Electricity costs have jumped significantly in the past few years. In fact, the consumer price index for electricity has increased by 45 percent since May 2020,1 15 percentage points higher than overall inflation.2 In that timeframe we have seen two oil and gas crises: one in 2022 following Russia’s invasion of Ukraine, and one beginning in February 2026 with the Israel and US war with Iran. In both cases, skyrocketing fuel costs have spiked electricity prices by increasing the cost of gas-generated electricity, demonstrating how the grid is closely tied to the volatility of fossil fuels and geopolitics so long as electricity generation relies on fossil fuels.
At the same time, the recent explosion in AI data centers is placing extreme demands on the existing electric grid and generation capacity, which has already driven up electricity bills. But the looming and less discussed cost of the AI boom will come in the next 5 to 40 years. Utilities are making plans today to invest $1.4 trillion on capital expenditure (capex) in the next five years—that is $300 billion more than total capital expenditure over the past five years.3 Because investor-owned utilities recuperate the costs of capital expenditure plus a return of equity—i.e., profit—by increasing rates for their customers, in the absence of urgent moratoriums or rate reforms, residential utility customers will pay higher electricity bills to finance these investments and utilities’ profits.
A recent report reviewed earnings calls, press releases, and 10-K financial performance reports to determine the motivation for these investments and found that $1.16 trillion are designated to expand grid and generation capacity to meet AI data center demand and/or extreme weather mitigation.4 Both investment priorities are the result of greed-driven corporate decision-making and expose utilities’ climate hypocrisy. On the one hand, AI expansion has driven additional fossil generation to support the energy demand of data centers, increasing emissions in data center hubs. AI tools themselves are also making the fossil fuel sector more productive at all steps of the production process. A recent report found that AI could help drive a 1.2–4.8 percent increase in global emissions from its contribution as a “productivity enhancer” in the fossil fuel sector.5 Capital spending that utilities undertake to support AI is implicated in furthering the climate crisis. On the other hand, utilities are also making capital investments to profit off their own climate accelerationism by charging ratepayers to make long-overdue repairs and upgrades to improve extreme weather resilience, post-disaster. Utility customers end up footing the bill for both the problem and the “solution,” while utilities profit.
Assuming a basic investor-owned utility cost recovery function,6$1.4 trillion in total utility capital expenditure means that households on average could spend an additional $376 per year for electricity by 2030. Over the full 40-year lifetime of these capital investments, residential customers can expect to pay a total of $10,000 on average. For middle-income households, this constitutes a 20.8 percent increase in electricity spending above 2024 averages.
These estimates do not account for state-specific rate regulations, the presence of moratoriums, or utility-specific discrepancies in investment, which will undoubtedly lead to disproportionate household burden across the country. However, they do demonstrate the potential scale of average disruption (e.g., from AI expansion and growing climate disasters) to households in the absence of reform. Moreover, they raise important questions about how infrastructure decisions are made and who they ultimately benefit. We desperately need grid investments to expand renewable generation, increase battery capacity, and add long-range high-voltage transmission. These infrastructure improvements are necessary to decarbonize electricity and break its connection to volatile fossil fuel markets, but also to deliver cheaper, stable electricity that households can afford and plan around.
Utilities’ capital expenditure plans should be carefully reviewed and questioned but are unfortunately usually created behind closed doors. These plans ultimately set the priorities for the future of the grid, and thus for affordability and decarbonization. Their prevailing approach is to prioritize quick returns and react to climate damages after the fact. AI demand has made the effort to prioritize proactive grid and renewable expansion even harder by displacing the political momentum behind decarbonization-aligned utility investments and occupying financing space that could otherwise contribute to reducing fossil fuel reliance and reducing costs to consumers.7 The trade-off between serving Big Tech–driven load demand versus serving proactive decarbonization goals is also a matter of energy affordability. Tellingly, none of the utilities reviewed listed meeting decarbonization goals as an investment priority. When investor-owned utilities continuously fail to make these necessary investments, and demonstrate how easily they can be bought by Big Tech, we are forced to consider alternative models for grid investment and planning.
At minimum, public utility commissions should (1) block rate hikes that support AI development at the cost of residential bill affordability and (2) meaningfully intervene in capital spending plans before they are set so that important overdue grid expansion and greening can take priority. Without strengthening public oversight and regulatory muscle, investor-owned utilities will not take on the renewable transition or affordability at the expense of their profits. Larger interventions, like establishing a federal investment body to take on grid expansion and renewable penetration, are likely needed to fill the gap private utilities persistently fail to address.8
The costs of climate disaster and AI expansion manifest in utility capital expenditure
Most reporting on the impacts of extreme weather and AI expansion have been state- or utility-specific. Importantly, these reports demonstrate how customers in AI hotspot states or in regions with frequent wildfires are experiencing record-breaking electricity costs. Missing from the discussion is an understanding of how increases to utility capital investments will raise average household expenditure on electricity in the coming decades, and what that will mean for the overall energy burden that households—especially lower-income households—could face. National average estimates can be compared with the Bureau of Labor Statistics Consumer Expenditure Survey, which tracks household spending on all categories reported in the consumer price index (CPI)—a key measure of inflation—by income quintile.
To develop an average household cost burden estimate, we make several assumptions that allow us to abstract from the hyper-specificity that characterizes utility regulation and cost allocation across the country.9 Whereas in reality, capex costs are passed onto ratepayers through individual rate cases on a utility, proposal specific basis, we assume, for the purposes of simplicity, that there is one utility in this model that plans to invest $1.4 trillion over the next five years and services the entire residential customer base across the country. This model utility employs a rate-of-return cost recovery formula—the most common among regulated investor-owned utilities—to cover its capital expenditure costs and its regulator-approved profit rate. A rate-of-return model allows utilities to recuperate the full cost of their investment and charge a rate of return on any undepreciated capital that the utility self-financed over the lifetime of the capital investment. We assume the rate of return is equal to the national average rate of return for utilities—9.7 percent in a year (and we assume the lifetime of these investments to be 40 years)—typical for utility infrastructure.
We assume that, in line with historical trends, residential customers will shoulder approximately 40 percent of the total cost of these investments, though this proportion is highly policy-dependent and thus future projections of this proportion are contested. The EIA reported that in 2025 residential customers accounted for 37.3 percent of total electricity sales,10 while other estimates range closer to 50 percent.11 We use 40 percent as a conservative baseline. Efforts to reduce residential customers’ burden could impact the average cost estimate if implemented at scale. Taking these variables as given, we estimate that the capital expenditure–driven cost burden will peak at $376 per year in 2030 when all five years of capex investment has been deployed, and reduce steadily over time. However, the cumulative costs will rise to $10,000 over the 40-year lifetime of planned investments.
Utilities are increasing the cost of electricity for residential customers due to capital expenditures. By 2030, each customer could spend, on average, an additional $376 per year on electricity, or $10,000 cumulatively by 2069.
Source: Climate and Community Institute, using data from PowerLines and the Energy Information Administration (EIA).12
For households in the lowest-income quintile, with incomes up to $30,000 per year, this represents a 29.5 percent increase by 2030 in their electricity expenditure compared to 2024. For households with annual incomes between $30,000 and $57,500, it represents a 22.5 percent increase. Finally, for households with annual incomes between $57,500 and $94,500, it represents a 20.8 percent increase. When 42 percent of US residents report having no emergency savings, a nearly $400 unexpected yearly increase to a household necessity like electricity can be debilitating.13 Insofar as this additional utility burden affects low-income households who often rely on rate assistance or bill assistance programs like California Alternate Rates for Energy (CARE) in California or the Low Income Home Energy Assistance Program (LIHEAP) nationally, these costs could also transfer onto state and federal program budgets that are already oversubscribed and underfunded.14
Planned capital expenditures will increase electricity bills by $376 on average in 2030, equivalent to a 29.5% increase for households making below $30,000 per year.
Source: Climate and Community Institute, using data from Bureau of Labor Statistics (2024).15
An approximation, not a prediction
For several reasons, this average cost burden estimate is not set in stone. On the one hand, this estimate reflects a "business-as-usual" scenario; federal, state, or municipal policies that either halt AI data center builds or require AI companies to pay for their own generation of grid upgrades are not reflected. If policy efforts are successful in shielding residential, and especially low-income, customers from shouldering the burden of data center–driven infrastructure, average cost burdens could be reduced. For example, if we reduce the residential customer burden assumption to 20 percent, the average additional cost burden by 2030 would drop to $188.
On the other hand, utility capital plans are themselves evolving. The PowerLines survey used to formulate this estimate notes that the $1.4 trillion in planned expenditure reflects only the plans utilities had shared at the time of their survey; some utilities they reviewed only proposed two- or three-year spending plans, and their numbers are likely to be revised upward. With both AI demand and the climate crisis showing no signs of slowing, compounding additions to these capital plans are likely.
As is the case with variable energy costs, these capital plans are also geographically clustered such that the average impact of the total planned expenditure will not reflect the higher and lower ends of the distribution. Some households will see very little impact from AI demand while others, in AI hotspots, will pay much more than $376 in one year.
Importantly, this narrow focus on the transferred costs of capital does not reflect the total burden customers are currently experiencing. This estimate does not take into account how additional AI energy demand is bidding up the per MWh cost of electricity in wholesale markets—the variable costs of electricity—in AI hotspots. This estimate only takes into account the fixed costs and assumes one form of cost recovery: rate of return. In reality, capital costs are passed to consumers in different ways depending on region and market. In some restructured regions of the United States, capital costs are not incorporated into bills purely through the capital expenditures of integrated utilities, but instead partially through a capacity market. In a capacity market, all of the capacity that bids in and is required to meet peak demand is paid the cost of the marginal unit of capacity, meaning that the amount consumers pay for existing capacity also increases when new, more expensive capacity is added to the system. In just one year in the Pennsylvania–New Jersey–Maryland (PJM) region, from the 2024–2025 service year to the 2025–2026 service year, the capacity market clearing price per MW-day jumped by 833 percent, from $29 to $270.16 The rate-of-return assumption makes the fixed-cost-to-customer estimate a likely underestimate as capital costs passed through markets typically lead to higher costs to customers.
Capital expenditure should support decarbonization, household affordability, and grid reliability
Projected electric utility capital expenditure has exploded in the past 10 years, with large increases due to increased natural disaster repair needs and growing AI demand. From 2026 to 2030, capex spending is projected to hit $1.4 trillion, meaning an average of $280 billion per year over the next five years.17 Utilities’ prior capital expenditure ranged from $104 billion in 2015 to $208 billion in 2025.18
Utilities’ projected capital expenditure in the next five years has exploded compared to recent years.
Source: Climate and Community Institute using data from Edison Electric Institute and PowerLines.19
Of the 51 utilities surveyed in the PowerLines report, 32 identified data center demand and load growth as a key driver, 28 identified system resiliency and extreme weather mitigation, and 16 identified replacing aging infrastructure. S&P Global analysis reports that data centers and other large industrial customers are “fueling the need for new power supplies through 2035, adding 374 TWh of energy demand and over 45 GW of peak load.”20 Some major utilities’ investment plans surveyed were justified only by data center demand and load growth. For example, NextEra plans to spend $94 billion to support AI-driven load growth in the state of Florida alone. In California and Texas, Sempra Energy plans to invest $64 billion. American Electric Power, which covers customers across the Midwest and Southeast, plans to spend $72 billion to support data center–driven load growth. All of these utilities serve customers in climate-vulnerable regions and have no named plans for retroactive climate resilience investments or proactive decarbonization- focused infrastructure.
Recent AI expansion has revealed a long-standing problem with utility capex spending, which is the trade-off between the types of capital investments utilities can undertake. Investor-owned utilities are incentivized by profit to invest in local transmission line upgrades or distribution line improvements to secure guaranteed returns. They systematically forgo investing in, for example, the expansion of high-voltage regional and interregional long-range transmission that allow for more renewable generation to come online, more effectively meet household demand, and reduce costs to ratepayers. In 2023, utilities built only 55 new miles of high-voltage transmission compared to 4,000 miles in 2013. This downward trend in transmission investment coincides with the upward trend in overall utility capital expenditure, indicating that utilities are finding ways to spend money, but not on what we need most. In fact, less than 10 percent of annual transmission investment is going to cost-effective, high-voltage transmission lines.21
In 2023, utilities built only 55 miles of high-voltage transmission lines, compared to 4,000 miles in 2013.
Source: Climate and Community Institute, adapted from Grid Strategies.22
AI demand has entered this pre-existing dynamic and utilities are quickly making plans to build local lines to serve specific data centers or rapidly bring new fossil fuel generation online to serve high-demand customers instead of building an expanded grid. Without policy intervention, a significant portion of these investments will be paid for by individual households who do not financially benefit from AI expansion. Communities are rightfully disturbed by the rapid increase in what should be predictable and stable components of their household budget.
Importantly, the infrastructure investments we need to decarbonize the grid are also cost-saving and have the potential to improve household affordability. Various studies have modeled the system cost reduction impact of planned regional investment in high-voltage transmission. One estimate from a recent Grid Strategies report shows that for every $1 in high-voltage transmission invested, customers save $3.80 to $4.70 in energy costs after accounting for the cost of the line itself.23 A more conservative estimate from the 2024 National Transmission Planning Study claims that every dollar spent on transmission creates $1.60 to $1.80 in savings.24
Investments in utility-scale battery storage and building efficiency programs save the system and customers money by reducing the need for long-range transmission and overall energy demand, respectively.25 The ultimate mix of these three types of infrastructure investments will need to be determined at the regional grid level. The research tackling these trade-offs is nascent but growing.26 These findings should be incorporated into robust grid planning efforts.
There is a crucial trade-off between investment in AI-dedicated grid and generation expansion and investments that serve household customers’ bottom line and the renewable transition. Capital investments to support AI expansion do only that. At worst, residential customers are saddled with outrageous bills to pay for infrastructure that they will not benefit from. At best, Big Tech is forced to foot the bill, but residential customers still will not benefit from the new infrastructure. However, they will still have to endure the environmental impact on their local water supplies, noise pollution, and gas power plant pollution.
In both cases, utilities are prioritizing local grid expansion and new generation to meet non-residential, non-essential demand, not the cost-saving and decarbonizing investments we desperately need.
Taking control of essential infrastructure, planning for the public good
The truth is that we desperately need investment in our grid infrastructure, but we need to decide for what, and for whom. The power to make these choices currently lies with private utilities and grid operators who enjoy a permissive relationship with state regulators. This black box of decision-making means that decarbonization goals are not systematically included in either private utilities’ or regulators’ priorities, and neither is protecting customers from climate disaster or Big Tech.
We need an alternative approach to infrastructure investment and planning that includes:
Federal investment in interregional transmission to fill the investment gap, relieve grid congestion, support renewable expansion, and lower costs.
Rate design reform to ensure that household ratepayers are not paying for data centers and extreme weather repairs.
Incentives for long-term resilience and affordability that drive investments in grid resilience, energy efficiency, etc., that ultimately lower costs to consumers.
These proposals can coincide and improve upon the existing investor-owned utility model. However, they will not eliminate the profit mark-up that any private utility–led investment will ultimately include. A larger effort to build out the prevalence and reach of public utilities will be essential to ensuring decarbonization goals can be deeply integrated in utility priorities and to keeping all electricity customers’ costs low and predictable while accelerating the transition to 100 percent renewable power.
Utility capitulation to the demands of AI is not a foregone conclusion.27 Community resistance to AI buildout is a justified response to significant changes to household financial conditions. This analysis demonstrates how localized spikes in electricity costs will show up in aggregate macro indicators with disturbing implications for energy burden and inequality. It is critical that this moment of upheaval leads to a deeper interrogation of the specific nature of utility capital expenditures, which have too long served profit-maximizing aims and not the strategic grid expansions needed to support affordability and decarbonization.
Investor-owned utilities are currently doing three things: profiting off of climate damages, profiting off of AI data center expansion, and hindering a grid-scale renewable transition that would reduce costs and mitigate climate change. If this continues throughout this critical decade, our ability to decarbonize in time to avoid catastrophic warming will be jeopardized and household energy burden will only continue to escalate.
Appendix
The average cost per residential customer was calculated using a simple model that treated all residential electricity ratepayers as if they were customers of one large vertically integrated investor-owned utility that uses rate-of-return cost recovery. This approach translated real, planned capex investment data into an average additional cost per household that is compatible with existing average measures of household expenditure on electricity reported in the Bureau of Labor Statistic Consumer Expenditure Survey.
Assuming that 50 percent of the investment is financed through debt with a financing cost of 5.5 percent, and 50 percent is financed by the utility on which it is entitled to charge 9.7 percent in profit, the weighted average cost of capital (WACC) that utilities charge customers on their total financing need is 7.5 percent in this model. This formula assumes that the total $1.4 trillion investment will be spent evenly over the next five years at $280 billion per year and with a 40-year depreciation timeline for each year’s total investment (thus one year of depreciation equals $280 billion divided by 40 years, or $7 billion per year).
I = initial investment = $280 billion r = WACC = 7.5% L = asset life (years) = 40 D=I/L = annual straight-line depreciation = $7 billion CRRt = capital revenue requirement
For just one round of annual investment, the cost to ratepayers for each of the following five years is calculated:
CRR1 = r (I) + D CRR2 = r (I - D) + D CRR3 = r (I - 2D) + D CRR4 = r (I - 3D) + D CRR5 = r (I - 4D) + D
Assuming an additional $280 billion is invested each year for the next five years, the cost recovery requirement compounds as additional investment is added. The total cost to ratepayers is calculated as follows:
CRR1 = [r (I) + D] CRR2 = [r (I - D) + D] + [r (I) + D] CRR3 = [r (I - 2D) + D] + [r (I - D) + D] + [r (I) + D] CRR4 = [r (I - 3D) + D] + [r (I - 2D) + D]+ [r (I - D) + D] + [r (I) + D] CRR5 = [r (I - 4D) + D] + [r (I - 3D) + D] + [r (I - 2D) + D] + [r (I - D) + D] + [r (I) + D]
These calculations continue until year 40 when the asset has completely depreciated. After calculating the total capital revenue requirement for each year, the formula assumes based on historical data that ~40 percent of total costs would be covered by residential customers. Then, the total residential cost burden per year is divided by the total number of residential customers in 2024 (143,144,185 customers) reported by the EIA, in order to calculate the cost per residential customer in that year. For simplicity, it is assumed that the total residential customer base will be unchanged for all years of this model.
CRRt = total capital revenue requirement h = residential customer revenue burden = 40% C = total residential customers = 143,144,185 Cost per residential customer = (h*CRRt)/C
Investment Period
2026 Investment
2027 Investment
2028 Investment
2029 Investment
2030 Investment
2026
$28,000,000,000
$27,475,000,000
$26,950,000,000
$26,425,000,000
$25,900,000,000
2027
$28,000,000,000
$27,475,000,000
$26,950,000,000
$26,425,000,000
2028
$28,000,000,000
$27,475,000,000
$26,950,000,000
2029
$28,000,000,000
$27,475,000,000
2030
$28,000,000,000
Total CRRt
$28,000,000,000
$55,475,000,000
$82,425,000,000
$108,850,000,000
$134,750,000,000
Cost per Residential Customer
$78.24
$155.02
$230.33
$304.17
$376.54
Cumulative Cost per Residential Customer
$78.24
$233.26
$463.59
$767.76
$1,144.30
For the first five years, new capital expenditure is added to the total cost burden each year, so in the year 2030 the annual average cost to each customer is at its peak of $376.54. To estimate an average cost per customer that reflects a lower percentage of residential burden, the value of h was altered to 20% in this model and the same calculation was conducted to arrive at an average additional cost of $188/household. In practice, the capital expenditure financing costs show up on utility bills as a combination of a fixed monthly payment and a volumetric cost—an additional cost per kWh. This analysis does not aim to parse out these payment structures. Instead, it estimates a cost per customer assuming each customer is charged an equal average cost per year.
Household Cost Burden by Income Quintile
To calculate the change in household cost burden by income quintile, 2024 averages for each quintile reported by the Bureau of Labor Statistics Consumer Expenditure Survey are used as a baseline. The average cost per residential customer is then added to the 2024 baseline for each quintile in each year of the model, and a percentage increase is calculated. These are inflation unadjusted numbers. Scroll right to view the full table.
Income Quintile
2024 Mean Electricity Exp.
2026 Total Electricity Exp + CapEx Impact
% Increase 2026
2027 Total Electricity Exp + CapEx Impact
% Increase 2027
2028 Total Electricity Exp + CapEx Impact
% Increase 2028
2029 Total Electricity Exp + CapEx Impact
% Increase 2029
2030 Total Electricity Exp + CapEx Impact
% Increase 2030
Quintile 1
$1,276.00
$1,354.24
6.13%
$1,431.02
12.15%
$1,506.33
18.05%
$1,580.17
23.84%
$1,652.54
29.51%
Quintile 2
$1,674.00
$1,752.24
4.67%
$1,829.02
9.26%
$1,904.33
13.76%
$1,978.17
18.17%
$2,050.54
22.49%
Quintile 3
$1,813.00
$1,891.24
4.32%
$1,968.02
8.55%
$2,043.33
12.70%
$2,117.17
16.78%
$2,189.54
20.77%
Quintile 4
$2,023.00
$2,101.24
3.87%
$2,178.02
7.66%
$2,253.33
11.39%
$2,327.17
15.04%
$2,399.54
18.61%
Quintile 5
$2,376.00
$2,454.24
3.29%
$2,531.02
6.52%
$2,606.33
9.69%
$2,680.17
12.80%
$2,752.54
15.85%
US Bureau of Labor Statistics, Consumer Price Index for All Urban Consumers: Electricity in U.S. City Average (CUSR0000SEHF01), FRED, Federal Reserve Bank of St. Louis, accessed August 2, 2026, https://fred.stlouisfed.org/series/CUSR0000SEHF01.
US Bureau of Labor Statistics, Consumer Price Index for All Urban Consumers: All Items in U.S. City Average (CPIAUCSL), FRED, Federal Reserve Bank of St. Louis, accessed August 2, 2026, https://fred.stlouisfed.org/series/CPIAUCSL.
This estimate comes from PowerLines, “Utility Spending is Rising,” appendices A and B. The total spend values for utilities that marked both data center demand and extreme weather mitigation as investment drivers were added to arrive at $1.16T in investment.
Winston Yau, Matt Haugen, Jesse Goldstein, Dustin Mulvaney, Hannah Story Brown, Kenny Stancil, and Sarah Knuth, “AI First: How the Federal Government is Prioritizing AI Over People and Planet,” Climate and Community Institute, June 30, 2026, https://stopgreedbuildgreen.climateandcommunity.org/posts/ai-first.
Using our estimate that utilities’ planned capex will result in an additional $376 in fixed electricity costs for residential customers in the year 2030 (see Appendix for methodology), we calculated the impact of that increase on 2024 electricity spending by income quintile. The source for income quintiles and mean expenditures on electricity per income quintile is the Bureau of Labor Statistics spreadsheet data on quintiles of income before taxes for the year 2024 within the table entitled “Calendar year aggregate expenditure shares across selected groups tables by demographic characteristics, 1989 forward.” See US Bureau of Labor Statistics, “Tables,” Customer Expenditure Surveys, last updated May 14, 2026, accessed August 23, 2026, https://www.bls.gov/cex/tables.htm. Income quintiles in the chart are rounded to the nearest $500.
Edison Electric Institute, “Industry Capital Expenditures”; Edison Electric Institute, “Electric Companies Projected to Invest Nearly $208 Billion in 2025”; PowerLines, “Utility Spending is Rising.”
Rangrang Zheng, Greg Schivley, Matthias Fripp, and Michael Roberts, “Optimal Transmission Expansion Modestly Reduces Decarbonization Costs of U.S. Electricity,” SSRN, September 25, 2025, https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5527781.
