AccelByte · Cloudflare
AccelByte ships "everything your game needs behind the scenes" — Foundations, Online, Multiplayer — for Krafton, Dreamhaven, Theorycraft, and AEXLAB across PC, console, mobile and VR. Every backend call is a player-perceived latency event, and Cloudflare's 330+ POPs are where AAA backends already live.

One stack for any game.
One edge for every player.

The pitch on accelbyte.io is clear: 80% of the backend every game needs, on day one. Identity, matchmaking, multiplayer servers, store, wallet, inventory, leaderboards, season pass — modular, customizable, battle-tested at scale.

3 packages
Foundations · Online · Multiplayer
46%
AEXLAB server-cost reduction on VAIL VR
2h → 20m
Striking Distance crash-resolution with ADT MCP

Three things converge in the AccelByte stack that map cleanly onto Cloudflare developer primitives — the kind of fit that's hard to find when most platforms ship for a single workload type:

GA · June 2026 Your AI bill is out of control. Cloudflare can fix it now.
AI Gateway ships dollar-denominated spend limits, identity-driven budgets, per-customer attribution, and PII redaction — across every model provider, in one log line. For the ADT MCP Server path (already cutting Striking Distance's crash resolution from 2hrs to 20mins), this is the cost-control layer that scales the AI value without scaling the AI bill. Read the announcement.

Three primitives that map directly to how AccelByte actually runs:

Workers + Durable Objects — for matchmaking, parties, presence, and game-session state at the player's nearest POP. Sub-150ms turn handling for any region. Replaces the central-region latency tax most BaaS platforms inherit from their hyperscaler.
R2 + Stream — for ADT Build Distribution and Crash Reporting artifacts. Zero-egress storage for builds delivered globally, with Stream for the playtest video uploads that come back.
AI Gateway + Workers AI — under the ADT MCP Server. Per-studio cost attribution, fallback routing, PII redaction on crash dumps. The runtime piece that makes the *"AI cuts crash resolution 6x"* story defensible at every studio you onboard, not just Striking Distance.

One question for the AccelByte platform team:

Is the bigger near-term pain on the multiplayer-edge side — pushing matchmaking and session state closer to the player than a hyperscaler region allows — or on the AI-economics side — keeping the ADT MCP Server's inference costs predictable as more studios adopt it? 20 minutes to find the right starting point.

Deeper Dive

The full architecture, ready when you are

A detailed primitive-by-primitive mapping is in progress — including the Foundations/Online/Multiplayer × Cloudflare primitive table, the request-flow diagram for a matchmaking call on Cloudflare's edge, the ADT MCP Server economics math, and the 90-day rollout. Tell me which slice would be most useful first.

Request the expanded version →
Grab 20 minutes →

One observation worth flagging: accelbyte.io is already fronted by Cloudflare today (HubSpot CMS behind our edge). So this isn't an "add a vendor" conversation. It's a "turn on what's already in the stack" conversation.

Matt Holscher Cloudflare · Developer Platform