US Electricity Prices Are Up 48% Since 2020 — AI Data Centers Are Making It Worse and Consumers Are Paying the Bill

Something has happened to the American electricity bill that most households have not been clearly told. In January 2020, the average American household paid 12.76 cents per kilowatt-hour for electricity — a rate that had been relatively stable for nearly a decade, sitting at around 13 cents through most of the 2010s while energy efficiency gains kept the total cost of keeping the lights on manageable. By March 2026, that same kilowatt-hour cost 18.83 cents. By September 2027, the US Energy Information Administration projects it will hit 19.01 cents. That is a near-50% increase across seven years, outpacing headline inflation for most of that period, and it is accelerating rather than slowing. If you have noticed that your monthly electricity bill feels heavier than it did a few years ago even though you have not changed how much power you use, you are not imagining it. And while the causes of rising electricity prices are multiple and genuinely complex, one factor stands out as both quantitatively significant and almost entirely invisible in most public debate: the power hunger of artificial intelligence.

The Numbers: How Much Prices Have Actually Gone Up

The jump from 12.76 cents to 18.83 cents per kWh represents more than a percentage-point shift in a utility rate. It is the difference between manageable and difficult for households that are already navigating the broader cost-of-living pressures of the past several years. Residential electricity prices have risen faster than the headline inflation rate every year since 2022, according to the EIA. Electricity costs jumped 6.9% in 2025 year over year, more than double the headline inflation rate of 2.9%, according to Goldman Sachs. In some states, the increases have been far more severe than the national average — Maine's residential electricity rate jumped 36.3% in a single year between May 2024 and May 2025. In states with particularly high concentrations of data centres like Virginia, residential electricity prices have increased by up to 267% over the past five years.

Goldman Sachs projects that electricity prices will continue rising through the end of the decade as data centres account for approximately 40% of electricity demand growth. The bank's research has been explicit about what this means for ordinary households: it will lower disposable income, drag down consumer spending, and slightly slow economic growth in the coming years. The EIA's most recent projections suggest prices could rise a further 40% by 2030 compared to 2025 levels. The average American household currently pays around $1,500 per year on electricity. A 40% increase by 2030 would add approximately $600 per year to that bill. For lower-income households, which already spend up to 20% of their income on energy compared to 3% for higher-income households, that additional burden represents a genuine hardship. Even before the data centre boom, one in four US households reported having difficulty paying their energy bills or maintaining safe indoor temperatures because of energy costs.

How AI Data Centers Are Straining the Grid

Data centres across the United States consumed 224 terawatt hours of electricity in 2025, representing 5.2% of the nation's total power consumption — a 21% increase from 2024. McKinsey projects that data centre electricity consumption could surpass 600 terawatt hours by 2030, which would mean these facilities account for 11.7% of all American electricity usage within six years. The International Energy Agency reports that AI-focused data centres are driving 30% annual growth in demand, compared to 9% growth for traditional server workloads.

The mechanism by which data centre demand raises electricity prices for households is not straightforward, but it is real. Data centres require enormous, round-the-clock electrical loads — a single large AI data centre can draw as much power as a mid-sized city. When a utility's territory suddenly needs to accommodate this kind of load, it must invest in grid upgrades: new power lines, transformers, substation capacity, and often new generation capacity. The cost to build a new natural gas power plant has tripled since 2022, with plants entering service in 2030 or later already reporting costs of $2,000 per kilowatt — a figure expected to rise to $3,000. Those infrastructure and generation costs are not absorbed by the tech companies that created the need for them. They are passed down to consumers through higher monthly utility rates.

In the PJM Interconnection — the regional grid that covers 13 states spanning the mid-Atlantic and Midwest, and which is home to one of the largest concentrations of data centres in the world — the impact has been particularly visible. The cost to secure power supplies in PJM has exploded in recent years, with $23 billion in increased costs attributable to data centre demand, according to the grid's watchdog Monitoring Analytics. That $23 billion does not sit in an account somewhere. It flows directly into the rates that every household and business in the PJM region pays for electricity. "Without mitigation, the data centres sucking up all the load is going to make things really expensive for the rest of Americans," said David Crane, CEO of Generate Capital and a former energy official in the Biden administration. He warned of brownout situations in some US power markets within the next one to two years if demand continues to outpace supply.

Ohio's $1.6 Billion Tax Break Blowout

The story of electricity pricing and AI infrastructure would be challenging enough if it only involved supply and demand dynamics. But it has a second dimension that has generated significant political backlash: the extraordinary public subsidies that state governments have extended to data centre operators, subsidies that were supposed to attract economic development but have instead cost taxpayers far more than anyone anticipated.

Ohio is the most striking case. The state has for years offered a 100% sales tax exemption on equipment purchased for data centre facilities — an incentive designed to attract technology investment to a state that can offer cheap land, reliable power, and a central geographic location. When lawmakers passed the exemption, they projected it would cost Ohio approximately $136 million in foregone tax revenue in fiscal year 2025. The actual figure was $1.6 billion — nearly twelve times the projection. In fiscal year 2024, the exemption had already cost $554 million, more than four times the projected amount, and the number more than doubled the following year as the pace of data centre construction accelerated far beyond anything the state had modelled.

The political fallout was swift. Ohio Governor Mike DeWine — who had previously supported the data centre incentive programme — paused all new exemption requests in late May 2026 after the actual cost figures became public. House Speaker Matt Huffman articulated the position that many legislators had reached independently: "We don't think we should be granting tax exemptions to multi-billion dollar corporations, especially when many of them are already coming here to build these data centres anyway." A bipartisan pair of bills — House Bill 975 and Senate Bill 374 — have been introduced to end the sales tax exemption permanently, with proposed effective dates of October 2026 and October 2027 respectively. A citizen-led ballot initiative to ban new large data centre construction in Ohio was gathering signatures as of June 2026, facing a July 1 deadline to collect over 400,000 voter signatures.

Ohio is not alone. In Georgia, data centre tax breaks are projected to cost $2.5 billion in 2026 alone, including $1.1 billion in losses to local governments. A state audit found that most data centre projects would have located in Georgia even without the subsidy. In Texas, the cost of the sales tax exemption for qualifying data centres grew from an estimated $14.6 million in the 2014-15 biennium to a projected $3.3 billion in 2028-29. The pattern has become familiar: Big Tech announces enormous capital investment. Politicians celebrate and offer incentives. Tax breaks flow. Then, years later, the actual cost arrives — many times larger than estimated — and the question of who pays becomes unavoidable.

What Ratepayers and Governments Are Doing to Push Back

The backlash against data centre electricity costs is no longer limited to advocacy groups and concerned communities. It has reached the White House. President Donald Trump, who acknowledged on the campaign trail that he would cut electricity prices in half within 18 months of taking office, has been confronted with the reality that AI infrastructure is working against that goal. Trump secured a public pledge from Microsoft in January 2026 that it would implement changes preventing US consumers from covering its AI data centre electricity costs. The administration has also signed a pact with several states that calls for tech companies to pay for new power plants in the PJM grid — a direct acknowledgement that the current model, in which ratepayers cover infrastructure costs driven by data centre growth, is politically unsustainable.

State-level legislative activity is accelerating. Beyond Ohio, multiple state public utility commissions have begun requiring tech companies to demonstrate that new data centre connections will not unreasonably burden existing ratepayers before approvals are granted. Virginia — home to the largest concentration of data centres in the world, known as "Data Centre Alley" in Northern Virginia's Loudoun County — has seen its electricity prices spike and is reviewing its utility rate structure to explore mechanisms for direct cost allocation to large commercial power users rather than distributing those costs across all ratepayers. Communities that once welcomed data centre investment are increasingly applying conditions: requiring facilities to procure renewable energy, contribute to local grid upgrades, and demonstrate net positive impact on local utility rates.

The federal picture is more complicated. Data centres qualify for a range of tax incentives under the Inflation Reduction Act's clean energy provisions, and the AI industry has significant lobbying infrastructure. But the political arithmetic is shifting. Electricity bills are one of the most visible and politically sensitive household costs, and the combination of documented ratepayer harm, multibillion-dollar tax break overruns, and communities experiencing abnormally low water pressure and grid strain from facilities they did not choose has created a constituency for action that crosses party lines.

Practical Steps for Households Dealing With Higher Bills

None of the policy debates at the state and federal level will show up in your electricity bill this month. While the structural question of who pays for AI infrastructure is resolved, there are concrete steps ordinary households can take to reduce their exposure to rising rates.

The single most impactful action for most households is a home energy audit — many utilities offer these for free or at low cost — which identifies where electricity is being lost or wasted and ranks the potential savings from specific improvements. Sealing air leaks around windows and doors, adding attic insulation, and replacing old HVAC filters are consistently among the highest-return interventions and require no significant capital expenditure.

For households with older refrigerators, washing machines, or dishwashers, the decision about when to replace aging appliances has become more financially urgent as electricity rates have risen. An ENERGY STAR-certified refrigerator uses roughly 40% less energy than one manufactured before 2001. At 18.83 cents per kWh, the operational savings from replacing an old refrigerator are now meaningfully larger than they were at 12.76 cents — the economics of appliance upgrades have changed with the rate increase.

Time-of-use pricing — offered by a growing number of utilities — allows customers who shift discretionary electrical loads like dishwashers, washing machines, and electric vehicle charging to overnight off-peak hours to pay significantly lower rates for that consumption. If your utility offers this option and you are not using it, you are paying peak rates for loads that could easily be shifted.

Smart thermostats consistently deliver among the largest electricity savings for the least upfront cost — typically $150 to $250 installed, with annual savings of $100 to $150 at current rates that will grow as rates rise further. And for the increasing number of households considering rooftop solar, the combination of declining solar installation costs, available federal tax credits, and rising utility rates has shifted the economics substantially toward solar adoption in most US markets. A solar array that made marginal financial sense at 12.76 cents per kWh is a meaningfully stronger investment at 18.83 cents — and stronger still at the 19.01 cents the EIA projects for 2027.

Finally, low-income households should investigate whether they qualify for the Low Income Home Energy Assistance Program (LIHEAP), which provides federal assistance with energy bills, and whether utility arrears or shutoff protection programmes are available in their state. These programmes are under-utilised relative to the population they are designed to serve, and the need for them is growing as electricity costs rise faster than incomes.

Conclusion

The 48% rise in US residential electricity prices since January 2020 is not an abstraction. It is showing up in monthly bills, in the budget calculations of lower-income households that are already stretched thin, and in the political calculations of officials from Ohio to the White House who are beginning to grapple with the question of who should pay for the infrastructure demands of the AI economy. The answer to that question — ratepayers by default, or tech companies by design — has not yet been settled. What is settled is that the current model, in which ordinary households absorb the grid infrastructure costs of some of the world's most profitable corporations while those corporations also receive multibillion-dollar tax exemptions, is generating the kind of political backlash that tends to precede regulatory change. Until that change arrives, the most practical thing any household can do is reduce the number of kilowatt-hours it is buying at whatever rate the utility is currently charging — because that rate is going up.

*This article is for informational purposes only. Data sourced from the US Energy Information Administration, Goldman Sachs, CNBC, Fortune, Signal Ohio, the Environmental and Energy Study Institute, Monitoring Analytics (PJM), and Bloomberg. All figures reflect published data as of June 2026.*