Multiplay

What Multiplay actually is

Multiplay is one of the oldest names in game server hosting — the company has run game servers since 1997, starting inside the UK LAN-festival scene (Multiplay (UK) Ltd, incorporated May 1997, founder Craig Fletcher) and growing into dedicated hosting for studios and publishers. The corporate trail is long: sold to retailer GAME in 2015; the digital hosting half went to Unity Technologies in November 2017 for £19 million; it ran as Unity Multiplay for several years; and at the end of 2025 Rocket Science Group — a company founded by Multiplay veterans, including the engineer behind the original platform code — bought it back and now runs it independently. Unity deprecated its own Multiplay platform in April 2026; the multiplay.com of today is the Rocket Science Group operation, relaunched publicly on 14 July 2026.

What it sells is not a game server you can rent from a sign-up form. Multiplay is an orchestration layer and managed service for live multiplayer games: you bring a build (Unity, Unreal, or custom engine), and Multiplay provisions VMs, containers, and bare metal across its fleet, applies OS, network, and security policies, handles patching and scaling, and runs live operations with a named engineer attached to your studio. The fleet model is per-region: builds, build configurations, and fleets, with independent scaling settings per region. It is the infrastructure behind AAA launches — the company cites titles like Titanfall 2, Rocket League, and Killing Floor among its hosted games — and it sits in our game server hosting category as the enterprise, hands-on end of the market.

The distinctive idea is the hybrid model, which Multiplay says it pioneered in 2016: steady-state load runs on owned bare metal at predictable cost, and cloud capacity across multiple providers absorbs launch spikes — "steady-state on bare metal, cloud only for the burst." The pitch on the platform page is a cost curve: all-cloud wastes money on elastic rates for capacity that never moves; all-bare-metal means owning hardware sized for your rarest peak; the hybrid mix lands at the bottom. In our view that argument is correct economics for live-service games with spiky demand, and it is the main technical reason to look at Multiplay rather than bolting servers together yourself.

Pricing: what you really pay

You will not find a price. As of late 2026 Multiplay publishes no pricing on its website or documentation: the commercial model is described as usage-based, onboarding is sales-led (access begins with a contact conversation), and no self-serve sign-up or free trial is offered. There are two engagement tiers — Assisted Integration, where your studio owns the build and Rocket Science co-designs it, and Fully Managed, where the sister studio Terminal Velocity takes the engineering end to end — and the cost of the managed model is set by contract. We have seen no credible published figures to quote, so we will not invent any.

What we can tell you is how to think about the cost. Multiplay's own argument is that the hybrid model lowers total spend versus all-cloud, and independent analysis in the game hosting market puts raw cloud game-server compute far above cheap bare metal — one 2026 industry breakdown estimates roughly four figures per month for a thousand concurrent players on cloud instances alone, before egress. If Multiplay's orchestration runs your baseline on owned metal, the savings can be material at scale. But at enterprise-contract pricing with a managed team attached, this is a product for studios with real revenue and real launch risk, not for a weekend prototype. Before signing, ask for: the per-hour or per-session rate card, egress terms, what the 24/7 live-ops service covers versus what is billed as professional services, and the exit path for your builds and data.

Performance & infrastructure

The footprint is global and hybrid: Multiplay reports more than 30 locations across six regions — North America, Europe, Asia, LATAM, MEA, and Oceania — on owned bare metal and multi-cloud, with redundancy at the region level and a stated target of sub-30ms latency to major population centers. One thing we want to flag honestly: the numbers in public sources do not line up cleanly. The platform page says "30-plus locations"; a third-party comparison notes the same page itemises 40 (16 North America, 9 Europe, 7 Asia, 3 LATAM, 3 MEA, 2 Oceania), and Rocket Science Group's own site cites 40+ across "five clouds and bare metal." Multiplay does not publish which locations are owned bare metal and which are cloud burst capacity. We record 30 as the vendor's own headline figure and treat the larger count as unconfirmed.

On reliability engineering, the platform story is strong where it counts for live games: automated scaling (proactive and reactive), fleet health monitoring with drift detection, zero-downtime patching, allocation tracing, crash reporting and logs, and single-click migration of a game between data centres or clouds if a site or provider degrades. The company's history backs the claims operationally — its about page describes an unannounced free-to-play title in 2019 that scaled to tens of millions of players within weeks, an order of magnitude above prediction, without turning players away. That is the kind of launch record studios are actually buying.

Security posture is documented on a public security page: an information security programme built on SOC 2 criteria, independent third-party assessments, at least annual penetration testing, least-privilege access, SSO and 2FA where available, and vendor risk management. DDoS is addressed at the platform level (the platform page lists it explicitly within its security properties), and Multiplay's own guidance stresses protocol-aware filtering for UDP game traffic rather than generic web scrubbing — though, as with the footprint, the specific mitigation commitments are not published in detail. Note also what is absent: no public uptime SLA figure exists as of mid-2026 (Unity's old platform advertised 99.95%, but that platform is deprecated and the figure does not carry over).

Where it falls short

Transparency, full stop. No pricing, no free trial, no self-serve onboarding, no published SLA figure, no clear bare-metal-versus-cloud breakdown of the fleet. Every serious buyer ends up in a sales process to learn the basics. For an enterprise product that is common, but it means we cannot verify price-performance for you, and comparisons with usage-priced competitors that publish rates are guesswork.

The sales-led model also filters the audience. If you want to spin up a server tonight for a community event or an indie title, Multiplay is the wrong door — you will not get an account without a conversation, and the engagement tiers assume studio-scale integration work. Individual gamers and small communities are explicitly outside the target.

There is also concentration risk in the new arrangement: Rocket Science Group is a small, veteran-founded company that bought the platform back from Unity at the end of 2025. The team is deep in this specific domain — arguably the deepest in the market — but the financial and organisational weight behind the contract is not what Unity's was. For a live-service game with a five-year horizon, we would want to understand their capacity commitments and continuity planning in detail. Finally, the opaque footprint cuts both ways operationally: if you cannot see which regions are owned metal versus burst cloud, you cannot fully model latency or failure behavior yourself.

Who should pick this

We pick Multiplay for studios running live-service or AAA multiplayer titles with spiky demand, launch-day scale risk, and the budget for a managed relationship: teams who want a named engineer, 24/7 live operations, and an orchestration layer that has carried unannounced launches before. It is especially compelling when your traffic profile is exactly what the hybrid model targets — predictable baseline with sharp peaks — and when multi-engine support (Unity, Unreal, custom) matters because your portfolio is mixed.

We would not pick it for indie or community-scale projects, for teams that want transparent rate cards and self-serve provisioning, or for anyone who needs contractual uptime guarantees in writing before signing — ask, but know that nothing is published. Alternatives in our game server hosting category include usage-priced orchestration platforms that publish their rates and general-purpose VPS and bare-metal providers for teams willing to own the orchestration themselves. If Multiplay is on your shortlist, the diligence questions above are the ones we would walk into the sales call with.

Key takeaways

• Game server hosting veteran: running servers since 1997; sold to GAME (2015), Unity (2017, £19m), and bought back by founders' new Rocket Science Group in late 2025; independent since, public relaunch July 2026.
• Managed orchestration, not self-serve hosting: builds, fleets, per-region scaling, patching, and 24/7 live ops with a named engineer; two engagement tiers including fully managed via sister studio Terminal Velocity.
• Hybrid model is the core idea and it is sound economics: owned bare metal for steady state, multi-cloud for bursts.
• Footprint of 30+ locations across six regions (public counts conflict up to 40+); sub-30ms population-center latency target; no published uptime SLA.
• SOC 2-criteria security programme, annual third-party pen tests, platform-level DDoS handling — specifics not published.
• No published pricing, no free trial, sales-led onboarding: enterprise budgets and studio-scale games only.

Sources

• https://multiplay.com/about-us
• https://multiplay.com/platform
• https://www.multiplay.com/security-1
• https://edgegap.com/comparison/edgegap-vs-unity-multiplay
• https://en.wikipedia.org/wiki/Multiplay
• https://pcgamesinsider.biz/profile/66474/why-did-unity-buy-multiplays-digital-business-for-19m
• https://itechguides.com/products/multiplay-game-server-orchestration-platform
• https://techyorker.com/compare/multiplay-game-server-orchestration-platform-vs-accelbyte-multiplayer-servers
• https://multiplay.com/multiplay-blog/8-questions-for-dedicated-game-server-hosting-in-2026
• https://multiplayguide.com/blog/dedicated-server-economics-multiplayer