Google Cloud API Gateway adds native MCP server support

On October 5, 2026, the Google Developers Blog announced that Google Cloud API Gateway can now act as a native remote MCP server. According to the post, this removes the need to build and maintain custom middleware just to expose REST APIs to AI agents.

The approach builds on specifications developers already have. Working from an existing OpenAPI 3.x specification, developers add specific annotations — the post gives x-google-api-management.mcp as an example — and standard REST operations become tools that AI agents can discover and call. The gateway then handles the protocol gap on its own: incoming MCP JSON-RPC requests are automatically translated into REST calls, so no intermediate layer has to be written, hosted, or operated by the developer.

By presenting this as native support inside the gateway rather than a separate service or sidecar, the post describes the resulting endpoints as discoverable and agent-ready. In practical terms, teams that already publish REST APIs described by an OpenAPI 3.x spec would not have to rewrite those services or stand up a parallel MCP layer; the annotation is added to the specification that already documents the API.

The announcement, as received, does not list pricing, regional availability, or a separate rollout schedule for the capability. The excerpt also cuts off mid-sentence, so the material does not make clear whether the feature has reached general availability or is still in preview. That is where the matter stands in the single source currently available.