Google is rolling out fully managed Model Context Protocol servers across its cloud platform, removing the need for developers to build and maintain their own connectors. The launch, announced December 10, covers four services at start: Google Maps, BigQuery, Compute Engine, and Kubernetes Engine.
The setup is simple. Developers paste a URL into their agent configuration and start making calls. No local infrastructure, no credential wrangling, no version mismatches. Steren Giannini, product management director at Google Cloud, told TechCrunch the process cuts setup time from a week or two down to minutes.
Each server targets a specific use case. Maps Grounding Lite feeds agents real-time location, routing, and weather data. BigQuery lets agents run SQL queries against enterprise data warehouses without loading records into context windows. Compute Engine and GKE servers expose infrastructure controls so agents can spin up VMs, resize clusters, or diagnose pod failures on their own.
Google is also opening Apigee, its API management product, to the MCP standard. Companies can convert existing internal APIs into discoverable agent tools with their current security policies and rate limits intact. Access controls run through Cloud IAM, and Google's Model Armor provides a defense layer against prompt injection and data exfiltration.
The servers are live now in public preview at no extra cost for existing customers. General availability is expected early next year, with new service integrations dropping weekly. Cloud Storage, AlloyDB, Cloud SQL, Spanner, Looker, and security operations tools are all on the roadmap.
The Bottom Line: Google is betting that managed MCP infrastructure will become the default way enterprises connect AI agents to production systems, and it wants that traffic running through its endpoints.
QUICK FACTS
- Launch date: December 10, 2025
- Initial services: Maps, BigQuery, Compute Engine, Kubernetes Engine
- Pricing: Free in public preview; included in existing Cloud billing at GA
- Expansion: 15+ services planned, including AlloyDB, Cloud SQL, Spanner, Looker
- Security: Cloud IAM, audit logging, Model Armor for prompt injection defense
