About
About this site
The Data Center Debate is a free, non-partisan resource on the US data center buildout. It does not take a side. It shows what each side says, the evidence behind it, and who paid for that evidence, so you can decide for yourself.
Why it exists
Data centers are being announced faster than power plants are being built, and the arguments about who pays and who benefits are happening in county board rooms, utility commission dockets and state legislatures at the same time. Most of what people hear about it comes from someone with a stake in the outcome. This site was built so that anyone, on either side or on no side yet, can see the whole picture and check it.
Contact
Corrections, questions and privacy requests go to contact@datacenterdebate.org, or through the form if you would rather not use email. Corrections to a source, a number or a map entry are especially welcome: say what is wrong and, where you can, link the document that shows it. Every error we have found so far is on the corrections page, and anything you find will go there too.
How this site is funded
A site that grades everyone else's funding should disclose its own. This site is a personal project, paid for out of its builder's own pocket. It carries no advertising and has no sponsors. Readers who find it useful can chip in toward keeping it going, and anything sold here exists only to cover them; contributions are voluntary, are not tax deductible (this is not a nonprofit), buy nothing, including influence, and are covered by the terms and disclaimer. The shirts are printed and shipped by Criterion B, DBA Canonball Creative Inc., the company this site's author runs, through its Fulfill Engine account. It pays the site's author a $6 royalty per shirt for the design and keeps what is left after costs; the disclosures page has the detail. That is a commercial relationship, and it is stated here because this site asks who funds everybody else. It is not a data center operator, utility, industry group or advocacy organization. No money is accepted from data center operators, utilities, industry groups, or advocacy organizations on any side of this debate. If that ever changes, this page will say so before it does.
The rules we follow
- Both sides get their best case. Every argument is presented the way its own advocates make it, with the other side's strongest response underneath and the reply to that. Nothing is a straw man.
- Every number links to its document. Numbered sources never change meaning: source 22 is always Virginia's JLARC study.
- Every document says who paid. Industry-funded, advocacy-funded, government, academic or press. Funding is not disqualifying, but you should know it before you quote the result.
- Each side's weaknesses are printed. The handbook tells each side where it is weak and where not to overclaim. Those sections are on this site in full.
- Modeled is labeled modeled; measured is labeled measured. A 5.5 million jobs figure from an input-output model and a 100 to 200 jobs figure from counting counties are both real numbers that answer different questions.
How the daily update works
Once a day, an automated research pass looks for what changed: new studies and reports, local zoning decisions, utility filings, company announcements, and legislation. Each new item is summarized neutrally, tagged with the arguments it touches and which side it strengthens, and given a short "check it" note on what to verify. New map projects, report center entries and tracked numbers are added the same way. The pass is automated, so it can make mistakes; everything it adds carries its source link so you can check it, and corrections are welcome.
The map
Existing facilities come from public map data (OpenStreetMap), a community-maintained map that is incomplete: many operating data centers are not yet drawn. Projects under construction, proposed, or blocked are curated from news and public records with a source for each. A dot means a project was reported, not that any claim about it is true.
Privacy
No tracking cookies, no accounts, no ads. Visits and handbook downloads are counted anonymously, analytics can be turned off with one click, browser privacy signals are honored, fonts and map software are served from this site so nobody else sees your visit, and raw records are deleted after about 13 months. The full statement, the opt-out switch and the contact form are on the privacy page.
No tracking cookies, no accounts, no ads. We count visits and handbook downloads two ways: a one-way daily hash on our own server that cannot be turned back into a person, and a self-hosted Matomo analytics instance run in cookieless mode, so no tracking cookie is ever set and no data goes to an advertising network. Both exist only so we know which pages help people and how often the handbook is downloaded. The poll uses a separate one-way hash that does not rotate, so one person can answer once; it cannot be turned back into a person either. Your browser's own storage holds two things only: your poll answer, and your choice on the button above. Analytics is on by default; the button turns it off for this browser, and we also honor the Global Privacy Control and Do Not Track signals automatically. No data is sold or shared with advertisers, and nothing here identifies you.
Corrections
If a source is misread, a number is stale, or a project's status is wrong, please say so through the contact form. Corrections are listed in the change log once applied.
The handbook
The site grew out of Data Centers: A Debate Handbook, compiled september 13, 2026. revised september 24, 2026 (point 8 added to both sides on the richland parish, louisiana project and who carries the cost when a power plant outlives its contract; a fifth and sixth genuine collision added on who funds the advocates and on the foreign influence question; the un digital agency's growth figure added to the agreed facts; source numbering aligned with datacenterdebate.org, so a note number here is the same source number there).. It is free to download and share.