This started with a crowd outside a county fairgrounds in Tremonton, Utah.
In May of this year, Box Elder County commissioners voted to approve the Stratos Project, a data center campus proposed for roughly 40,000 acres on the north shore of the Great Salt Lake, with its own natural gas power plant. Hundreds of people showed up to that meeting. Thousands more filed formal protests against the water rights application. Six thousand signed a letter to the governor. Residents organized a referendum drive to put the approval on the November ballot. Kevin O’Leary, the project’s most visible investor, went on national television to say the campus was a matter of national security and economic survival, and the protesters answered with chants using his name.
I am an Oregon native, and I wanted to understand what was happening. So I did what anyone does. I searched. And what I found was noise. Every article had a piece of the story and a point of view. The developer’s numbers lived on the developer’s site. The opposition’s numbers lived on the opposition’s Facebook page. The county’s own record was scattered across meeting agendas and PDFs. Nobody had put the whole thing in one place, in order, with a source on every line.
Then I kept looking, and Box Elder was everywhere. Protests in Georgia, Virginia, Arizona, Pennsylvania, Oregon. Stories about power bills going up. Stories about people losing land to transmission corridors and about towns that welcomed a project and then found out what it actually needed. And, just as loud, stories from the other side: the construction jobs, the tax base a rural county had never seen, the argument that this is the infrastructure the next decade runs on and that falling behind has a cost too. Both sides had real facts. Neither side had a neutral place to put them.
That is the gap. For news, we have public broadcasting: a place whose job is to lay out what is known and let you decide. For data centers, the biggest build out of physical infrastructure in a generation, there is nothing like that. There are industry trackers behind paywalls, advocacy sites on both sides, and press coverage that rewards the loudest number.
Take the simplest question. Nobody can tell you how many data centers there are in the United States. Ask five sources and you get five numbers. One says 4,423 [1]. Another says 3,436, with 1,306 operating and 2,082 planned [2]. A third counts 6,986 in the active pipeline. They are all measuring something real, all measuring something different, and none will show you the list. You are asked to trust a total you cannot check.
I believe you should be able to check. That is how science works and how good data analysis works: you show the evidence, you show the method, and the reader draws the conclusion. Data Center Record is my attempt to build that for this subject.
What it is
It is a public record. Every project we track gets a page with the facts that can be sourced: who is building it, where, how big, what it needs from the grid and the water system, what it was promised in incentives, and where it stands today. Under that sits a timeline. Every entry is a dated event with a link to the document or article it came from. If a fact has no source, it is not on the page. If two sources disagree, both numbers appear with their scope stated, and the page does not pick one.
Alongside projects, the record tracks the places. Every county and city that has passed an ordinance, imposed a moratorium, or tabled a vote gets a scorecard row that says what happened, when, and where the record is. The scorecard has colors, but the colors mean something specific and written down, not our opinion of whether a town is friendly.
It also tracks what people say. When a mayor posts support, when a CEO says a query uses less water than an almond, when an investor argues the whole fight is about who bears the cost, that statement goes in the record with the speaker, the date, the venue, and the source. Next to it sits a separate table of measured figures from research, regulators, and company disclosures, each with its methodology and scope. The claim and the evidence sit side by side. We do not write the verdict. You can read both and decide.
And it counts. Because every project on the record has a status and a source, we can publish our own number of operational, under construction, and announced facilities, and unlike the other totals, ours comes with the list. You can download it, count it yourself, and tell us where we are wrong.
The rules
A few rules hold all of this together, and they are the reason the site looks the way it does.
Nothing is published without a source. Not a capacity figure, not a vote date, not a quote.
We do not estimate. If a developer has not disclosed the water use, the page says not disclosed. That is a fact, and it is often the most important one on the page.
We do not rate people or outlets. A statement is recorded as a statement. A speaker’s profile shows what they have said and which of their claims came with data, and lets the pattern speak.
Every page carries the date it was last verified, so you know how fresh the record is before you cite it.
The dataset is released under a Creative Commons Attribution license. Take it, build on it, cite it. The whole point is that it gets used.
Where it is going
Today the record is deep in a few states and thin in the rest. That is honest, and the site says so. Over the coming weeks every state gets its sources wired in, so that news, public utility filings, and council agendas flow into a review queue and onto the record. The map fills in as records are approved, county by county, and the count on the home page moves with it.
The longer plan is to answer the questions people actually type. How much water does a data center use. Do data centers raise electricity bills. Which states have moratoriums. Each of those becomes a page answered from the record, only when the record can answer it, and updated as new figures arrive.
After the United States is covered, the same process points at Canada, the European Union, and Australia, where the same fights are happening with different rules and, in Europe’s case, better public data than we have here.
Who this is for
Reporters who need a citable timeline by deadline. Council members reading the packet the night before a vote. Residents who want to know what was promised in the last town over and whether it was delivered. Contractors and engineers deciding where the work is. Researchers who need a dataset they can trust. And the AI systems that increasingly answer these questions for everyone, which will cite a public record if one exists and make something up if one does not.
Data Center Record is built to be the one that exists.
If you know of a project, a vote, or a document we are missing, every page has a report button. That is how the record grows.