Google Launches Managed Remote MCP Servers for Cloud

Google announced the introduction of fully managed remote Model Context Protocol (MCP) servers for its cloud services last year, December 10, 2024. This release significantly shortens the time needed to integrate AI with actual company data and tools by enabling artificial intelligence agents to safely connect to Google Cloud services, such as BigQuery, Google Maps, Compute Engine, and Kubernetes Engine, via a single open standard.

This shows the recent discovery by Google for large language models (LLMs) to communicate with third-party services, the Model Context Protocol (MCP) has emerged as the de facto standard. In order for this to happen, Google says that the third-party services must have an MCP server that the LLM can connect to. Even though MCP is just a year old, almost all online services now have an MCP server. For several of its Google and Google Cloud services, Google is now expanding its MCP support by providing fully managed MCP servers, with many more to come.

Up until now, developers were forced to use Google’s community-built local servers or implement open source solutions, which were frequently challenging to set up and maintain, as Google points out. Google is eliminating all of this extra work by providing these new managed MCP servers, which makes it much simpler for a developer to create an agent that is supported by Google’s new Gemini 3 model, for instance, and that connects to Google Maps and the BigQuery data service to add geospatial information to this data.

We are making sure that developers and agents can effortlessly engage with data and take action with these expanded and new MCP capabilities. In today’s announcement, Michael Bachman, Google’s VP and GM of Google Cloud, and Anna Berenberg, Google Cloud engineering fellow, write, “Google is dedicated to leading the AI revolution not just by building the best models but also by building the best ecosystem for those models and agents to thrive.”

For the MCP servers are first being released by Google for four services:

• Google Maps: “Grounding Lite” data is available. This makes it possible for agents to respond to questions about location, weather, and routing.
• BigQuery: Enables agents to natively run SQL queries on enterprise datasets and read schemas.
• Google Compute Engine (GCE): Facilitates automated infrastructure operations by exposing provisioning and scaling tasks.
• Google Kubernetes Engine (GKE): Provides cluster management resources with organised access.

Also the Google Compute Engine (GCE), which will enable agents to manage infrastructure workflows and provision and resize instances, for instance, and the Google Kubernetes Engine (GKE), which will provide a server that enables agents to work with the Kubernetes and GKE APIs, are among the first services to receive the new managed MCP servers, along with BigQuery and Maps.

The Core Features in which the Managed MCP Server will offer are the managed remote servers, in contrast to local MCP servers, are hosted on Google’s infrastructure. They can be accessible through public HTTP endpoints like bigquery.googleapis.com/mcp. 

• Managed Transcoding: By using Apigee, businesses may make their own unique APIs discoverable tools for AI agents without changing the code that already exists.
• Unified Access Layer: This will offers a globally uniform endpoint that is compatible with external hosts like VS Code or Claude Code as well as typical MCP clients like Gemini CLI and AI Studio.
• Enterprise Security: Its a Cloud IAM and this is integrated with servers to provide authorisation and authentication. Model Armour provides further protection by looking for agent-specific dangers in prompts and responses.
• Instant Setup: When entering a URL to a managed endpoint, developers can establish a connection between agents. As a result, setup time is decreased.

In the upcoming months, the following more services will also receive MCP support:

• Compute, storage, and projects: Cloud Run, Cloud Storage, Cloud Resource Manager
• Analytics and Databases: Dataplex Universal Catalogue, AlloyDB, Cloud SQL, Spanner, Looker, and Pub/Sub
• Google Security Operations (SecOps) is responsible for security.
• Cloud operations: cloud monitoring, cloud logging
• Google services: Android Management API, Developer Knowledge API

Google is leveraging the Cloud API Registry, which is still in preview, to find, control, utilise, and keep an eye on these services. The Registry will give the agents all the information about the tools and APIs that are available to them, as well as the capabilities of these tools that they can utilise.

The statement was made today, one day after Anthropic gave the Linux Foundation the MCP protocol specifications as part of the new, vendor-neutral Agentic AI Foundation (AAIF).

“More developers will be able to create agentic AI applications thanks to Google’s support for MCP across such a wide range of products and their close collaboration on the specification,” stated David Soria Parra, an Anthropic technical team member and co-creator of MCP. “We are getting closer to agentic AI that functions seamlessly across the tools and services people already use as adoption increases among leading platforms.”

Extended MCP support is also being added to Google’s Apigee API management service, enabling businesses to manage their custom MCP servers and how agents access them using the same tool they now use to manage their APIs.

The Google Cloud documentation will also provides developers with implementation details for MCP server management. To know about this visit, here.