Dedicated Game Server Specs by Player Count: CPU, RAM, Bandwidth, and NVMe Numbers (2026)

Most “game server hosting” advice stops at vague tiers like “medium” and “high-end.” That is useless when you are trying to keep a 200-player Rust wipe from stuttering at 03:00. This guide works the other direction: it starts from player count, tick rate, and per-player network cost, then derives the CPU, RAM, NVMe, and port speed you actually need to order.

If you want to see how those specs map onto real products, compare dedicated server plans by CPU, RAM, and data-center region on the main site before you read the numbers below.

Start with the three variables that decide everything

• Tick rate or simulation rate — the number of authoritative world updates per second. A 128-tick CS2 server processes 2x the simulation work of a 64-tick server for the same player count.
• Concurrent players — not registered accounts. Most community servers see 15–25% concurrency at peak.
• Per-player network cost — upstream kilobits per player per second, which sets your port speed and monthly transfer.

Everything else — core count, memory, storage tier — follows from those three.

Dedicated game server specs by game and player count

Game	Players	CPU (clock-first)	RAM	Storage	Bandwidth/player
Minecraft (vanilla)	50	4 vCPU @ 4.0 GHz+	8–12 GB	NVMe SSD	≈30–60 Kbps
Minecraft (heavy modpack)	30	6 vCPU @ 4.5 GHz+	16–24 GB	NVMe SSD (Gen4)	≈80–120 Kbps
Rust	150	8 vCPU @ 4.7 GHz+	24–32 GB	NVMe SSD	≈40–80 Kbps
ARK: Survival Ascended	50	6 vCPU @ 4.5 GHz+	24–32 GB	NVMe SSD	≈60–120 Kbps
Counter-Strike 2 (128 tick)	20	6 vCPU @ 4.8 GHz+	16 GB	NVMe SSD	≈100–200 Kbps
Valheim	20	4 vCPU @ 4.2 GHz+	8–16 GB	NVMe SSD	≈50–90 Kbps
Palworld	32	6 vCPU @ 4.5 GHz+	16–32 GB	NVMe SSD	≈70–120 Kbps

Treat the RAM column as a floor, not a target. Mod packs, world size, and view distance push real usage 30–60% above the idle baseline.

Why clock speed beats core count on almost every game tick

The vast majority of survival and FPS servers run their authoritative simulation on one hot thread. A 3.4 GHz, 32-core Xeon will lose to a 5.0 GHz, 6-core Ryzen on a Rust server with 150 players, because the tick loop never gets off that single core. Core count buys you session density — the number of independent game instances you can stack — not the smoothness of any one of them.

Practical rule: reserve the two fastest cores on the box for the main game process, then budget one additional core per 50 concurrent players for physics, AI, and networking threads.

Bandwidth math you can do on a napkin

Take the per-player figure above, multiply by peak players, then add 20% overhead for protocol headers and retransmits:

• 150-player Rust server × 64 Kbps ≈ 9.4 Mbps steady, ~24 Mbps during a wipe-night burst.
• 32-player Palworld × 100 Kbps ≈ 3.2 Mbps, but world-save uploads spike to 100+ Mbps briefly.
• 128-tick CS2, 20 players × 150 Kbps ≈ 3 Mbps — trivial in bandwidth, brutal in packet rate (pps).

Packet rate, not raw megabits, is what starves cheap hardware. A 128-tick server sends roughly 128–256 packets per second per player; at 20 players that is 2,500–5,000 pps that your NIC and firewall must process without jitter. That is why a 1 Gbps port with a weak CPU and no NIC offload can still feel laggy.

Storage: NVMe is not optional for save-heavy worlds

Open-world survival games rewrite large save files on every autosave — Palworld and ARK can push 500 MB–2 GB per cycle. On SATA SSD this produces 200–600 ms stalls; a Gen4 NVMe drive cuts that to 40–90 ms and eliminates the classic “server freeze every 5 minutes” complaint. Budget at least 2x your world size in free space so the write does not have to read-modify-write in place.

Provider tier comparison for the specs above

Tier	Typical 2026 price	Hardware	Fits
Entry bare metal	$55–$95/mo	Ryzen 3600/E5, 32 GB, NVMe, 1 Gbps	Minecraft, Valheim, 1 small instance
Mid-range dedicated	$110–$190/mo	Ryzen 7 / EPYC 4004, 64–128 GB, NVMe, 1–2.5 Gbps	Rust, ARK, Palworld, 2–4 instances
High-clock gaming box	$180–$320/mo	Ryzen 9 / Core i9, 5.0 GHz+, 128 GB, 10 Gbps	128-tick CS2, multi-instance FPS, modded MC

Match the tier to the highest single-instance requirement, not the average of everything you plan to run. Then check the provider comparison section for the data-center region closest to your players — a 4.8 GHz server 40 ms away is worse than a 4.2 GHz server 8 ms away.

Sizing checklist before you order

• Measure peak concurrent players for a full week, not launch day.
• Confirm the CPU’s single-thread passmark, not just core count.
• Confirm NVMe (Gen4 preferred), not SATA SSD, for any survival game.
• Confirm port speed AND pps filtering capacity for FPS titles.
• Leave 25% CPU and RAM headroom for autosaves and player spikes.