Every time you ask an AI a question, somewhere a machine drinks your water and spikes your electric bill. The infrastructure behind artificial intelligence is being built on the backs of communities — and most people have no idea.
Global data center electricity consumption hit 415 TWh in 2024. By 2030, it is projected to more than double — matching Japan's entire national power use. AI is the primary engine of this surge. Goldman Sachs projects U.S. data center power demand will surge 165% by 2030 vs. 2023. A single ChatGPT query uses approximately 10x the electricity of a Google search.
The power grid was not designed for this. The PJM Interconnection — the grid serving 67 million people across 13 states — has already failed to procure sufficient generating capacity for the first time in its history. PJM's own market monitor testified this shortfall is "almost entirely the result of large data center load additions." By June 2027, PJM projects it may fall below federal reliability standards — rolling blackouts during heat waves are a real near-term risk.
Residential electricity rates rose 46% since 2019, hitting 19¢/kWh nationally by end of 2025. In Virginia — the world's data center capital — wholesale electricity costs rose up to 267% near data center clusters. Residential bills could rise 25–75% by decade's end. The costs do not land on the companies building the data centers. They land on you.
A typical large data center uses 3–5 million gallons of water per day — equivalent to the daily water use of a city of 50,000 people. That water cools the servers. When it evaporates, it is gone from the local watershed permanently. And companies are deliberately building these facilities in the driest parts of the country.
Over 32% of U.S. data centers sit in high or extremely high water stress areas. Two-thirds of data centers built since 2022 are in drought-prone regions. U.S. data centers directly consumed 17 billion gallons in 2023 — plus an additional 211 billion gallons indirectly through electricity generation required to power them. Texas data centers alone are projected to consume 399 billion gallons per year by 2030.
The companies involved have the data. Google documented its data centers consumed 6.1 billion gallons in 2022 — a 22% increase from 2021. Microsoft: 1.7 billion gallons (FY2023, up 34%). Meta: 2.8 billion gallons. Amazon: partially disclosed. These are not rounding errors in a tight water budget. They are the budget.
In 2025 alone, Microsoft, Google, Amazon, and Meta committed over $380 billion to AI infrastructure — more than twice what the entire U.S. electric utility industry invests annually. The investment is extraordinary. Who pays for what it requires is the question almost no one is asking at the local level.
The answer is consistently: the public. Grid upgrade costs land on all ratepayers. Water is withdrawn from shared watersheds under minimal permitting. Tax incentives — often negotiated before communities know a deal exists — remove hundreds of millions from state budgets that would fund schools and roads. In Virginia, 25 of 31 communities with data centers had signed nondisclosure agreements with local officials before residents knew what was being built.
The decisions that determine where data centers are built, what tax breaks they receive, and who pays for grid upgrades are made at the state and local level — in city councils, utility commission hearings, and legislature committee rooms that are almost always empty. That emptiness is the strategy.