in which these projects are denominated. It does not need to obtain dollars before it spends. When Congress authorizes spending, the Treasury and Federal Reserve credit bank accounts. The dollars are created in the act of spending.
Under current practice, deficits are followed by Treasury issuance and interest payments to bondholders. But that is an institutional choice, not an operational necessity. A sovereign currency issuer could fund infrastructure directly without embedding decades of private interest costs. Interest payments are a policy decision – a transfer of income to holders of government securities – not a funding requirement.
Once this is understood, the debate changes.
The relevant constraint is not “How will we pay for it?” It is: do we have the real resources? Skilled labour, transformers, copper, steel, manufacturing capacity, logistical coordination – these determine whether infrastructure can be built without triggering inflation. If those resources are idle, public spending mobilizes them. If they are fully employed, additional demand will raise prices. Inflation not insolvency is the binding constraint.
The deeper issue is political economy. In American capitalism, the needs of private capital routinely trump the needs of the real economy. Projects are evaluated not first on physical feasibility or social necessity but on whether they satisfy required returns to investors. When interest rates rise, socially necessary investments are delayed, not because the country lacks labour or materials, but because the financial yield is insufficient.
That inversion – subordinating the real economy to the financial sector – is what creates artificial scarcity.
None of this implies unlimited spending. If supply chains are tight or critical inputs must be imported, inflationary pressures can emerge. Real resource limits are real. But they are not financial limits.
What makes projects look unaffordable is the decision to route them through private capital markets and embed long-term interest streams. Use public money instead, and the question becomes the only one that matters: does the United States have, or can it build, the real productive capacity required?