Case file — 50B69695
The idea
“MCP server marketplace — developers publish MCP servers, Claude/Cursor users discover and install them in one click. Takes 15% of revenue on paid servers. Similar to how Shopify App Store works but for AI tool integrations. MCP adoption is growing 40% month-over-month.”
The panel
Your idea is already executed. MCP Market and MCP Marketplace both exist and are live—the first shows 29,642 servers indexed, the second explicitly positions itself as "where developers publish, distribute, and monetize MCP servers" with payments and discovery handled. You're entering a market with entrenched players offering identical value propositions at launch. The red flag: 40% MoM growth sounds strong but masks a critical unknown—total addressable user base and whether that growth sustains post-hype. MCP is still niche; you'd be taking 15% from an unproven monetization layer atop an immature protocol. Your genuine advantage: neither existing marketplace appears heavily VC-funded or corporate-locked based on the live data. If you move fast with superior UX, creator tools, or revenue-share terms, you could capture early mindshare before the market consolidates. But you're not first—you're third or fourth.
Technical Reality Check: Your core underestimation: MCP server standardization is still in flux. Anthropic controls the spec evolution, meaning your marketplace could become obsolete overnight if they change versioning, authentication, or deployment models. You're building on shifting sand. Build-vs-buy trap: Don't build payment processing or subscription management. Stripe or Gumroad handle this; your edge is curation and discovery, not infrastructure. No moat exists yet. Any player with Anthropic relationships (including Anthropic itself) could launch this tomorrow. First-mover advantage evaporates if the network doesn't lock in fast. What's solid: One-click installation is genuinely hard and genuinely valuable. If you nail the UX for dependency resolution and version compatibility, that's defensible for 12-18 months. The real risk? You need MCP adoption to stay outside Anthropic's direct control. If they integrate a marketplace natively, you're finished. Validate this dependency risk before building anything.
The brutal math: You're pre-revenue with zero users. At 40% MoM growth, you need the marketplace to reach critical mass before the unit economics matter—but you have no path to that yet. Your real CAC problem: convincing developers to publish (supply-side chicken-egg), then users to discover (demand-side chicken-egg). You're solving neither. Pricing delusion: 15% take rate assumes developers won't build their own distribution or use competing platforms. Once you have traction, they'll negotiate hard or leave. Runway cliff: You'll burn through capital on product and marketing for 12–18 months before you know if either side shows up. No paying customers = no leverage to extend. What works: MCP is genuinely early-stage infrastructure. If adoption sustains and you move fast, you capture a defensible layer before competitors notice. First-mover advantage in platform picks matters here.
Timing verdict: Late, but not dead. MCP servers exist in a pre-standard phase—Anthropic hasn't blessed a canonical marketplace, and adoption is still niche developer territory. You're entering after early movers have already built integrations directly. The 40% MoM growth is real but off a tiny base; you need critical mass of both supply and demand that doesn't exist yet. Macro factor: Anthropic's product roadmap. If they launch native marketplace features in Claude or formalize MCP standards, you're obsolete. If they stay hands-off, fragmentation creates opportunity—but only if you move before they decide. Window status: Closing in 12 months. Either Anthropic centralizes distribution or the ecosystem settles into direct integrations. You have maybe two release cycles to become the de facto standard. One genuine advantage: Cursor's rapid user growth (and independence from Anthropic) creates a parallel demand side you can exploit faster than Anthropic can respond. Build for Cursor first.
Competitors found during analysis
Live dataMCP Market
29,642 servers indexed, live
MCP Marketplace
Official monetization platform, live
RapidAPI
Existing API marketplace mentioned as precedent
Cause of death
You're not first, you're not second — you're a latecomer with no differentiator
MCP Market has 29,642 servers indexed. MCP Marketplace is already live with payments and distribution baked in. You have an idea and zero users. "Better UX" is not a strategy — it's what every third entrant says before they discover that the first mover already iterated twice. You need a reason to exist that isn't "we'll do it better," and right now you don't have one.
Anthropic is your landlord, your regulator, and your potential competitor — simultaneously
The MCP spec is in flux and Anthropic controls its evolution. They could change versioning, authentication, or deployment models and render your entire catalog incompatible overnight. Worse, they could launch native marketplace features inside Claude — which would be the rational move once the ecosystem reaches meaningful scale. You're building a storefront on land you don't own, with a landlord who's also in the retail business.
The two-sided marketplace cold start problem with no capital or traction
You need developers to publish (why would they choose you over existing marketplaces or direct distribution?). You need users to discover (why would they come to you instead of searching within Claude or Cursor directly?). You have zero of either. The finance panel is right: you'll burn 12-18 months solving this chicken-and-egg problem, and the 15% take rate gives developers every incentive to leave the moment they have any leverage. Marketplace businesses are capital-intensive precisely because you have to subsidize one side — and you haven't even thought about which side or how.
⚠ Blind spot
You're treating "MCP adoption growing 40% MoM" as validation of your marketplace, but it's actually evidence against it. Rapid protocol adoption means developers are already finding and distributing MCP servers without a centralized marketplace. The ecosystem is growing through GitHub repos, direct integrations, and word of mouth. Your implicit assumption — that discovery is a bottleneck — may simply be wrong. If developers can share MCP servers via a config file and a URL, the marketplace layer you're building is a convenience, not a necessity. Conveniences don't command 15% take rates.
What would need to be true
Anthropic must not launch a native MCP marketplace inside Claude within the next 18 months — if they do, every third-party marketplace becomes a redundant middleman overnight.
MCP server quality and security must become a real pain point as adoption scales — if the ecosystem stays small and trust stays informal, there's no demand for curation or verification at a price anyone will pay.
At least one major IDE or AI client (Cursor being the best candidate) must be willing to integrate your registry as a default or preferred source — without platform-level distribution, you're just another website competing with GitHub search.
Recommended intervention
Don't build a marketplace. Build a trust and security layer for Cursor specifically. The timing panel flagged it: Cursor's rapid, independent growth creates a demand side that Anthropic doesn't control. Cursor users are installing MCP servers from random GitHub repos with no verification, no sandboxing, and no compatibility guarantees. Build a verified registry — think npm meets the Chrome Web Store — that Cursor integrates natively. Charge enterprises for verified/audited server packages (security compliance is a real budget line item), not individual developers 15% of their hobby project revenue. This gives you a wedge that existing marketplaces haven't claimed, a customer (Cursor + enterprise teams) who actually has money, and a moat (trust/verification) that's genuinely hard to replicate. Go talk to the Cursor team this week.
Intervention unlocking
5seconds
No account needed. One email, no follow-ups.
Want your idea examined? Free triage or full panel →