We track 148 private servers. We have played none of them. That makes this the hardest kind of article to write honestly, because every claim about what players will do is a prediction, and predictions are the thing this site publishes least.
So: a small number of structural claims, each of which follows from data we hold, and no predictions dressed as insights.
The claim we can support: the scene has been consolidating for a year
Our own data shows it. 13 dead entries, and the shutdown reasons are not all the same kind of failure:
- Ascension and four related realms, shut down 4 September 2026 after a Blizzard cease-and-desist.
- Stormforge, two realms, closed after a cease-and-desist.
- Turtle WoW, whose source is dead while at least three rebranded continuations are live, tracked separately.
- Project Alterac, shut down.
- ArgusWoW, dead since May 2022, which we only found in September because its page still showed a live player count.
Two observations from that list. Legal action is a real cause of closure, not a background worry, and it killed the largest classless project in the scene. And successors survive rebrands, which is a different survival mechanism from staying put.
The claim we can support: builds and patches create a scene, not just a server
The Forever beta has produced four builds in five days and a class deep dive programme with numbers in it. Every time Blizzard publishes something specific, the scene has material to talk about.
That is a mechanism, and we can point at where it ran: Luntares, the only Midnight server we track, has published five changelog parts in September. A server with dated changelogs and a closed beta is doing exactly what a scene-forming server does.
The claim we can support: what we publish after launch is the operators' data, not ours
This is the one that governs everything else we do on 33 November.
On launch day we will have: Blizzard's build number, whether regions are in parity, and whatever dated notes exist. We will not have played the game.
Our population bands are operators' own widgets, read on a date and labelled self-reported. Six entries are unknown this week because we could not find a website, and they are unknown rather than dead because a missing website is not a shutdown. If that discipline holds on launch day, our launch-day numbers are a reading of what servers chose to publish, which is a biased sample and we will say so.
What we will not claim
We will not say which servers will survive. We do not have the data and neither does anyone else yet.
We will not rank servers we have not played. Our existing guides are explicitly written from first-party reading, and that limitation is stated in each one.
We will not use launch-day population as a quality signal. A 255 funserver and a blizzlike progressive server will both show numbers, and the numbers do not measure the same thing.
The uncomfortable part
Our tracker is most useful when the scene is stable and least useful in the week it is not, and launch week is the least stable week in the cycle.
A dead server with a live-looking widget is the specific failure that cost us four years of ArgusWoW. A domain that lapsed mid-scene is what took Atlantiss. A project that announces a date and then slips is the same pattern we corrected on Firestorm and uWoW.
All three of those failure modes get worse in the 33 days before a launch and the weeks after, because that is when the most projects are moving at once.
So our launch coverage will be built around the correction policy rather than around excitement: dated readings, struck-through values, and "we could not verify this" where that is the truth. It is less engaging than a launch-day top ten and it is the only version of it we would be willing to sign.
SOURCES: all figures are from our own tracker data on 2 October 2026: 148 entries, 13 dead with their recorded shutdown dates and reasons, and the six entries downgraded to unverifiable this week. The Luntares changelog count is from its own site, read on 1 October 2026. We have played no private server and have no beta client, so every statement about player behaviour is a prediction and is labelled as one. Related: the 33-day launch guide, how we review a private server, and what our beta coverage cannot tell you.