Anthropic's Model Context Protocol crossed 97 million monthly SDK downloads in March 2026, cementing its transition from experimental standard to the foundational infrastructure layer for agentic AI. The growth trajectory is extraordinary: from approximately 2 million downloads at launch in November 2024, to 8 million by January 2025, 22 million by April 2025, and now 97 million — a 4,750% increase in 16 months. For context, the React npm package took approximately three years to reach 100 million monthly downloads.
The adoption is no longer optional. Every major AI provider now ships MCP-compatible tooling as a default integration mechanism. OpenAI adopted it in Q2 2025, Microsoft Copilot in Q3 2025, and Google DeepMind and Amazon Bedrock followed in Q4 2025. The ecosystem has grown to over 5,800 community and enterprise MCP servers spanning developer tools (1,200+), business applications (950+), web and search (600+), and AI automation (450+) — giving agents standardised access to databases, CRMs, cloud providers, and productivity tools.
The protocol's governance evolved in December 2025 when Anthropic donated MCP to the Linux Foundation, establishing it as a vendor-neutral standard. This move was critical for enterprise adoption — organisations that were hesitant to build infrastructure around a single company's protocol could now treat MCP as an industry standard with shared governance and long-term stability guarantees.
For context engineers, MCP's dominance has practical implications for every project. Building an AI agent in 2026 without MCP support is like building a web application without HTTP — technically possible but practically pointless. The protocol defines how agents connect to external tools and data sources, and understanding MCP server architecture is becoming as fundamental as understanding REST APIs was a decade ago.