News

State of Minecraft Servers — September 2026

· · 7 views

Our server list pings every listed Minecraft server around the clock — status, player counts, versions, and MOTDs, sampled continuously and stored. That adds up to a dataset almost nobody publishes from: not opinions about the server scene, but measurements of it. This is the first monthly report from that data.

Numbers below come from 360 tracked Java Edition servers over the last 30 days (methodology and caveats at the bottom). Short version: Minecraft server activity peaks Saturday evening, a handful of giants hold almost all the players, a fifth of servers are barely alive at any moment, and nearly half the ecosystem now sits behind multi-version proxies.

When Minecraft actually gets played

Across every server we track, combined player counts peak on Saturday at 19:00 UTC — averaging around 307,000 concurrent players on tracked servers — and bottom out on Monday around 11:00 UTC at roughly 69,000. That's a 4.5× swing between the week's high and low.

Saturday busiest day
Sunday
Friday
Tuesday
Thursday
Wednesday
Monday quietest day

The daily curve is exactly what you'd expect from a player base of students and after-work adults: activity climbs steadily from noon UTC, plateaus through the European and American evening (16:00–21:00 UTC), and drains overnight.

If you run a server: schedule events for Saturday evening in your dominant timezone, and do restarts and maintenance Monday mornings. The audience difference is more than fourfold.

The 95/5 economy

Minecraft servers are a superstar economy, and the data is stark:

The top 5% of servers hold 95% of all players. The top 10% hold 98%. The median server averages 0.7 concurrent players — and just over half of all tracked servers average fewer than one player online.

This isn't a knock on small servers — most are private SMPs for a friend group, perfectly healthy at eight players on a Saturday. But it reframes what "average server" means: the typical Minecraft server is a quiet place that spikes to life a few evenings a week, not a bustling hub. If your server averages even 5 concurrent players, you're comfortably in the upper tier of the ecosystem.

Uptime: most servers are solid — a fifth are ghosts

99–100% uptime 66% of servers
90–99% 9%
50–90% 6%
Below 50% 19% — effectively dying or dead

Two-thirds of servers post 99%+ uptime over 30 days, which says good things about modern hosting. The interesting cohort is the bottom: 19% of servers were reachable less than half the time, and at any given moment about 21% of listed servers are offline. Separately, 16% of servers haven't answered a single ping in two weeks — the server graveyard grows continuously, and most owners never formally shut down; they just stop paying the host.

Version adoption in the 26.x era

1.21.x 45% — current major line
Multi-version proxies 47% — Velocity/Bungee accepting 1.7 → 26.x clients
Older versions (1.8–1.20) 8% — the legacy tail

Counting what servers advertise: about 45% still identify as 1.21.x, while nearly half the ecosystem now answers through a multi-version proxy — overwhelmingly Velocity — accepting clients from 1.7 all the way to the current 26.x releases. Genuinely old servers (1.8 anarchy holdouts, 1.12 modpack relics) are down to a rounding error at around 8%.

The proxy takeover is the real story here. For half the ecosystem, "what version is your server?" no longer has a single answer — the proxy speaks every protocol and the backend version is invisible from the outside.

What 380+ MOTDs told us

We also analyzed the message-of-the-day lines servers broadcast. The most common meaningful words:

  • survival (60 servers) — still the undisputed king of gamemodes
  • SMP (40) — the friend-group survival multiplayer identity is its own brand now
  • join (27) and discord (26) — roughly 1 in 12 servers advertises a Discord in the MOTD itself
  • season (23) — seasonal resets are now standard practice for SMPs
  • offline (22) — servers whose MOTD is a maintenance message that never got removed, a quiet marker of abandonment

(Writing your own? Our free MOTD generator handles the color codes and the escaping.)

One more number

In just the first week of sampling player names from ping responses, we observed 3,678 distinct players across tracked servers. That sampling powers per-server unique-player estimates we're rolling out to server owners — expect a deeper dive on player retention in a future report.

Methodology & caveats

Sample: the 360 Java Edition servers listed on minecraft-java-servers.com, pinged via the standard Server List Ping protocol roughly every 30 minutes, with 30 days of history behind the uptime, activity, and population numbers. The sample skews toward small community servers and SMPs (that's who lists on server lists), with several large networks included. Player-count concentration uses per-server 7-day averages. Version figures reflect what servers advertise, which for proxied servers is the proxy's supported range. This is not a census of all Minecraft servers — it's a longitudinal measurement of a real, growing sample, and the same methodology will apply next month.

Cite freely. If you use these numbers in a video, article, or post, a link to this page is all we ask. Data questions or corrections: contact us. Next edition: October 2026.

Tags: minecraft server statistics state of minecraft servers minecraft data server uptime version adoption minecraft trends
More from the blog
Minecraft Server Backups That Actually Restore (2026 Guide)
Aug 22, 2026
Most Minecraft server backups are quietly corrupt — copied while the server was still writing region files. Here's a set…
How to Fix Minecraft Server Lag: 2026 Optimization Guide
Aug 22, 2026
Most lag advice is cargo-culted. This guide shows you how to tell TPS lag from network lag in 30 seconds with spark, the…
How Much RAM Does a Minecraft Server Need? (2026 Guide)
Aug 22, 2026
The honest answer, with numbers: 4 GB covers most vanilla and Paper servers up to ~20 players, modpacks need 6–10 GB, an…