How player count actually scales your resource use

Performance & Lag Reviewed September 6, 2026 2 min read

«How many players can this plan hold?» has no single answer, because players do not cost a fixed amount each. What they cost depends almost entirely on how spread out they are.

Why it happens

The cost of a player is mostly the chunks they keep loaded. Ten players standing together at spawn share almost all of their loaded chunks and cost barely more than one. Ten players exploring in ten directions each load their own area, and the server ticks all ten at once.

Rough guidance, vanilla-ish Paper

PlanComfortable for
4 GBUp to about 10 players, light plugins, a normal world
8 GBAround 20–30 players, a full plugin set
12 GB40 or so, or a small network with a proxy
16 GBA real public server, or several game modes
These numbers assume nothing pathological

One villager hall or one badly-written plugin moves every row of that table down a tier. A 16 GB server can struggle with eight players if one of them built a machine that ticks a thousand entities. Player count is a weak predictor; what those players built is a strong one.

Working out your own number

  1. Watch memory at your actual peak

    Not the average, and not right after a restart. The number that matters is the highest it reaches with your busiest evening online.

  2. Watch TPS at the same moment

    Memory and TPS fail differently. Full memory means crashes; low TPS means everything feels slow. They need different fixes.

  3. Add a margin

    Running at ninety percent of capacity means the first unusual evening is an outage. Sixty to seventy percent at peak is a server that survives a surprise.

Exploration is the expensive activity

A new player exploring generates terrain, which is the heaviest thing a server does. Pregenerating the area you expect people to use converts that repeating cost into a one-off, and it is the single best thing you can do before opening to a group.