Data Centers Have Been Cutting Your Electric Bill

A new Electric Power Research Institute working paper finds that between 2015 and 2024, data center growth was associated with lower, not higher, retail electricity prices nationally. That runs directly against the public assumption driving local opposition to new data center projects, and against separate industry projections suggesting the pattern is already starting to reverse.

Key Points

• EPRI's study found that for every doubling of data center capacity, average retail electricity prices fell 3.5% nationally and roughly 6% at the state level between 2015 and 2024
• A YouGov poll of 1,000 Americans found more than two-thirds expect electricity prices to rise if a data center is built in their area, the opposite of what EPRI's historical data shows
• PJM, the largest U.S. power grid operator, projects a $6.3 billion increase in consumer electricity costs over the next three years, attributed mostly to data center power demand
• Virginia, the state with the most data centers, saw residential electricity prices rise more than 13% in the past year, according to EIA data
• Bloomberg projects data centers could consume nearly 20% of U.S. electricity by 2035, up from about 5.9% today, a more than threefold increase in demand

Why the Historical Trend Ran Opposite to Public Expectation

EPRI researcher Asa Watten explained the mechanism to Fortune: electricity pricing works on cost recovery rather than production cost, meaning fixed infrastructure costs get divided across however many kilowatt hours are consumed. More demand from data centers means those fixed costs spread across a larger base, pushing average prices down, while also bringing more efficient generation online to meet the added load. A separate Manhattan Institute analysis of the same EPRI paper noted that the states seeing the sharpest electricity price increases, led by California, aren't the ones with the most data centers, suggesting other factors like climate policy costs and wildfire mitigation spending are doing more of that work.

The Reversal Risk

Watten's own caveat is the most important detail in the research: the price-lowering effect depends on data center demand actually materializing at the scale utilities are building for. "If the grid builds capacity, expecting a lot of demand from data centers, and that doesn't show up, that could be a clear story of how data centers could increase prices in the future in a way that they did not do in the past," he told Fortune. Fixed costs get spread across a smaller base than planned, and the same mechanism that lowered prices works in reverse.

Some Investors Are Already Pricing In That Risk

Mark Cuban raised a version of this concern on the All-In podcast, arguing that AI is becoming cheaper to run through efficiency gains fast enough that some of the data center capacity being built now may go unused, joking that "a lot of data centers...are going to be turned into pickleball courts." Tesla and Alphabet share prices fell after both companies announced increased AI capital expenditures the same week, a signal that at least some investors are questioning whether committed spending matches actual demand. The $7 trillion buildout figure cited across coverage is aspirational infrastructure spending, not a guarantee that revenue or usage will scale to match it.

What the Near-Term Data Actually Shows Diverging

PJM's $6.3 billion three-year cost projection and Virginia's 13% year-over-year residential rate increase both suggest the reversal Watten described may already be underway in specific markets, even as the national historical average through 2024 still shows a net decrease. Those two facts aren't contradictory: aggregate national data through 2024 and forward-looking regional projections for 2026 onward are measuring different things, and it's the more recent regional numbers that better answer whether the pattern is turning.

Growth and Budget Planning

For companies weighing AI infrastructure investment, cloud costs, or facility decisions tied to power-intensive operations, the take-home point isn't that data centers are good or bad for electricity prices. It's that the answer depends entirely on whether AI demand growth matches the capacity being built to serve it, a question that's still genuinely unresolved. That uncertainty is worth factoring into any growth strategy work tied to AI infrastructure costs or long-term technology budgeting, rather than assuming either the historical trend or the more alarming forward projections will simply continue.

Microsoft, Amazon, Google, Oracle, and Meta are all investing directly in nuclear, geothermal, and other generation sources specifically to avoid overloading local grids, a sign the hyperscalers themselves see the demand-mismatch risk as real. Teams tracking how AI infrastructure economics might affect their own technology costs down the line should keep watching how this plays out rather than treating either the good news or the bad news as settled. If you're factoring AI infrastructure trends into a broader marketing or technology budget, that's a conversation our AI marketing services team can help you think through.

This article discusses publicly traded companies and investment-adjacent commentary for informational purposes only. It is not financial advice, and Winsome Marketing is not a financial advisor. Consult a licensed professional before making investment decisions.