MCP Explained: The Protocol Anthropic Gave Away, and Why Everyone Uses It Now

If your AI assistant can now read your Google Drive, post to Slack, or pull a ticket out of Linear, the plumbing behind that is likely the Model Context Protocol, or MCP — and if your assistant's quietly gotten more capable this past year with no clear explanation, this is a big part of it. Anthropic introduced it in November 2024 as a narrow piece of infrastructure — a standard way for an AI model to talk to outside tools and data. In December 2025, Anthropic donated it to a new Linux Foundation initiative, handing shared control to the same companies it competes with, and why says a lot about what MCP actually is.

The problem MCP was built to solve

Before MCP, every AI product that wanted to connect to, say, GitHub or a database had to build a custom, one-off integration. Multiply that by every assistant and every tool, and you get an N×M mess: every provider re-implementing the same connections with no shared standard. MCP's pitch was simple — one protocol for "here's how a model asks a tool for information or asks it to do something," that any AI product can speak to any supporting tool. People compare it to a USB-C port for AI: instead of a different cable for every device, one plug everything agrees on.

How fast this actually moved

The adoption curve is unusually fast for a technical standard. Anthropic shipped MCP in November 2024; by March 2025 OpenAI announced support across the Agents SDK, ChatGPT's desktop app, and the Responses API — notable, since OpenAI was adopting a rival's standard. Google DeepMind followed in April, shipping native Gemini support that May; Microsoft, GitHub, Cursor, and VS Code added support around the same stretch. By the December 2025 donation, Anthropic reported over 97 million monthly SDK downloads and roughly 10,000 active public servers.

Why Anthropic gave it away

Anthropic donated MCP's governance to the Agentic AI Foundation, a new directed fund under the Linux Foundation, co-founded alongside Block and OpenAI, with Google, Microsoft, AWS, Cloudflare, and Bloomberg also backing it. Anthropic said its own involvement wasn't changing — still investing and maintaining core infrastructure — but decision-making no longer sits with one company. That matters because a standard controlled by a competitor is a harder sell to everyone building on top of it; once rivals share the same governance structure, no single vendor can quietly bend the protocol toward its own product.

What this actually changes for you

Most people won't ever touch MCP directly — it's a protocol, not a product. But it's why the "connectors" menu in whatever AI assistant you use has gotten so much longer without each one needing a special app. If your assistant can search your company's wiki or query a database, someone stood up an MCP server for that source. The upside: adding a capability increasingly looks like installing a browser extension rather than waiting for a vendor to build a bespoke integration.

Where to be careful

An MCP server is still software with access to your data or accounts, and connecting to one is closer to installing a browser extension than downloading a PDF — it deserves the same scrutiny. Prefer servers published by the tool vendor itself over unofficial community ones unless you've checked who maintains them, and pay attention to exactly what permissions a server requests. The protocol's own 2026 roadmap acknowledges this is unfinished business: maintainers have flagged audit trails, single sign-on, and more consistent gateway authentication as priority work, since enterprises adopting MCP at scale need better guarantees than "trust the server."

The bigger point

What's notable about MCP isn't the technical design — plenty of integration protocols existed before it. It's that a standard built by one AI lab got adopted by its direct rivals fast enough, and became load-bearing enough, that the lab building it decided sole ownership was a liability rather than an advantage. That's a rare sequence in software, and a decent signal that MCP is less a feature of any one AI product and more a permanent layer underneath all of them — worth knowing the way you'd know what HTTP does, even without ever writing a request by hand.

Sources: Anthropic's donation announcement, MCP blog on the Agentic AI Foundation, TechCrunch on OpenAI adopting MCP, TechCrunch on Google adopting MCP, MCP's 2026 roadmap.