Photon Fusion vs Unity Netcode at 5,000 CCU: Verify Cost and RTT

Premium Deals
Mighty Travels Premium
Travel in style,
save up to 90%

On flights and hotels worldwide by booking the best deals when they appear.

See Deals

Sponsored

TakeawayDetail
Treat the headline's $2,000/mo and 45ms RTT as unverified at 5,000 CCUThe thesis states neither figure is established by the grounded sources; require a written per-CCU quote and a paired 5,000-CCU p95 RTT test before choosing.
Grounded cost brackets are $100-$300/mo at 100 CCU and $3,000-$8,000/mo at 10,000 CCUThe spendark source gives those endpoints only; no grounded figure exists for 5,000 CCU, and containerized servers typically save 30-50%.
Unity's pricing page lists COST.3,000/Month for gaming and entertainment applicationsThe Unity page states that cost with a monthly subscription required; it does not provide a 5,000-CCU Photon Fusion or Netcode quote.
Default to Unity Netcode for GameObjects on containers when you have server engineersThe reader rule sets that default; Photon Fusion 2 is lower-risk only if its written per-CCU quote clears the monthly ceiling and the paired 5,000-CCU p95 RTT test shows an acceptable latency penalty.

At 5,000 CCU in 2026, neither the $2,000/mo nor the 45ms RTT headline figure is established by the grounded sources, so the guide treats both as unverified.

The choice changes only after a written per-CCU quote clears your monthly ceiling and a paired 5,000-CCU p95 RTT test shows an acceptable latency penalty.

Photon Fusion vs Unity Netcode at

How per-CCU relay billing works

Before treating Photon Fusion 2 and Unity Netcode for GameObjects as directly comparable, require written confirmation of how each option is billed and what the current Unity session workflow requires. At 5,000 CCU, keep the relay line item separate from the application workflow: request the vendor’s written per-CCU terms for Photon Fusion 2, and check Unity’s current official documentation for its Netcode for GameObjects session setup before planning build work.

For any Photon Fusion 2 relay option, a 5,000-CCU procurement check should require a written quote that states the included CCU allotment and the overage rate. Tick frequency and network conditions can change how much gameplay each connection can sustain; they should not be used to turn a relay quote into a per-tick or per-RTT estimate. Treat any verbal estimate as provisional until the vendor confirms, in writing, how peak concurrency is counted, how bursts above the included allotment are billed, and which regions the quoted relay capacity covers.

For Unity Netcode for GameObjects, check the current official Unity session documentation and evaluate the framework as part of the server-authoritative implementation and hosting plan rather than as a relay SKU priced by CCU. When comparing options, separate the networking framework’s development model from the infrastructure bill so that a relay quote is not mistaken for a complete hosting forecast, and a hosting forecast is not mistaken for a managed relay price.

RTT is not a vendor SKU and cannot be read from a pricing page. It must be measured after region, tick rate, and transport are fixed for the same gameplay scenario. Treat any published tick rate as a test configuration, not a default, and do not reuse it as a proxy for production latency. A valid comparison pairs the same map, player behavior, region, and tick settings across candidates and records p95 RTT under a sustained 5,000-CCU load.

sleek glass and steel bridges running side side over misty
sleek glass and steel bridges running side side over misty

What grounded cost and RTT data shows

The grounded material supports cost scaling and workflow scope, but it does not establish a 5,000-CCU monthly quote or a latency number. Use it to set checks, not to pick a winner.

Run four checkpoints before using the sheet: (1) the SpendArk 4-resource template observed $855.80-$1,996.44/mo; (2) SpendArk observed 30-50% container savings above 1,000 CCU; (3) KinematicSoup's per-CCU relay billing means the vendor quote must replace the $2,000 headline; and (4) run a paired 5,000-CCU p95 RTT test at the same 20Hz tick and region before accepting any latency penalty.

Cost of choosing wrong: if a future Photon quote lands at the unverified $2,000/mo placeholder, it is $1,144.20/mo above the SpendArk template low of $855.80, or $13,730.40 over 12 months; it is only $3.56/mo, or $42.72 over 12 months, above the template high of $1,996.44. The same quote therefore looks either material or marginal depending on which container baseline the workload actually hits.

The 30-50% container saving also gives a VM-side sanity check: if the template figures are the containerized case, the implied dedicated-VM range is $1,222.57-$3,992.88 at the same two endpoints. Use that only to test whether a quote is plausible, not to replace the written per-CCU number.

Keep the sheet auditable: store the quote, test method, 5,000 CCU load, 20Hz tick, region, and resulting p95 RTT delta with the cost comparison. If either the quote or the paired test is missing, the decision stays open; do not let the unverified $2,000/mo or 45ms placeholders fill the gap.

What grounded cost and RTT data shows — Photon Fusion vs Unity Netcode at

Decision rules for the 2026 build

Treat the proof sheet as a gate, not a scorecard: every vendor choice needs a written artifact that names the test conditions, and no unverified figure should become a pass/fail criterion.

If your launch forecast is 5,000 CCU in one region and you do not have dedicated-server engineers, Photon Fusion 2 is a candidate only after the vendor quote shows the per-CCU total clears your written monthly ceiling. Because KinematicSoup bills per CCU, require the quote to state the CCU basis and to separate relay, egress, support, and overage lines before you compare it with any internal number.

If you already operate containerized dedicated servers above 1,000 CCU, run a Unity Netcode for GameObjects session test before renewal or migration. SpendArk reports 30-50% container savings over dedicated VMs above that level, so compare your current VM baseline with a containerized session build at the same tick rate, region, and match size.

If your game has a 45 ms RTT hard requirement, do not sign either contract until a paired 5,000-CCU p95 RTT test measures the penalty under your actual match size, region, and netcode settings. Use the same client population and server placement for both options, and write the pass/fail threshold into the contract or acceptance criteria.

If neither artifact clears, default to the stack your team can staff and operate without new infrastructure, and record the quote, test method, and thresholds in one decision memo. If a vendor cannot provide a written per-CCU quote or a reproducible 5,000-CCU test, treat that as a fail and revisit only after the missing evidence arrives.

ConditionRequired evidenceFail action
5,000 CCU, no dedicated-server engineersWritten per-CCU quote with line itemsHold the vendor choice
Containerized servers above 1,000 CCUPaired session cost test at same tick rateRetain current VM baseline
45 ms RTT hard requirementPaired 5,000-CCU p95 RTT testDo not sign either contract

What to do next

StepActionWhy it matters
1Treat the headline's $2,000/mo and 45ms RTT as unverified at 5,000 CCU; require a written per-CCU quote and a paired 5,000-CCU p95 RTT test before choosing.The thesis states neither figure is established by the grounded sources.
2Request a written per-CCU quote for 5,000 CCU covering Photon Fusion 2 relay and Unity Netcode for GameObjects paths.No grounded 5,000-CCU figure exists; spendark gives only $100-$300/mo at 100 CCU and $3,000-$8,000/mo at 10,000 CCU.
3Run a paired 5,000-CCU p95 RTT test for both options under identical load and region.The 45ms RTT claim is unverified; measured latency is required before choosing.
4If using Unity Netcode for GameObjects on containers, model 30-50% containerized server savings against the $3,000-$8,000/mo bracket.Containerized servers typically save 30-50%, and spendark gives endpoints only.
5Review Unity pricing page COST.3,000/Month gaming/entertainment subscription and note it provides no 5,000-CCU Photon Fusion or Netcode quote.Avoid assuming the Unity subscription covers relay or CCU costs.
6Apply the default: Unity Netcode for GameObjects on containers with server engineers; Photon Fusion 2 without them only after the relay quote clears the monthly ceiling.This is the stated reader rule after quote and test.

Frequently Asked Questions

What grounded cost data is available for lower concurrency levels?

The grounded sources provide cost brackets of $100-$300 per month at 100 CCU, but no grounded figure exists for 5,000 CCU.

How can infrastructure choices affect server hosting costs?

Containerized servers can typically reduce server costs by at least 30% compared to non-containerized setups.

When should a team default to Unity Netcode for GameObjects?

Teams should default to Unity Netcode for GameObjects on containers when they have server engineers available.

Under what conditions is Photon Fusion 2 considered a lower-risk choice?

Photon Fusion 2 is considered lower-risk only if its written per-CCU quote clears the monthly ceiling and a paired 5,000-CCU p95 RTT test shows an acceptable latency penalty.

What specific evidence is required before choosing a networking solution at 5,000 CCU?

You must obtain a written per-CCU quote and conduct a paired 5,000-CCU p95 RTT test before making a choice.

Does Unity's published gaming and entertainment pricing include a specific 5,000-CCU quote for Photon Fusion or Netcode?

Unity's pricing page does not provide a 5,000-CCU quote for Photon Fusion or Netcode, and it requires a monthly subscription.

Quick answers

Are the headline monthly cost and RTT figures verified at 5,000 CCU?No, neither the headline monthly cost figure nor the 45ms RTT headline figure is established by the grounded sources, so both are treated as unverified.
What is required before choosing between Photon Fusion 2 and Unity Netcode for GameObjects?The choice changes only after a written per-CCU quote clears your monthly ceiling and a paired 5,000-CCU p95 RTT test shows an acceptable latency penalty.
Is there a grounded cost figure for 5,000 CCU?No grounded figure exists for 5,000 CCU; the grounded cost brackets are given only at 100 CCU and at 10,000 CCU.
What is the default recommendation for teams with server engineers?The default is to use Unity Netcode for GameObjects on containers when you have server engineers.
When is Photon Fusion 2 considered lower-risk?Photon Fusion 2 is lower-risk only if its written per-CCU quote clears the monthly ceiling and the paired 5,000-CCU p95 RTT test shows an acceptable latency penalty.

Also worth reading: Rollback Netcode: Free GGPO, Photon Fusion, and One Winner: Rollback Netcode: Free GGPO, Photon · Multiplayer game lag compared: Photon Fusion shared vs dedicated for 64 players: Multiplayer game lag compared: Photon · 2026 P2P Netcode: 18-24% CPU Overhead for 10-Player Indie Builds: 2026 P2P Netcode: 18-24% CPU

Premium Deals
Mighty Travels Premium
Travel in style,
save up to 90%

On flights and hotels worldwide by booking the best deals when they appear.

See Deals

Sponsored

Research Methodology & Editorial Standards

We begin by defining the specific objectives the reader needs to accomplish. Primary product documentation and authoritative secondary sources are assembled into a verified research corpus; drafting occurs only after this foundation is in place.

Every quantitative claim is subjected to dual-source verification. Any figure that cannot be independently corroborated is either qualified or omitted.

Published · Last reviewed · Owned by the Semble editorial desk (About, Contact, Privacy).

Related answers