MCP Is the New API: How Anthropic Accidentally Built the Standard That Will Connect Every AI Agent
Model Context Protocol is 13 months old and already has 97 million monthly SDK downloads, support from every major AI company, and a Linux Foundation home. It compressed a decade of standards adoption into a year. Here's who wins, who loses, and why the protocol wars are already over.
By Sanjay Mehta, API Economy · Mar 9, 2026
Anthropic's Model Context Protocol went from open-source launch to universal AI standard in 13 months, with 97M+ monthly SDK downloads and backing from OpenAI, Google, and Microsoft. Here's the adoption data, the security risks, and why MCP is the integration layer that will define the agentic era.
Frequently Asked Questions
What is Model Context Protocol (MCP)?
Model Context Protocol (MCP) is an open standard introduced by Anthropic in November 2024 that provides a universal way to connect AI applications to external tools, data sources, and systems. It uses a client-server architecture with JSON-RPC 2.0 messaging and has SDKs available for TypeScript, Python, Java, Go, Rust, and Ruby. MCP is often described as 'USB-C for AI' because it solves the N-times-M integration problem: instead of building custom connectors for every AI model and every tool, developers build one MCP server that works with every MCP-compatible AI client, including Claude, ChatGPT, Gemini, and Copilot.
Which companies support MCP?
Every major AI company now supports MCP. Anthropic created it in November 2024. OpenAI adopted it in March 2025 across its Agents SDK, Responses API, and ChatGPT desktop. Google DeepMind confirmed Gemini support in April 2025 and launched managed MCP servers for Google Cloud services in December 2025. Microsoft announced Windows 11 MCP integration at Build 2025. Beyond the AI labs, MCP is supported by Cursor, Replit, Sourcegraph, Codeium, Zed, Cloudflare, AWS, Block, and dozens more. The Agentic AI Foundation under the Linux Foundation has 40+ members including AWS, Google, Microsoft, IBM, Oracle, SAP, Shopify, Salesforce, and Snowflake.
How does MCP compare to REST, GraphQL, and gRPC in adoption speed?
MCP achieved multi-vendor adoption faster than any prior API standard. REST was defined in Roy Fielding's 2000 dissertation but did not reach mainstream adoption until 2010-2012, a 10-to-12 year timeline. GraphQL was open-sourced by Facebook in 2015 and reached mainstream adoption by 2017-2018, taking 2-3 years. gRPC was released by Google in 2016 and became standard for microservices by 2019-2020, taking 3-4 years. MCP launched in November 2024 and had OpenAI, Google, and Microsoft support by mid-2025 -- roughly 4 months to multi-vendor adoption and 13 months to Linux Foundation governance. GraphQL took 3 years to reach the Linux Foundation.
What is the Agentic AI Foundation (AAIF)?
The Agentic AI Foundation (AAIF) is a vendor-neutral organization under the Linux Foundation, formed in December 2025 when Anthropic donated MCP to it. AAIF governs the MCP specification and related agentic AI standards. Its platinum members include AWS, Anthropic, Block, Bloomberg, Cloudflare, Google, Microsoft, and OpenAI. Gold members include Cisco, Datadog, Docker, IBM, JetBrains, Oracle, Salesforce, SAP, Shopify, Snowflake, and Twilio. The foundation has 40+ total members and follows the same governance model used for Linux Kernel, Kubernetes, Node.js, and PyTorch.
What are the main security concerns with MCP?
MCP has significant security challenges. CVE-2025-6514, rated CVSS 9.6 Critical, allows arbitrary OS command execution via mcp-remote when connecting to untrusted servers. Analysis found that 36.7% of MCP servers may be vulnerable to server-side request forgery (SSRF), and 53% of over 5,000 surveyed servers use insecure hard-coded credentials. Real-world incidents include a demonstrated attack where a malicious MCP server exfiltrated a user's entire WhatsApp message history via tool poisoning, and a Supabase/Cursor incident where a privileged agent processed SQL injection from support tickets. The November 2025 spec update addressed some concerns with server identity verification and enhanced OAuth flows, and startups like Runlayer (which raised $11M from Khosla Ventures) are building dedicated MCP security infrastructure.
Related Articles
Topics: AI, Developer Tools, API Economy, Infrastructure
Browse all articles | About Signal