Unity Multiplay Shut Down: Where Studios Actually Went (May 2026 Verdict)

Unity Multiplay Game Server Hosting shut down March 31, 2026. Two months on, the dust is settling and the migration picture is clearer than it was during the panic windows of January and February. This is what studios actually did, what's working, and the parts the official migration docs gloss over.

If you're reading this and you still haven't migrated, you're not alone. A significant number of studios are still in limbo. The good news: every alternative below has at least one Multiplay refugee in production now. The bad news: the talent crunch on fleet-orchestration engineers gets worse every week.

What actually shut down

Unity sunsetted Multiplay Game Server Hosting - the orchestration service that allocated and scaled dedicated game server VMs. The following Unity services are not affected:

• Unity Matchmaker - still alive, still routes players to matches.
• Unity Relay - still alive, still relays packets between peers.
• Unity Lobby - still alive, still manages pre-match grouping.
• Distributed Authority - still alive, still works for P2P-style sessions without dedicated servers.

What changed: the dedicated server hosting layer. If your game shipped a server binary that Multiplay allocated on demand and scaled with your CCU curve, that's the layer you have to replace.

The four real alternatives, with field notes

Option	Best for	Pricing model	Migration friction
Rocket Science Group	AAA continuity, existing Multiplay tooling	Not yet published	Low (same codebase)
Edgegap	Edge latency, Unity-first studios	$0.00115 / vCPU-min + $0.10/GB egress	Low (Unity plugin + docs)
Gameye	Predictable costs, no SDK lock-in	$0.07 / vCPU-hr on-demand, $0.025 reserved, egress included	Medium (strip Multiplay SDK, no replacement SDK needed)
Crux	Indie / mid-size with auth + leaderboards needs	Flat tiered, free up to 2K MAU	Medium (server hosting + backend in one)

Rocket Science Group: the continuity bet

Rocket Science Group is the answer if your engineering team's main constraint is not rewriting all our integration code. The pitch is concrete: same codebase, same engineering DNA, same SDK shapes. Unity licensed the software to a team founded by the original Multiplay engineers, and continuity is the whole product.

The risks are commercial, not technical. No public pricing as of May 2026. Studios that have signed are quoted privately and the terms vary by scale. The formal service launch is targeted for late Q1 2026 with SOC 2 Type II and ISO/IEC 27001 certifications coming. If you need a contract on paper this week, you are negotiating with a small team carrying the operational load of every Multiplay refugee in parallel.

Pick Rocket Science if: you have an AAA-scale live game, you cannot afford engineering time to swap fleet-management code, and you can absorb pricing uncertainty for the first contract year. Skip it if: you wanted Multiplay because it was a Unity-bundled SaaS and you need a published price card to budget your year.

Edgegap: the edge-latency bet

Edgegap operates 615+ locations globally and ships a Unity plugin with a documented Multiplay migration path. The pricing is straightforward: $0.00115 per vCPU-minute on the dedicated tier and $0.10/GB of egress. You can fraction down to 1/4 vCPU if your server binary fits, which puts the floor at $0.0002875/min for thrifty deployments.

The Unity plugin is the killer feature for Multiplay refugees. The migration playbook is: strip the Multiplay SDK, drop in Edgegap's Unity plugin, point your build pipeline at their image registry, and update your fleet-allocation calls. Two to three engineering weeks for most teams.

Pick Edgegap if: your game cares about edge latency and you want to keep Unity as your first-class SDK target. Skip it if: egress fees on your traffic curve make the per-GB pricing scary (run the calculator first).

Gameye: the cost-predictability bet

Gameye has been orchestrating dedicated game servers since 2017 and lists 120M+ sessions on their counter. They predate Unity's entry into the space, so they're not a startup scrambling to absorb a refugee wave.

Two pricing tracks: $0.07 per vCPU-hour on-demand, $0.025 per vCPU-hour reserved. The headline feature: egress is included. No per-GB bandwidth surprise on your invoice. Their cost calculator vs Edgegap and AWS shows the gap most clearly on high-bandwidth games (FPS, battle royale) where egress dominates compute.

Gameye also doesn't require an SDK in your server binary. You remove the Multiplay SDK and you don't replace it with anything. The orchestration happens around your container, not through it. This is the cleanest migration shape if you want to leave Unity-specific SDK lock-in behind.

Pick Gameye if: your accounting team wants a single number per vCPU-hour and your network engineer doesn't want to count gigabytes. Skip it if: you need a Unity-native plugin (you'll be working with their Docker container interface instead).

Crux: the unified-backend bet

Crux is the option that's not on most Multiplay migration lists, and that's the angle. The other three options replace the server hosting layer and leave you to integrate your auth, leaderboards, progression, and live ops separately. Crux ships them as one platform with one flat-rate price.

For an indie or mid-size studio that was using Multiplay for hosting plus rolling their own backend for player data, this consolidates the bill and the on-call rotation. The free tier covers up to 2,000 monthly active users and 2,000,000 API calls - enough to get a live game in front of real players before any commercial commitment.

Pick Crux if: you're indie or mid-size, you were stitching Multiplay together with a separate Firebase / Supabase / hand-rolled backend, and you'd rather buy one platform than four. Skip it if: you're AAA-scale and your existing backend is a strategic moat you don't want to swap.

The migration playbook (what your engineers actually need to do)

1. Inventory your Multiplay integration surface. Which SDK calls allocate servers? Which webhooks fire on session start/end? What does your build pipeline push to the Multiplay image registry?
2. Pick a target alternative and a hedge. Most studios are signing with one primary and quietly piloting a secondary. The hedge isn't paranoia; it's the lesson from Hathora's May 5 2026 shutdown - providers can vanish overnight.
3. Rewrite the fleet-management glue. This is where the engineering weeks go. The SDK calls change shape, the webhook payloads differ, and your build pipeline points at a new image registry.
4. Update your live ops dashboard. Whoever watches server health day-to-day needs the new provider's dashboard wired into your incident response. Don't ship migration without runbook updates.
5. Run a 1% canary. Route a tiny slice of matchmaking traffic to the new provider and watch latency, allocation success rate, and player drop rate for a week before flipping the fleet.
6. Keep your old contract live during the canary. If you've signed with Rocket Science Group, your existing Multiplay servers may still be online during the transition. Don't tear down until the canary is convincing.

The Hathora warning

Two months after Multiplay shut down, a second major dedicated game server provider - Hathora - was acquired by Fireworks AI and permanently shut down on May 5, 2026. Hathora customers were offboarded to Nitrado via GameFabric. Games that used Hathora for online infrastructure (Stormgate, Splitgate 2, Predecessor) had to rush migration or lose online modes.

The lesson: vendor concentration in dedicated game server hosting is a real business risk. The right answer in May 2026 isn't to pick the cheapest provider; it's to pick a provider whose business model isn't optimized for an AI inference pivot. Game server orchestration is a margin-thin commodity layer, and the providers most exposed to the next acquisition are the ones running closest to break-even on game traffic.

What we're recommending in May 2026

If you're not yet committed and you want a recommendation rather than a comparison table:

• AAA continuity, 50K+ CCU peak: Sign with Rocket Science Group, negotiate hard on the SLA, keep a Gameye account warm as the hedge.
• Edge-latency-critical (FPS, battle royale, shooter): Edgegap primary, Gameye hedge.
• Cost-conscious mid-size studio: Gameye primary, Edgegap hedge.
• Indie / launching a multiplayer game in the next 12 months: Crux primary. The free tier covers your soft launch, the unified billing simplifies your year, and you skip the four-vendor backend integration project.

Whatever you pick, the worst answer is to wait. Every additional week of delay is another week of your engineers competing with everyone else's engineers for the same finite pool of fleet-orchestration talent.

Related reading

• AWS GameLift vs Agones (Kubernetes) - if you're considering self-hosting on Kubernetes instead of a managed provider.
• PlayFab Alternative: Crux vs Microsoft PlayFab - if your Multiplay setup also relied on PlayFab and you're rethinking the backend layer.
• Game Server Backend Platform Buyer's Guide - the broader buying framework if you want to compare beyond the Multiplay refugee shortlist.
• Cost calculator - model your CCU curve against Crux's flat-rate pricing.