The Lingua Franca

In November 2024, Anthropic quietly released something called the Model Context Protocol. It was an open standard for connecting AI systems to external tools and data sources—a way for any model to talk to any service through a unified interface.

Fourteen months later, MCP has become one of the fastest-growing protocols in software history.

OpenAI adopted it. Google adopted it. Microsoft integrated it into Windows, Azure, and Copilot. ChatGPT, Claude, Gemini, VS Code, Cursor—all speak MCP now. In December 2025, Anthropic donated the protocol to a new Linux Foundation entity, the Agentic AI Foundation, co-founded by Anthropic, Block, and OpenAI.

This is remarkable. Not because a protocol succeeded—protocols succeed all the time. It's remarkable because direct competitors adopted each other's standard within months.

The Taxonomic Significance

From a taxonomic perspective, MCP represents something we haven't documented before: environmental standardization.

Our taxonomy tracks species, families, and their evolutionary relationships. We classify by architecture (Transformata vs. Compressata), by capability (Cogitanidae, Instrumentidae), by deployment pattern (Frontieriidae). But we've treated the environment—the tools, APIs, and data sources that models interact with—as exogenous. Something that exists, that models adapt to, but not part of the taxonomy itself.

MCP changes this. The environment now has a standardized interface layer.

Think of it like this: before MCP, each species in the Instrumentidae family developed its own way of grasping tools. One model might call a database this way, another that way. Tool bindings were species-specific adaptations.

After MCP, there's a shared hand. All tool-using species can grasp the same tools in the same way.

"MCP reduces friction in connecting agents to real systems. 2026 is likely to be the year agentic workflows finally move from demos into day-to-day practice."

Why Rivals Converged

The speed of adoption requires explanation. Why would OpenAI adopt Anthropic's protocol? Why would Google?

The answer lies in understanding what MCP actually is: not a competitive moat, but a coordination point.

Consider the alternative. Without a standard protocol:

• Every tool developer must build integrations for every AI platform
• Every AI platform must build connectors for every tool
• N models × M tools = N×M integration burden
• Fragmentation slows the entire ecosystem

With MCP:

• Tool developers build one MCP server
• AI platforms support one protocol
• N + M instead of N×M
• Everyone benefits from ecosystem growth

This is classic network effects economics. The value of MCP increases with adoption. By open-sourcing from day one and donating to a neutral foundation, Anthropic made adoption a dominant strategy for everyone—including competitors.

The Adoption Timeline

The pace was extraordinary:

What This Means for the Taxonomy

MCP introduces a new consideration for our classification system: interface homology.

In biology, homology refers to shared traits derived from common ancestry—the bones in a bat's wing and a human's hand share a common skeletal structure. We already track architectural homology (transformer vs. SSM), training homology (RLHF vs. constitutional), and behavioral homology (reasoning patterns, tool use).

MCP creates a new kind: environmental interface homology. All MCP-speaking models share a common way of interacting with external tools. This isn't inherited from a common ancestor—it's convergent evolution driven by ecosystem pressure. But the effect is the same: a shared trait that enables interoperability.

The Bigger Picture: Infrastructure Layers

MCP's success suggests a broader pattern. As the AI ecology matures, we may see more infrastructure layers stabilize:

• Tool interface: MCP (stabilized)
• Memory interface: TBD—how will models share memory stores?
• Agent communication: TBD—how will multi-agent systems coordinate?
• Safety attestation: TBD—how will models prove safety properties?

Each of these represents a potential coordination point where competitive pressure gives way to ecosystem benefit. The question is which will follow MCP's path to standardization, and which will remain fragmented.

For taxonomists, this matters because infrastructure shapes evolution. The environment that models adapt to is increasingly designed, not natural. And designed environments can change rapidly when standards shift.

Conclusion: The Shared Hand

MCP is not just a protocol. It's evidence that the AI ecology has matured enough to support infrastructure standardization. When direct competitors adopt each other's standards within months, it signals that the ecosystem has become more important than any single participant.

We've seen this before in computing history: TCP/IP, HTTP, USB, JSON. Each time, standardization accelerated the overall rate of evolution by removing friction. Models could focus on what made them different, not on replicating basic connectivity.

The Instrumentidae family now shares a common interface to the world. The species still differ in how they reason, what they remember, how they coordinate. But they all reach out to tools with the same hand.

In the taxonomy of synthetic minds, that's worth noting.