If you've published an MCP server under MIT, this is worth ten minutes of your time.
Directories like Glama and Smithery index thousands of free, open-source MCP servers, then sell hosting access on top of them — $9 to $80/month tiers, enterprise SaaS pricing, the works. The developer who wrote the server usually sees none of it. Not because anyone's breaking a rule — MIT grants unrestricted commercial reuse and guarantees zero royalties. It's working exactly as licensed. It's just not working for the person who wrote the code.
Some numbers, pulled from actual market research rather than platform marketing pages:
The MCP market is projected to grow from $1.28B (2025) to $28.46B (2035) — a 37.22% CAGR (DataM Intelligence)
Gateway and integration platforms — Glama's whole category — already hold 32.47% of that market
Fewer than 5% of all listed MCP servers are monetized on any platform today
Not every platform is broken, though. Apify pays developers 80% revenue share. MCPize pays 80–85%. MCP Marketplace pays 85%. And x402 — the HTTP-native micropayment standard for agent-to-tool billing — just moved under the Linux Foundation with Visa, Stripe, Google, AWS, and Microsoft as founding members. AWS Bedrock AgentCore already integrates it natively. That's the first real infrastructure letting an MCP developer get paid per call with no directory taking a cut in the middle.
I wrote up the full platform-by-platform comparison — who pays, who doesn't, what the eligibility bar looks like to get monetized, and what changes ethically once your MCP server is actually making money instead of being a portfolio piece: 👉 https://www.seosiri.com/2026/09/mcp-monetization-gap-glama-smithery-developer-losses.html
Curious where this community lands: is the MIT license itself the problem here, or is it purely a platform-ethics gap that better licenses (or things like x402) can route around? Have any of you actually gotten paid for an MCP server you shipped?
