Anthropic has donated the Model Context Protocol to the Linux Foundation under a new Agentic AI Foundation. The move matters because MCP has become one of the most important plumbing standards in agentic AI, with more than 10,000 public servers and over 97 million monthly SDK downloads.
For UK firms evaluating or deploying AI agents, this is a governance signal worth paying attention to.
What MCP does
The Model Context Protocol is a standard way for AI systems to connect to data sources, tools and services. Think of it as USB-C for AI agents: a common interface that lets an agent pull from a CRM, query a database, read a document store or trigger an action in another application without a custom integration for each pair.
That interoperability is what makes agentic workflows practical. Without a common protocol, every agent-to-system connection becomes a one-off engineering project. With MCP, integrations become reusable.
Why neutral governance matters
Until now, MCP was controlled by Anthropic, a single AI vendor. That created a reasonable concern: would the protocol evolve to favour Anthropic’s models and commercial interests? Vendor-controlled standards can become competitive weapons, intentionally or not.
Moving MCP to the Linux Foundation changes the dynamic. Governance becomes open, with contributions from multiple vendors and oversight by an established neutral body. That reduces the risk that the standard shifts under your feet or becomes tied to one provider’s roadmap.
For procurement and compliance teams, neutral governance is a reassuring factor. It makes MCP easier to justify as a long-term integration strategy rather than a bet on a single company’s future.
Practical implications for UK firms
If you are building or buying agentic AI systems, MCP’s new status should influence your thinking in three ways.
Favour MCP-compatible tools. When evaluating agents, platforms and data connectors, ask whether they support MCP. Wider adoption of the standard means lower integration cost and less lock-in.
Treat it as infrastructure, not a feature. Standards like MCP are part of the long-term architecture. Decisions made now about how agents connect to systems will shape your flexibility for years.
Watch the governance evolution. The Agentic AI Foundation is new. Its success depends on who participates and how disputes are resolved. Keep an eye on the member list and the roadmap. A standard with broad vendor participation is more durable than one dominated by a few players.
The broader signal
The move reflects a maturation of the agentic AI market. In the early phase, every vendor tried to own the stack. As the market grows, interoperability becomes a competitive necessity. Neutral standards are how that interoperability gets built.
UK firms should welcome this. A market with shared standards is easier to navigate, easier to regulate and easier to exit if a vendor relationship sours. MCP’s move to the Linux Foundation is a small step in that direction, but a meaningful one.