Every Major Cloud Now Has Official MCP Servers — and It Changes Everything

AI agents managing cloud infrastructure via MCP natural language commands

MCP Goes Enterprise

Anthropic's Model Context Protocol (MCP) — dubbed the "USB-C for AI" — has moved well beyond AI startups. Every major cloud provider now has official MCP servers, letting AI agents manage cloud infrastructure using nothing more than natural language commands. Cursor, VS Code, Claude, Codex, and GitHub Copilot all support MCP out of the box.

Here's a breakdown of what each hyperscaler is offering.

AWS — 60+ Official MCP Servers

Amazon Web Services leads the pack with a suite of over 60 official MCP servers spanning the entire AWS product catalog — documentation, infrastructure provisioning, containers, Lambda, AI/ML, data analytics, messaging, cost analysis, and more.

The general-purpose AWS MCP Server is the flagship. It connects agents to the latest docs, API references, and standard operating procedures (SOPs) for multi-step workflows. A prompt like "Investigate increased 5xx errors in prod over the last 30 minutes" will trigger the server to pull metrics, logs, and config data across services to surface a root cause.

AWS is also migrating toward Streamable HTTP, an improved transport protocol. The investment is significant and clearly long-term.

Microsoft Azure — 40+ MCP Tools

Azure's MCP Server breaks into 40+ individual tools covering best practices, AI/ML, analytics, compute, containers, databases, DevOps, IoT, and storage. Natural language prompts like "Show me all my resource groups" or "List blobs in my storage container named 'documents'" work out of the box.

Azure provides particularly clear documentation with hand-holding for setup, tool parameters, and controls to enable or disable agent access to sensitive functions.

Google Cloud — 4 Servers (Preview)

Google Cloud announced official MCP servers in December 2025. Currently in preview, GCP offers four remote servers covering BigQuery, Compute Engine, Google Kubernetes Engine (GKE), and Google Security Operations.

Prompts like "Kill my running VM in project 0009 in the east zone" invoke the stop_instance tool directly. A standout feature: Google logs all MCP interactions for audit purposes — useful for cloud administrators who need compliance trails.

Oracle — OCI, Databases, MySQL

Oracle brings MCP to its established enterprise platforms — Oracle Cloud Infrastructure (OCI), Oracle Database (via SQLcl), and MySQL. Prompts like "Connect to my side project and tell me about what kind of data I have there" return saved Oracle connections and schema descriptions in plain language. Some of Oracle's MCP work remains in proof-of-concept phase, but the direction is clear.

IBM Cloud — Experimental but Promising

IBM's MCP servers are experimental, functioning as a knowledge-gathering layer between AI assistants and IBM Cloud environments. Unlike the others, IBM's Core MCP Server runs locally rather than as a remote server — an interesting architectural choice that may appeal to enterprises with strict data residency requirements.

The Bottom Line

MCP just went enterprise. When AWS, Azure, Google Cloud, Oracle, and IBM all ship official MCP servers within months of each other, it stops being a trend and starts being infrastructure. The ability to manage cloud operations — spin up VMs, query databases, investigate production errors — via natural language commands is no longer a demo. It's production. The agent-forward cloud era has started.