The Model Context Protocol (MCP) has recorded approximately 97 million monthly SDK downloads in March 2026, cementing its position as the dominant infrastructure layer for agentic artificial intelligence applications. This open standard, released by Anthropic as open source on November 25, 2024, has grown from roughly 100,000 monthly downloads at launch to nearly a thousandfold increase in 16 months.
Standardization Sweeps Across AI and Crypto
MCP defines how AI models connect to external tools, databases, APIs, and workflows. A single MCP server can serve any compatible client—including Claude, ChatGPT, Gemini, Microsoft Copilot, Cursor, and VS Code Copilot—without requiring custom code per model. OpenAI added full MCP support to ChatGPT and its Agents SDK in March 2025, a turning point analysts say accelerated adoption. Google, Microsoft, AWS, and dozens of other platforms followed by mid-2025.
In December 2025, Anthropic donated MCP to the newly formed Agentic AI Foundation (AAIF) under the Linux Foundation, with OpenAI and Block as co-founders. Platinum members include AWS, Google, Microsoft, Cloudflare, GitHub, and Bloomberg. This vendor-neutral, community-governed structure mirrors that of Kubernetes and PyTorch.
As of March 2026, over 10,000 active MCP servers are in public and enterprise deployments. The crypto sector has been quick to build MCP infrastructure: BitGo launched an official MCP server in March 2026, enabling AI tools and dev environments to interact with the institutional digital asset custody platform via natural language. Coinbase released a Payments MCP via its Developer Platform in late 2025, linking AI agents to crypto wallets, on-ramps, and stablecoin transactions. Crypto.com published a Market Data MCP server delivering live quotes, order books, and candlestick charts. CoinGecko launched its own server providing real-time data for over 15,000 cryptocurrencies and 1,000+ exchanges. The cross-chain protocol deBridge implemented an MCP server in February 2026, enabling non-custodial swaps and bridging between EVM chains and Solana.
From API Fragmentation to Network Effects
Previously, building separate connectors for each AI platform was costly and inefficient. With MCP, companies publish a single server and gain instant interoperability with every major AI client—a network effect proprietary integration approaches cannot match. Security researchers have noted that many public MCP servers have not undergone formal audits. While the Linux Foundation governance enforces standardized authentication and transport requirements, implementation-level security remains the responsibility of individual server operators.
The MCP specification, SDKs, and reference implementations are available at github.com/modelcontextprotocol, with core documentation at modelcontextprotocol.io.
AI-Crypto Convergence Enters New Phase
The explosive growth of MCP signifies that the fusion of AI and crypto infrastructure is moving from concept to reality. When AI agents can seamlessly call BitGo’s custody, Coinbase’s payment rails, CoinGecko’s market data, and deBridge’s cross-chain routing through a single protocol, decentralized finance (DeFi) automation and intelligence enter a new paradigm. Crypto firms are racing to secure their position in this emerging infrastructure layer, betting that MCP will be the USB-C of the AI-DeFi stack.