Issue 004 · Part 1 of 2 · Data Centers & the Grid
Your electricity bill is going up. Here's the part nobody explained.
This is an infrastructure post. It explains how data center construction affects electricity prices, water supplies, and grid stability, using documented data from utility filings, the US Energy Information Administration, and the International Energy Agency. Regardless of your politics, understanding data centers and how they work is useful information. I had no idea what a data center was until last week and I figured if I didn’t know, there were probably others who didn’t,
In Part 1 I’m going to cover three fears people have about data centers, and Part 2 covers what you can do, things like engaging the public processes that govern these decisions and looking at your environmental footprint. A standalone preparedness issue on outage readiness is coming after that, and if I hear from all of you that there are more fears I’ll add a Part 1.5 to discuss those.
FEAR 01
"My electricity bill keeps going up and I don't understand why, or when it will stop."
U.S. electricity prices rose 27% between 2019 and 2024 according to the Energy Information Administration. That increase came after a decade during which residential electricity prices had been essentially flat, so the rising costs came as a shock to most of us. Analysts at the EIA and the Energy and Environment Study Institute believe there will be a further increase of roughly 40% by 2030. Those two numbers together are huge: 27% increase already, with a 40% increase projected. That represents a big change for a cost that most of us had treated as relatively predictable background noise.
Before going further, I want to note that there are two separate pressures are stacked on your electricity bill right now. This post is about the structural one, the grid, because this account is AI focused, but the war in Iran that began in late February 2026 has added a second level of pressure through the global gas market. I’m going to use the phrase “as I understand it” a lot in this series because I’m learning about all of this as I go and I’m bound to get some of it wrong, but as I understand it, oil prices/the war and data centers are related but distinct: data centers are bending the long arc of prices up, and oil prices/the war are the reason why that arc went up more sharply than expected.
But back to the grid, here’s why prices stayed flat for so long, as I understand it: between roughly 2010 and 2019, demand growth in the U.S. was modest, natural gas prices were low, and renewable energy was staying/becoming cheaper. The grid had slack. Utilities weren't under pressure to build new capacity, so prices stayed flat.
What changed was demand. Data center construction has accelerated in ways that weren’t even projected a few years ago. Here are some shocking numbers: the five largest technology companies spent a combined $300 billion on data center construction in 2025 alone. The Electric Power Research Institute estimated in 2024 that data centers could account for up to 9% of total US electricity consumption by 2030, up from roughly 4% in 2023.
That demand has to be met somehow, which means the grid needs to be expanded to meet it. The problem for most of us is that the cost of that expansion doesn't land only on the companies building the centers. When utilities file for rate increases with state public utility commissions, which they are required to do before raising rates, the new infrastructure costs get spread across all ratepayers. For example, a family in Fairfax County, Virginia, or outside Columbus, Ohio pays a portion of the transmission line costs built to serve the nearby data center they will never enter.
In January 2026, NPR found that families in Ohio had seen electricity bills rise more than 60% over five years in areas with a high density of data centers. Specifically,130 data centers had been built in in Ohio. The utility's rate increases, filed and approved by the state public utility commission, cited infrastructure expansion as the primary reason.
Basically, the increase isn't because your personal usage changed. It's because the grid your household connects to is serving a substantially larger and faster-growing set of demands than it was designed for, and the cost of accommodating that demand is allocated across all ratepayers by default.
What you can actually do?
None of us can stop the rate trajectory at the household level. What each of us can do, though, is reduce the number of kilowatt-hours your household uses, which will shrink the dollar impact of each percentage increase in the rate itself. Thinking about that 40% projected increase by 2030 means that efficiency investments now will give a bigger return on investment in the next few years than they ever have.
→ LED lighting is still the highest single efficiency upgrade for most homes. A household typically saves $150–$250 per year.
→ A programmable or smart thermostat will help tame your bill. Heating and cooling costs are about half of the average residential bill. A thermostat that reduces air conditioning during unoccupied hours can save 10–15% on that portion of the bill.
→ Check whether your utility offers time-of-use rates. Many now charge substantially less during off-peak hours, which are typically overnight, and daytime on the weekends. Running high-draw appliances like the dishwasher, laundry, or EV charging during off-peak windows is a real saving that doesn’t require you to spend anything.
→ Look up your state's financing programs. A number of states and utilities offer zero-interest loans for efficiency upgrades (insulation, heat pumps, window replacements) repaid through your monthly bill. The improvements reduce your usage so the savings will partially offset the repayment.
→ Standby power draw from devices left plugged in but not in active use adds up to 10% of average household electricity consumption. Power strips with individual switches on entertainment centers and home office setups are a practical way to cut costs.
FEAR 02
"I had no idea AI used this much water. Now I'm worried about what that means for my water supply."
Water use of AI infrastructure is the least widely reported and, in water-stressed regions, potentially the most locally significant. Data centers use water, the same water you and I use, for cooling. Servers generate heat, so cooling them requires moving that heat somewhere, and evaporative cooling using water is the most efficient method at scale. A large facility can use up to 5 million gallons of water per day. That's the daily water use of a city of 10,000 to 50,000 people, and its drawing from the same municipal supply as the surrounding households.
In 2023, U.S. data centers directly consumed an estimated 17 billion gallons of water. That’s the annual household water use oof a city the size of Philadelphia, and that figure is projected to more than double by 2028.
Northern Virginia, home to the largest cluster of data centers in the world, draws from the Potomac River basin. Data centers in central Arizona draw from an aquifer system already under pressure from population growth and agricultural use. Are these water systems designed for the combined demands of the existing population and large-scale data centers?
Most of approvals happen at the county zoning level and the state environmental permitting level, in processes that don't get significant public attention before permits are issued. The water draw projections are part of the public record in most cases, in environmental impact assessments and utility water permit applications, but accessing them requires knowing where to look.
Water restrictions don't typically come with an explanation of what changed on the supply or demand side. If you're in a region that has experienced tightening water availability in recent years, data center growth is a variable worth factoring into your understanding of why. Not as the only explanation, but as one that is rarely mentioned in the public communications that accompany restrictions.
What you can actually do
Water resilience at the household level is a separate topic covered in other issues. What this fear specifically opens up is awareness of a demand-side pressure on your local water supply that most people in affected areas don't know about.
→ If you're in a water-stressed region like the Southwest, parts of the Southeast, and any area that has seen multi-year drought conditions, look up whether large data center permits have been filed in your county. County planning department websites and state environmental agency permit databases are the primary sources. Search for "data center," "cloud campus," or "technology campus" in the permit search.
→ Your state's environmental agency will typically publish water permit applications for large commercial facilities. These include projected daily water draw figures. Knowing the scale of what's been approved in your area gives you a more accurate picture of the demand-side pressures your local system is facing.
→ If you have a well rather than municipal water, data center water draw is less likely to affect you directly, although groundwater depletion from nearby large facilities is a documented concern in some aquifer-dependent areas of the Southwest. The USGS publishes groundwater monitoring data by region.
FEAR 03
"I'm worried about blackouts and grid instability."
The US electrical grid was largely built in the mid-20th century and its capacity was designed for demand that looked very different from our current one. Utilities have known for years that the grid needs substantial investment, but what has changed recently is the pace and scale of new demand layered on top of that already existing need.
The North American Electric Reliability Corporation. NERC, the body that monitors grid reliability across the continent, has flagged elevated reliability risk in its annual assessments for the past several years. The 2024 Long-Term Reliability Assessment identified the Midwest, the Southeast, and parts of the West as regions where the reserve margin, which is the buffer between peak demand and total available capacity is narrowing. A narrower reserve margin means the grid has less room to absorb unexpected demand spikes or equipment failures before reliability is affected.
The specific concern for individual households is how grid stress will distributes. Utilities manage peak load through demand response programs, purchasing power from adjacent regions, and in shortage conditions, rolling blackouts. Critical infrastructure stays online first. I used to live next to a hospital, and I saw this firsthand. The blocks around the hospital were never without power for very long because the hospital need to stay online. Large commercial customers are next, and residential customers are typically in the last tier of protection. This means any rolling blackouts will fall disproportionately on residential neighborhoods.
Many data centers have backup generators powerful enough to enough to run the entire facility, which means during a grid emergency they can unplug from the public grid and keep operating on their own power. This means the facility stays online but the houses around it don't get the same option.
Several states are now requiring grid impact assessments before approving new data center permits, in part because regulators have realized that a facility that can disconnect during a shortage event doesn't actually relieve the grid. The grid was built and sized to serve that facility under normal conditions, and households are paying for that infrastructure either way. When the data center pulls itself off the grid in a crisis, the protection is private. The remaining load, the homes and small businesses that can't generate their own power, still has to be managed by whatever capacity is left.
The infrastructure gap between what the grid can handle today and what's being asked of it is real, and that’s a documented assessment from NERC, not an alarmist projection. What it means practically is that the grid is less resilient than it was a decade ago in the regions where demand growth has been sharpest, and that residential customers are not at the top of the protection order when resilience is tested.
What you can actually do
→ A high-capacity power bank (20,000mAh or more) will keep phones and small devices running through a 12–24 hour outage. My research led me to the Anker PowerCore 26,800mAh as the standard recommendation because it charges a phone six to seven times. Regardless of whatever power bank you use, test it every 90 days.
→ Know your utility's outage map and notification system. Most utilities now have real-time outage maps online and offer text or email alerts. Signing up takes five minutes and means you know what's happening and when restoration is expected rather than waiting in the dark.
→ If you have medical equipment that requires power like CPAP machines, oxygen concentrators, home dialysis equipment, register with your utility as a medical baseline customer. Most utilities maintain lists of customers with life-sustaining equipment needs and prioritize restoration to those addresses. The registration is free.
→ Understand what in your home runs on gas versus electricity. I never thought about this, but our gas stove and gas water heater will continue to operate during a power outage (true in most households, but not all). Knowing which of your critical systems have no electrical dependency is basic resilience information.
→ If you're in a region NERC has flagged for elevated reliability risk like the Midwest, Southeast, and parts of the West, and you own your home, a battery storage system paired with solar panels is now a financially reasonable investment, not primarily for environmental reasons but for rate and resilience reasons.
COMING IN PART 2
Part 2 will cover the broader takeaway on where we stand with data centers right now.
In the meantime, keep in mind: the infrastructure covered in this issue isn't a projection. Electricity prices are already up 27%, and building permits are already filed and approved, meaning construction is already underway. Part 2 is about what to do about that.
@ai.prepper · The AI Prepper · AI news. Life readiness. No fluff, no doom.