- Amazon is building managed MCP servers and has opened a beta for AI agents to interact with its advertising platform through the protocol.
- AWS engineer Clare Liguori is actively shaping the Model Context Protocol spec as Amazon tests draft MCP concepts in production.
- Anthropic recently donated MCP to the Linux Foundation’s new Agentic AI Foundation, cementing it as the industry standard for agent-to-tool connections.
The protocol war for agentic AI appears to be over—and Amazon just placed a very large bet on the winner. The tech giant is building managed Model Context Protocol servers and has opened a beta allowing AI agents to interact with its advertising platform through MCP, The New Stack reported Wednesday.
For the uninitiated, MCP is the connective tissue that lets AI agents talk to external tools and services. Think of it as the USB port for artificial intelligence—a standardized way for an agent to read your calendar, query a database, or, in Amazon’s case, manage advertising campaigns. Anthropic created the protocol last year, and adoption has been explosive.
Why MCP Matters for Amazon’s Advertising Empire
Amazon’s advertising business generated $17.3 billion in Q4 2025 alone, making it the third-largest digital ad platform behind Google and Meta. Opening that revenue engine to AI agents through MCP isn’t a science experiment—it’s a strategic move to ensure that as companies deploy agentic AI for marketing, Amazon’s ad platform is the first one those agents learn to use.
The implementation is being led by AWS engineer Clare Liguori, who is actively contributing to the MCP specification while Amazon builds managed MCP servers for its cloud customers. The company is reportedly testing draft MCP concepts in production—a sign that Amazon isn’t waiting for the spec to be finalized before deploying it at scale.
Amazon Ads has already opened a beta for its new MCP server, giving developers and AI platforms a structured way to programmatically create, manage, and optimize ad campaigns. The beta enables agents to handle everything from bid adjustments to creative testing without human intervention.
The broader MCP ecosystem is also hardening into an industry standard. Anthropic recently donated the protocol to the Linux Foundation’s newly formed Agentic AI Foundation, alongside projects like Goose and AGENTS.md. The move signals that MCP has transcended its origins as an Anthropic project and become infrastructure that no single company controls—which is exactly what Amazon needs before building critical business operations on top of it.
The practical upshot for developers is straightforward. If you’re building an AI agent that needs to interact with Amazon’s services—ads, product catalogs, AWS resources—MCP is now the supported integration path. No custom APIs, no proprietary SDKs. The same protocol that connects an agent to a Slack workspace or a GitHub repository now connects it to Amazon’s $17 billion ad machine.
As of Wednesday, the Agentic AI Foundation counts contributions from Amazon, Anthropic, Google, Microsoft, and over 40 other organizations—making MCP the most broadly supported protocol in the agentic AI stack.
