The strongest case, stated plainly · 10 reasons · Evidence attached
Every search, payroll run, 911 dispatch, MRI scan, and small-business sale runs through a building like this. The question isn't whether the digital economy gets built. It's whether your community captures the value when it is.
Local concerns about data centers deserve honest answers. That's what the rest of this site is for. But the national campaign against American AI infrastructure is not entirely organic, and communities deserve to know who is funding the messaging they're hearing.
The U.S. Attorney for the Southern District of New York has opened a grand jury investigation into the finances of the activist network funded by Neville Roy Singham, a U.S.-born tech entrepreneur based in Shanghai and one of the largest private funders of U.S. activist organizing. Source: reporting on the SDNY probe →
House Ways and Means, House Oversight, and the Select Committee on the CCP are investigating whether Singham-funded nonprofits should register as foreign agents under FARA, and have demanded internal documents on their funding and CCP connections. Source: Ways & Means subpoenas →
Bitcoin Policy Institute research documents network outlets collaborating with Chinese state media on anti-data-center content, with same-day publication patterns across affiliated sites. It estimates roughly $23.6 billion in proposed AI infrastructure investment has been delayed, scaled back, or blocked where network-linked organizers mobilized. OpenAI separately identified Chinese-linked influence operations generating opposition content. Source: BPI report → · OpenAI findings → · Cotton referral →
The pressure is producing real policy: 54+ local moratoriums enacted, a dozen states weighing statewide pauses, and a federal bill (S.4214 / H.R.9442) that would freeze new AI data center construction nationwide. A national freeze is precisely the outcome a strategic competitor would want. Source: bill announcement →
Ask who funds the campaign in your town. Local residents with real questions deserve engagement; imported, professionally coordinated opposition deserves scrutiny. Both can be true at once.
Recent posts worth your time: rate data, deal announcements, and good analysis. Updated as new material gets flagged.
Loading feed…
The four claims you'll hear most, against the published numbers. Government and national-lab sources, linked inline.
Measured: every data center in America combined uses about 17 billion gallons a year for direct cooling (DOE / Berkeley Lab). U.S. golf courses use 531 billion, about 30× more (GCSAA/USGA survey). Home lawns: ~3 trillion, nearly 200×. Crop irrigation: ~43 trillion, roughly 2,500×. Data centers are about 0.04% of U.S. irrigation withdrawals. Nobody is protesting the back nine. And per facility, water is a design choice: a 25 MW closed-loop facility runs on ~2,700 gallons a day: nine houses.
The grain of truth: an old-style evaporative facility dropped on a small system without basic water engineering can stress it. That's a permitting checkbox (specify closed-loop or air cooling), not a reason to say no. Even in dry basins, the modern designs make the issue disappear.
Measured: the national evidence runs the opposite direction. EPRI found residential rates would be ~6% higher without the data centers built 2019–2024, and Berkeley Lab found the fastest-growth states generally saw prices fall, because a big, steady, 24/7 customer spreads the grid's fixed costs across more kilowatt-hours and pays demand charges that fund upgrades everyone shares.
And where the grid is tight, the fix already exists: developers increasingly bring their own power. The data center funds or builds its own generation and covers its own demand. It doesn't even need to sell power back to the grid. Its supply costs never touch anyone else's bill, while its fixed-cost payments still flow to the utility. Existing customers keep the contribution without inheriting the load. That's how a data center lowers rates even in a constrained region.
The grain of truth: PJM put concentrated new load on standard tariffs in a supply-short market and bills rose. That's a deal-structure failure, the exact one bring-your-own-power exists to prevent.
Measured: U.S. data centers used 4.4% of the nation's electricity in 2023 (DOE / Berkeley Lab), headed toward 6.7–12% by 2028. That's real growth worth planning for, and it's the best kind of load growth a grid can get: steady, predictable, creditworthy, and paying the demand charges and interconnection costs that finance new substations, transmission, and generation for everyone. Utilities have spent twenty years fighting flat sales; this is the demand that gets things built again.
Measured: a modern facility runs about 45–55 dBA at the property line, quieter than a normal conversation (~60 dBA). Sound drops ~6 dB every time distance doubles, so standard setbacks bring it to suburban nighttime ambient. Virginia's legislative audit measured 40–59 dB even at the facilities that drew complaints, and the complaint cases trace to a zoning failure (facilities permitted within 200 feet of homes), not to the buildings themselves. Setbacks, acoustic enclosures, and modeled noise limits in the permit solve it, and operators routinely accept all three.
SOURCES // DOE & Lawrence Berkeley National Laboratory 2024 U.S. Data Center Energy Usage Report · GCSAA/USGA water surveys · EPA WaterSense & USGS · EPRI · Virginia JLARC · full list in the footer. A county-level version of this analysis is coming soon →
Spotted a better stat, a broken claim, or a community this analysis should cover? Send it. This site is free, ad-free, and stays that way.
A tool that analyzes whether your county can support a data center: which utilities serve it, whether the grid has headroom, the right-sized project for your area, and what it would actually do to local rates, taxes, and water. Sourced, with the arithmetic shown. Here's an early sample: Kootenai County, Idaho.