Unity Multiplay is ending direct support on March 31st, 2026. Unity is licensing the software to Rocket Science Group, but studios using Multiplay Hosting need a migration plan immediately.
Hathora announced on March 4th that its game server hosting platform will shut down on May 5th, 2026. Hathora is joining Fireworks AI, with GameFabric by Nitrado as the official recommended migration path for game studios.
Two shutdowns. Two tight deadlines. Hundreds of live games caught in the middle.
Why the “official” migration paths deserve scrutiny
When a platform shuts down, the easiest path isn’t always the right one.
GameFabric is a Kubernetes-based platform sharing technical similarities with Hathora’s containerised approach. For simple server-spin-up needs without matchmaker dependencies, it may work adequately.
However, most studios aren’t that simple. The majority running on Hathora use matchmakers — Pragma Engine, Nakama, or FlexMatch. GameFabric has no documented integration with Pragma or Nakama. FlexMatch is the one area where GameFabric has published documentation — though it still requires custom Lambda infrastructure to implement.
For Multiplay studios, the deadline is even more pressing. Studios must move quickly, but moving hastily onto the wrong platform creates downstream technical debt lasting years.
How to think about this
“Infrastructure decisions have a long tail. The right question isn’t ‘what can I migrate to fastest?’ — it’s ‘what do I want to be running on in two years?’”
Evaluate these factors seriously before committing to a migration path:
Matchmaker compatibility. If your game uses a matchmaker today, your migration choice is largely determined by which platforms your matchmaker integrates with natively. Custom connectors mean your team owns that integration forever — maintenance, edge cases, and all.
Provider focus. Platforms where game server hosting is a core product — not a side offering — tend to move faster on developer needs and carry less strategic risk of another pivot.
Pricing transparency. Opaque custom pricing with mandatory sales consultation is a warning sign. You need to know upfront what you’re getting into before you’ve already rebuilt your stack around it.
The hidden engineering cost of “managed Kubernetes”
GameFabric manages the Kubernetes cluster — studios don’t operate nodes. But managed infrastructure doesn’t mean zero engineering overhead. The cost sits at the application layer, and it compounds.
Agones SDK integration. GameFabric is built on Agones, which means your game server binary must implement three lifecycle calls: sdk.Ready() on startup, sdk.Allocate() when players connect, and sdk.Shutdown() on teardown. The Agones orchestrator depends on these calls to know server state. A missed or mistimed call produces sessions the platform believes are broken, or servers that never terminate. Getting this right requires dedicated testing, and maintaining it means the Agones lifecycle is in scope for every future game server build.
Fleet configuration. Studios define and tune their own ArmadaSets — the scaling unit that determines how many servers sit ready between matches. To avoid cold starts, you pre-provision a standing fleet. Those idle servers accrue compute cost between sessions. Gameye bills per second on active sessions only.
Matchmaker glue code. Each matchmaker integration is engineering work your team builds and owns indefinitely. For FlexMatch, that means writing your own AWS Lambda + SNS pipeline from scratch. For Pragma and Nakama, there is no documented GameFabric path at all.
The sum of these — SDK integration, fleet tuning, and custom matchmaker connectors — represents a real engineering investment that doesn’t appear in a per-vCPU comparison. It’s the cost that outlasts the migration.
Where Gameye fits
Gameye is a game server orchestration platform purpose-built for multiplayer studios, with over 120 million sessions orchestrated since 2019 for titles including Chivalry 2 and games from Bungie.
For migrating studios, Gameye offers:
- Native matchmaker integrations — Pragma Engine and Nakama work out of the box. FlexMatch is available; contact us for setup details.
- Provider-agnostic orchestration — runs across cloud and bare metal, not locked to a single provider.
- Transparent pricing — $0.07/vCPU/hr, no egress fees, 99.99% SLA. Numbers you can model before you commit.
- Fast onboarding — sandbox access provisioned within 24 hours. You can be running sessions and validating performance before your next sprint ends.
For Pragma studios specifically, Gameye is an official Capacity Provider within Pragma’s own codebase. Migration means updating a config, not rebuilding your matchmaking architecture.
The broader signal
These shutdowns are a reminder that game infrastructure is a specialised domain. It rewards platforms that are fully committed to it. When infrastructure providers pivot — toward AI, other markets, or acquisition — studios left behind pay the price in emergency migrations, technical debt, and player-facing disruption.
This is why Gameye exists. Multiplayer game server orchestration is not a feature we offer. It’s the only thing we do.
If your studio is affected by either shutdown, book a free migration call — we’ll map your current setup to a Gameye environment and tell you honestly whether it’s a good fit.