Migration Guide
Unity Multiplay Has Shut Down. Here's Where to Go.
The dedicated server fleet is gone. Matchmaker, Relay, and Lobby continue. Here are your options — and why most studios still haven't migrated.
Updated April 11, 2026
- Unity Multiplay Game Server Hosting shut down March 31, 2026. Unity Matchmaker, Relay, Lobby, and Distributed Authority are unaffected.
- Three main options: Rocket Science Group (Multiplay continuation, least change), Edgegap (Unity plugin, self-serve), Gameye (no SDK required, $0.07/vCPU/hr, no egress, sandbox in 24 hours).
- Migration scope: Replace the server hosting layer. Your game logic, networking, matchmaking rules, and client connection flow don't change.
On December 4th, 2025, Unity announced that Multiplay Game Server Hosting — the dedicated server fleet orchestration layer — shut down on March 31, 2026. Unity wound down internal operations and licensed the underlying software to Rocket Science Group, a company founded by veterans of the original Multiplay team.
The most important thing to understand is what did not shut down: Unity Matchmaker, Relay, Lobby, and Distributed Authority are all unaffected. Unity has also updated Matchmaker to support third-party hosting providers before the deadline, so your matchmaking logic doesn't have to change — only where your servers actually run.
- Multiplay Game Server Hosting
- Unity fleet management dashboard
- Multiplay server SDK (
IMultiplayService) - Built-in Multiplay features in Unity 6.0+ LTS
- Unity Matchmaker (now supports third-party providers)
- Unity Relay
- Unity Lobby
- Distributed Authority
- Netcode for GameObjects
- Unity Gaming Services dashboard
What are your options now?
How do the options compare side by side?
| Rocket Science (Multiplay) | Edgegap | Gameye | |
|---|---|---|---|
| Migration effort | Minimal — same service | Low — Unity plugin | Low — API swap |
| Server-side SDK required | Multiplay SDK (carried over) | Edgegap SDK (optional) | None |
| Works with Unity Matchmaker | Yes | Yes | Yes |
| Pricing model | TBD (not published) | Pay-per-second, transparent | $0.07/vCPU/hr, no egress |
| Egress / bandwidth fees | Unknown | Varies | None — included |
| Provider-agnostic | No — single operator | Edge network (own infra) | Yes — bare metal + cloud |
| Session start time | Fleet-based, minutes warm-up | Just-in-time, sub-second | ~0.5s average |
| Sandbox access | Contact required | Self-service | 24 hours |
| Games-only platform | Yes | No — broader edge | Yes — Founded in 2017, 120M+ sessions orchestrated |
What changes in your codebase when you migrate?
Regardless of which provider you choose, the technical surface of the migration is the same. Here's what you're replacing:
Remove the Multiplay SDK initialization (IMultiplayService) from your dedicated server build. With Gameye, there is no replacement SDK — Gameye requires no code in your game server binary. Your server just needs to listen on the port provided in the session metadata. Other providers may require their own SDK integration.
Wherever your code requests a server from Multiplay's fleet, replace that call with your new provider's session creation endpoint. For Gameye, this is a single POST request to the Session API with your image name, region, and any session metadata.
Multiplay's fleet concept (regions, scaling profiles, server profiles) maps to equivalent configuration in your new provider. This is a one-time setup step, not an ongoing code change.
Both Gameye and Edgegap run your server in a Docker container. If you're not already building your dedicated server as a Docker image, this is the main new engineering step. Unity's Dedicated Server build target produces the binary; Docker wraps it for reproducible deployment.
What doesn't change: Your authoritative server logic, networking model (Netcode for GameObjects or otherwise), client connection flow, matchmaking rules, Relay sessions, Lobby structure, and player-facing behaviour. The migration scope is the infrastructure layer — not your game.
Why do studios migrating from Multiplay choose Gameye?
Multiplay charged for bandwidth. Gameye doesn't. $0.07/vCPU/hr is the entire compute bill — no per-GB charges on top. For studios running high-traffic sessions, this is often where the biggest savings come from. Model your own numbers →
Gameye spins up a session in ~0.5 seconds on average. Players don't wait between "match found" and "server ready." No fleet warm-up, no cold-start delays.
Gameye orchestrates across bare metal and multiple cloud providers. If one region or provider has an incident, sessions fail over automatically. You're not replacing Unity lock-in with Gameye lock-in — your server image can run anywhere Gameye schedules it.
Unity has already updated Matchmaker to support third-party providers. Gameye plugs into your existing Unity Matchmaker configuration — you keep your matchmaking logic, tickets, and rules exactly as they are.
Because Gameye is API-driven and provider-agnostic, you can run Multiplay and Gameye in parallel by region or game mode while validating performance and costs. No hard cutover required.
You can have sessions running and validated today. Gameye provisions sandbox access within 24 hours — no sales conversation required.
Frequently asked questions
Does Unity 6.3 LTS affect anything beyond server hosting?
Unity 6.3 LTS will remove the built-in Multiplay features from the engine. This primarily affects the Multiplay SDK integration in the Unity editor and Dedicated Server build tooling that was tightly coupled to Multiplay's fleet. Core networking (Netcode for GameObjects), Relay, Lobby, and Matchmaker are separate packages unaffected by this change.
Is Rocket Science Group's Multiplay a safe bet long-term?
Rocket Science was founded by the original Multiplay team, which is a genuine signal. They understand the service deeply. The risk factors to consider: they haven't published pricing, they're a newer standalone company rather than a division of a larger organization, and their long-term roadmap isn't public. For studios that want to keep maximum continuity and accept that trade-off, it's a reasonable path. For studios that want infrastructure they control or diversify across, Gameye or Edgegap give you that.
Can I use Gameye without a matchmaker — just on-demand sessions?
Yes. Gameye's Session API is just an HTTP endpoint. Any system that can make an HTTP request can trigger a session — a custom backend, a serverless function, a client-side request (with appropriate auth), or a Unity Matchmaker ticket. You don't need a specific matchmaker to use Gameye.
How does Gameye handle server warm-up compared to Multiplay fleets?
Multiplay used fleet-based orchestration: you pre-allocated server capacity and servers could sit idle waiting for sessions, with cold-start times reported as high as 15–16 minutes when fleet capacity wasn't pre-warmed. Gameye allocates on-demand in ~0.5 seconds using pre-built container images scheduled across available infrastructure. There's no fleet to warm up — you pay for sessions when they run, not for reserved capacity.
What if I'm using Unity Relay for peer-to-peer sessions, not dedicated servers?
Unity Relay is unaffected by the Multiplay shutdown and continues as normal. Relay-based sessions (peer-to-peer with relay fallback) don't use Multiplay's fleet at all — no migration needed for that part of your stack. Only dedicated server sessions that allocate through Multiplay's fleet management are affected.
Deadline: March 31, 2026
Sandbox access in 24 hours — test before the deadline.
Request your API token, push your Docker image, and run your first Gameye session before Multiplay goes offline. Most studios complete parallel testing within a week.