On June 5, 2026, Senator Elizabeth Warren said that if you live near a large data center, your electricity bill has risen by as much as 267 percent over five years. PolitiFact checked it and rated the claim Mostly False. That single exchange captures why this topic is so hard to get a straight answer on: the number was real but described the wrong thing, the underlying concern was legitimate, and both the outrage and the debunking ended up misleading people. Meanwhile a separate figure — $23 billion in customer price increases, with data center demand named as a primary cause — is not seriously disputed by anyone. Here is what is actually happening to American power bills, what AI has to do with it, and what the honest evidence supports.
This matters beyond the politics because electricity is one of the few costs almost every US household pays and almost nobody can opt out of. If AI infrastructure is quietly moving that number, you deserve to know by how much — and to not be told a scary figure that turns out to describe something else entirely.
Quick summary: US average residential electricity costs rose about 42 percent over five years, and data centers are one real driver among several — not the sole cause, and not innocent either. The most-quoted alarming number, 267 percent, refers to wholesale nodal prices rather than household bills, and the supply component those prices affect is only roughly 30 to 50 percent of a typical bill. What is solid: PJM capacity costs rose 174 percent for 2025-26, its independent market monitor named data center load growth as the primary reason, and that flowed into roughly $23 billion in customer price increases expected to persist to at least the end of 2028. Effects are intensely regional — Washington DC, Maryland, Maine and New York saw far steeper rises than the national average.
The Number Everyone Quotes, and Why It Is Wrong
The 267 percent figure is not invented, but it does not mean what it was used to mean. It describes wholesale nodal electricity prices — the price at which power is traded into the grid at a particular point — not the bill that arrives at your house. Yale economist Kenneth Gillingham explained the gap plainly: wholesale nodal prices only raise the supply component of an electricity bill, and that supply component is roughly 30 to 50 percent of what a consumer actually pays. The rest is transmission, distribution, fees and taxes, which do not move with wholesale prices in the same way. So a large wholesale swing produces a much smaller household swing. Source: PolitiFact.
The real national figure is that US average residential electricity costs rose about 42 percent over five years. That is a serious increase — it is simply not 267 percent, and conflating the two makes it easy for anyone defending data centers to dismiss the entire concern by debunking the headline. Source: PolitiFact.
What Is Genuinely Established
Strip out the rhetoric and a well-documented chain remains. PJM, the grid operator covering all or part of 14 mid-Atlantic and Midwest states, saw capacity costs rise 174 percent for the 2025-26 period. Its Independent Market Monitor — not an advocacy group, but the body whose job is watching that market — concluded that data center load growth is the primary reason for recent and expected capacity market conditions. That fed into roughly $23 billion in customer price increases that are expected to persist until at least the end of 2028. Harvard Law's Ari Peskoe summarised the mechanism: data centers are causing tens of billions of dollars of price increases in wholesale power markets, and those costs are spread to all ratepayers by the utility. Sources: PolitiFact; Fortune.
That last clause is the part most worth understanding. You do not need to live near a data center to pay for one. Costs enter the wholesale and capacity markets and are then distributed across the utility's whole customer base. The geography of who pays is much wider than the geography of where the buildings are.
| Claim | Status | What the evidence says |
|---|---|---|
| Bills rose 267% near data centers | Mostly False | That figure is wholesale nodal prices, not household bills |
| US residential power costs rose ~42% in 5 years | Supported | National average; regional variation is large |
| PJM capacity costs rose 174% for 2025-26 | Supported | Market monitor names data center load growth as primary reason |
| ~$23B in customer price increases | Supported | Expected to persist to at least end of 2028 |
| Data centers are the sole cause of rising bills | Not supported | One driver among several; effects concentrated regionally |
Where It Actually Hurts: This Is a Regional Story
National averages hide almost everything interesting here. Over the five-year window, Washington DC saw residential electricity costs rise about 94 percent, Maryland about 74 percent, Maine about 73 percent, and New York about 58 percent — against a national average around 42 percent. If you live in the PJM footprint or the Northeast, your experience of this story is genuinely different from someone in a region with surplus generation and little data center construction. Arguments about whether this is a crisis or a non-issue often turn out to be two people describing different states. Source: PolitiFact.
The Serious Counterargument
It would be dishonest to present this as settled. Researchers at the New Jersey State Policy Lab at Rutgers published an assessment titled, essentially, that data centers are mostly not raising your electric bill yet — arguing the measurable household effect to date is smaller than the coverage suggests, even if the trajectory is concerning. Yale Climate Connections made a related but sharper point: home electricity bills are rising steeply while data centers themselves are often not absorbing proportionate cost, because large industrial customers negotiate rates households cannot. Sources: New Jersey State Policy Lab, Rutgers; Yale Climate Connections.
Read those together with the PJM data and a coherent picture emerges that neither side is advertising. The wholesale-market effect is real and large. The pass-through to any individual household bill so far is real but smaller than headlines imply. And the distribution question — whether households or hyperscalers absorb the cost of new capacity — is the actual fight, and it is being decided in state regulatory proceedings that get almost no coverage.
Why This Gets Worse Before It Gets Better
The demand curve is the part nobody disputes. A January 2026 Bloom Energy analysis projected that total US data center energy demand will nearly double between 2025 and 2028, from about 80 gigawatts to about 150 gigawatts. Adding 70 gigawatts of demand in three years is roughly equivalent to adding the electricity needs of a mid-sized European country. Generation and transmission cannot be built on that timeline, which is precisely why capacity prices spike: the market is paying a premium for scarce firm supply. Source: Bloom Energy analysis, January 2026.
Meanwhile utilities are asking for more. They sought about $9.4 billion in rate increases in the first quarter of 2026 alone, following a record year in 2025 in which roughly $31 billion in increases were requested nationally. Not all of that is AI-driven — grid hardening, storm recovery and aging infrastructure account for a great deal — but the AI buildout is arriving on top of an already-strained system rather than instead of those pressures.
What You Can Actually Do About It
- Check whether your state has a pending rate case. These are public proceedings at your public utility commission, they accept public comment, and they are where the household-versus-hyperscaler cost split is actually decided.
- Look at your bill's supply component specifically. That is the portion wholesale prices move. In many states you can switch supplier while keeping the same utility delivery, which is the only part of the bill you have direct leverage over.
- Check whether your utility offers time-of-use rates. Data center demand is relatively flat around the clock, which makes household off-peak shifting more valuable than it used to be in strained regions.
- Treat any single scary percentage with suspicion, including from people you agree with. The 267 percent claim came from a senator and was rated Mostly False. Ask whether a number describes wholesale prices or your actual bill.
- If you are choosing where to live in the Northeast or mid-Atlantic, electricity cost trajectory is now a real variable, not a rounding error.
There is a quieter point in all of this for anyone using AI daily. Every model query runs in one of these buildings, and the electricity cost of inference is ultimately in the price of the tools. That is an argument for using the right-sized model rather than reflexively reaching for the largest one — a small fast model answers most everyday questions at a fraction of the compute. LumiChats makes that easy by keeping many current models under one login at a pay-per-day price, so picking a lighter model for a light question is a normal habit rather than a separate subscription decision.
01Are AI data centers really raising my electricity bill?
Partly, and it depends heavily on where you live. Data center demand is documented as the primary driver of a 174 percent rise in PJM capacity costs for 2025-26, feeding roughly $23 billion in customer price increases. But data centers are one driver among several, and the household effect so far is smaller than the most-shared headlines suggest.
02Is the 267 percent increase figure true?
No, not as it was used. PolitiFact rated the claim Mostly False in June 2026. The 267 percent refers to wholesale nodal electricity prices, not household bills, and wholesale prices affect only the supply component of a bill — roughly 30 to 50 percent of what you pay.
03How much have US electricity bills actually risen?
US average residential electricity costs rose about 42 percent over five years. Regionally it varies enormously: about 94 percent in Washington DC, 74 percent in Maryland, 73 percent in Maine and 58 percent in New York.
04Do I pay more only if I live near a data center?
No. As Harvard Law's Ari Peskoe put it, data center costs enter wholesale power markets and are then spread to all ratepayers by the utility. You can pay a share of new capacity costs without a data center anywhere near you.
05Will this get worse?
Demand is projected to grow sharply. A January 2026 Bloom Energy analysis put US data center energy demand nearly doubling from about 80 gigawatts in 2025 to about 150 gigawatts by 2028. Generation cannot be built that fast, which is what drives capacity prices up.
06Is there any argument that data centers are not to blame?
Yes, and it deserves a hearing. Rutgers' New Jersey State Policy Lab argues the measurable household impact so far is smaller than coverage implies. Yale Climate Connections argues the sharper problem is distribution — that households face steep rises while large industrial customers negotiate rates households cannot access.
The bottom line: your power bill is going up, AI infrastructure is one genuine reason among several, and the scariest number in circulation does not mean what people think it means. The useful question is not whether to be outraged but who ends up paying for the new capacity — households through spread costs, or the hyperscalers driving the demand. That gets decided in state regulatory proceedings that almost nobody watches, which is exactly why it is worth knowing they exist.
