From Anthropic Project to Linux Foundation Standard: How MCP Left Vendor Control

MCP went from an internal Anthropic project to a vendor-neutral standard governed by a Linux Foundation body in about 13 months: open-sourced November 25, 2024, donated to the Agentic AI Foundation (AAIF) on December 9, 2025. This page traces why Anthropic gave up unilateral control, who adopted the protocol along the way, and what actually changes in governance versus what stays the same.

Open-source and vendor-neutral by design, not as an afterthought

Anthropic released MCP as an open standard on November 25, 2024, explicitly to solve the "M x N integration problem": every AI application needing custom code for every data source it connects to, producing information silos (What Is the Model Context Protocol (MCP): the Open Standard Connecting AI Models to Tools and Data). — https://www.anthropic.com/news/model-context-protocol

From the outset Anthropic committed to keeping MCP "neutral, open, and community-driven," rejecting proprietary control; this framing was later reaffirmed by Anthropic itself as the reasoning behind the December 2025 donation. — https://www.anthropic.com/news/donating-the-model-context-protocol-and-establishing-of-the-agentic-ai-foundation

GitHub's engineering blog notes that shared stewardship — not Claude-only usage — was central to MCP's original vision: the protocol only reaches its standardization goal if the entire AI ecosystem adopts it, not just Anthropic's own products. — https://github.blog/open-source/maintainers/mcp-joins-the-linux-foundation-what-this-means-for-developers-building-the-next-era-of-ai-tools-and-agents/

The 2025 adoption race: OpenAI, Google DeepMind, Microsoft

Once open-sourced, MCP was picked up rapidly by Anthropic's own competitors:

- March 2025: OpenAI integrated MCP across its Agents SDK, Responses API, and ChatGPT desktop; Sam Altman: "People love MCP and we are excited to add support across our products." — https://www.pento.ai/blog/a-year-of-mcp-2025-review - April 2025: Google DeepMind CEO Demis Hassabis confirmed MCP support in upcoming Gemini models, calling it "a good protocol and it's rapidly becoming an open standard for the AI agentic era." — https://techcrunch.com/2025/04/09/google-says-itll-embrace-anthropics-standard-for-connecting-ai-models-to-data/ - May 2025: VS Code announced native MCP support in GitHub Copilot Agent Mode. — https://guptadeepak.com/the-complete-guide-to-model-context-protocol-mcp-enterprise-adoption-market-trends-and-implementation-strategies/ - September 2025: OpenAI added MCP support to ChatGPT apps; the MCP Registry launched and grew to nearly 2,000 server entries within months (MCP Registry and Server Discovery — how a developer finds, publishes, and vets a server before trusting it enough to connect a client). — https://www.pento.ai/blog/a-year-of-mcp-2025-review - By December 2025: Microsoft shipped native MCP support across GitHub, Copilot Studio, Dynamics 365, Azure AI Foundry, Semantic Kernel, and Windows 11. — https://devblogs.microsoft.com/semantic-kernel/ - Google's Gemini API managed agents gained the ability to connect directly to remote MCP servers, reaching private databases and internal APIs. — https://blog.google/innovation-and-ai/technology/developers-tools/expanding-managed-agents-gemini-api/ (dated July 7, 2026)

Why this matters for governance

by late 2025, MCP's biggest users were Anthropic's direct competitors — a structural argument for moving control out of Anthropic's hands (MCP Adoption and the July 2026 Spec — what changed under you while you weren't looking).

December 9, 2025: MCP donated to the Agentic AI Foundation

On December 9, 2025, Anthropic donated MCP to the Agentic AI Foundation (AAIF), a newly formed directed fund under the Linux Foundation. — https://www.linuxfoundation.org/press/linux-foundation-announces-the-formation-of-the-agentic-ai-foundation

MCP became a founding project of AAIF, alongside goose (built by Block) and AGENTS.md (built by OpenAI). — https://blog.modelcontextprotocol.io/posts/2025-12-09-mcp-joins-agentic-ai-foundation/

Scale at time of donation

97+ million monthly SDK downloads across Python and TypeScript, more than 10,000 active public servers, and adoption across ChatGPT, Claude, Cursor, Gemini, Microsoft Copilot, and Visual Studio Code. — https://www.anthropic.com/news/donating-the-model-context-protocol-and-establishing-of-the-agentic-ai-foundation

The Linux Foundation was chosen to provide "the neutral home" for MCP, with open governance structures modeled on those used for Kubernetes and PyTorch. — https://github.blog/open-source/maintainers/mcp-joins-the-linux-foundation-what-this-means-for-developers-building-the-next-era-of-ai-tools-and-agents/

Founding and supporting members

- Platinum members (co-founders and lead supporters): Anthropic, Block, OpenAI, AWS, Google, Microsoft, Cloudflare, and Bloomberg. — https://www.linuxfoundation.org/press/linux-foundation-announces-the-formation-of-the-agentic-ai-foundation - Gold members: 18 total, including Adyen, Docker, IBM, JetBrains, Oracle, Salesforce, Snowflake, and Twilio. — https://www.linuxfoundation.org/press/linux-foundation-announces-the-formation-of-the-agentic-ai-foundation - Silver members: 23 total — Apify, Chronosphere, Cosmonic, Elasticsearch, Eve Security, Hugging Face, Kubermatic, KYXStart, LanceDB, Mirantis, NinjaTech AI, Obot.ai, Prefect.io, Pydantic, Shinkai.com, Solo.io, Spectro Cloud, Stacklok, SUSE, Uber, WorkOS, Zapier, and ZED. — https://www.linuxfoundation.org/press/linux-foundation-announces-the-formation-of-the-agentic-ai-foundation - GitHub is named as a supporting member in reporting, but (unverified) whether it holds a distinct governance seat or is folded into an existing Platinum member's representation (e.g. Microsoft) is not established in the sources read for this page.

Changes

- Control is no longer centralized at Anthropic. Previously Anthropic had unilateral authority over the spec; governance is now distributed under AAIF. — https://github.blog/open-source/maintainers/mcp-joins-the-linux-foundation-what-this-means-for-developers-building-the-next-era-of-ai-tools-and-agents/ - Specification Enhancement Proposals (SEPs) are the formal mechanism for changing MCP, modeled on Python PEPs and Rust RFCs; no single member, including Anthropic, can bypass the process. — https://modelcontextprotocol.io/community/governance - Roadmap ownership is now consensus-based, set by technical steering committees; no member has unilateral control over direction. — https://techcrunch.com/2025/12/09/openai-anthropic-and-block-join-new-linux-foundation-effort-to-standardize-the-ai-agent-era/ - Funding does not equal control: AAIF runs on membership dues from Platinum/Gold/Silver tiers, but dues don't grant veto power over technical decisions. — https://techcrunch.com/2025/12/09/openai-anthropic-and-block-join-new-linux-foundation-effort-to-standardize-the-ai-agent-era/

Stays the same

- The existing three-tier maintainer hierarchy (Contributors → Maintainers → Core Maintainers → Lead Maintainers) continues under Linux Foundation stewardship; Core Maintainers still meet bi-weekly to review SEPs with public notes. — https://modelcontextprotocol.io/community/governance - Current Lead Maintainers David Soria Parra and Den Delimarsky (Origins and Development of the Model Context Protocol: From Internal Frustration to Public Launch) retain their roles; Core Maintainers include Peter Alexander, Caitie McCaffrey, Kurtis Van Gent, Clare Liguori, Paul Carleton, Nick Cooper, and Nick Aldridge. — https://modelcontextprotocol.io/community/governance - The Linux Foundation itself provides infrastructure and neutrality, not technical control — actual technical decisions stay with the MCP Steering Group. — https://blog.modelcontextprotocol.io/posts/2025-12-09-mcp-joins-agentic-ai-foundation/

Why it matters

GitHub's analysis frames the donation as trading Anthropic's product-strategy interests for long-term ecosystem stability, reducing the risk enterprises take on when they integrate MCP deeply (MCP Limits: Registry Trust and Supply Chain Risks — why 'it's in the registry' is not a safety guarantee before you connect a client covers a separate, still-open risk category: registry vetting). — https://github.blog/open-source/maintainers/mcp-joins-the-linux-foundation-what-this-means-for-developers-building-the-next-era-of-ai-tools-and-agents/