Anthropic Built MCP, Got Everyone to Use It, Then Gave It Away

Anthropic announced it’s donating the Model Context Protocol to the Agentic AI Foundation, a directed fund under the Linux Foundation. The founding members are Anthropic, Block, and OpenAI — three companies that compete directly — along with Google, Microsoft, AWS, and Cloudflare as supporting members.

That combination tells you something important.

MCP is the protocol that lets AI assistants connect to external tools and data sources — your calendar, your database, your code editor, your APIs. It’s what makes Claude or Copilot or Cursor actually useful inside a real environment instead of just answering questions in a chat window. Anthropic shipped it in late 2024. It now has 10,000+ active public servers and runs inside ChatGPT, Gemini, VS Code, and Microsoft Copilot.

This is the USB moment.

USB didn’t belong to Intel after it got adopted. It became plumbing — something the industry builds on without worrying about who controls it. When a protocol reaches that status, the vendor that created it stops being a dependency and starts being a contributor. The governance moves from one company’s roadmap to a standards body where no single player can unilaterally change direction.

For anyone building on MCP — connecting AI to production systems, building agents, writing MCP servers — the biggest risk was always single-vendor control. Anthropic could have changed the spec, deprecated tools, or steered the protocol toward their own commercial interests. That risk is now substantially smaller.

Competing companies agreeing on shared governance doesn’t happen unless all of them have concluded that the cost of fragmentation is higher than the cost of cooperation. They’ve concluded that. MCP stopped being Anthropic’s thing the moment these companies agreed to govern it together — it’s the industry’s plumbing now.

Build on it accordingly.