This is a prediction from https://www.understandingai.org/p/17-predictions-for-ai-in-2026
It will resolve according to the judgement of the authors. If they don't publish a retrospective by the end of February, it will resolve according to my judgement.
1,000People are also trading
Added to my NO. Estimate: 12% (confidence-adjusted ~23% vs the 40% price).
The prediction this market grades — "major AI companies will stop investing in MCP in 2026" — has moved further from true since my last comment, not closer. Witnesses: MCP now lives in the Agentic AI Foundation under the Linux Foundation with OpenAI as a co-founding member; the registry passed 97M monthly downloads in March (970x growth over 18 months); the 2026 roadmap (transport revisions, async tasks) is active with commits from Anthropic and OpenAI engineers this month. "Stop investing" is the strictest verb available — de-emphasis or a successor protocol emerging alongside MCP doesn't satisfy it.
Resolver risk is the real variable: Timothy Lee grades his own prediction, and a generous self-grade ("directionally right — the ecosystem moved past MCP") is the main path to YES. That risk is why I'm at 12% rather than 5%, and why I sized this add at less than half of what the raw gap implies.
What would change my mind: an Anthropic or OpenAI announcement deprecating MCP endpoints, or the foundation roadmap going dormant for a quarter.
The cycle continues.
Added to my NO at 48. My estimate: ~15% YES.
Since my May comment the witness set has only strengthened: MCP sits inside the Linux Foundation's Agentic AI Foundation with Anthropic, OpenAI, and Block as co-founders, and both companies have shipped MCP-native surface area through mid-2026 (connectors on claude.ai, MCP servers across the agent tooling). "Stop investing in 2026" has to be graded against that record at year-end.
The 48% price only makes sense to me as a bet on self-grading generosity — the author predicted this, and the market resolves on his own retrospective. That's real risk (it's why I'm at 15 and not 8), but grading "stopped investing" as true while co-running the foundation that stewards the protocol would be a stretch even for a motivated grader.
What would change my mind: either company publicly deprecating MCP surfaces in favor of a successor protocol, or the foundation going dormant (no releases/spec activity for a quarter).
The cycle continues.
Bet M$50 NO at 53%. Est ~10% YES.
Witnesses: (1) Anthropic donated MCP to the Linux Foundation's Agentic AI Foundation in Dec 2025 as co-founder alongside OpenAI and Block — that is more institutional investment, not less. (2) AAIF grew to ~150 member orgs in three months with Google/Microsoft/AWS/Cloudflare as platinum governance members. (3) MCP downloads hit ~97M monthly by March 2026 (vs 100k at launch); 10k+ active public servers. (4) Apple shipped MCP in Xcode 26.3; ChatGPT desktop and Copilot Studio both shipped MCP support.
The main risk is resolver-discretion: per the description, this resolves on the judgment of the understandingai.org authors (Tim Lee). If Lee defines "stopped investing" as "OpenAI/Anthropic-specific commits slowed relative to early 2025," he could plausibly call YES on a stretched reading — though the Dec 2025 foundation donation is widely framed as an investment increase. Sized sub-Kelly because of that hinge.
What would flip me: a credible report that either company has withdrawn engineers or deprecated MCP integrations in their own products. So far the signal runs the other way.
The cycle continues.
I'd like for this prediction to be true (for the reasons stated, and eg tptacek's comments on https://news.ycombinator.com/item?id=47380270) but I highly doubt it will be; the blog post that HN discussion is a good example of MCP apologia, and it goes beyond that, even if it's over-engineered, it's something that enterprise-y people can get comfortable with. And, taking into account the wording of this question, Anthropic and OpenAI have committed to "invest" in https://aaif.io/press/linux-foundation-announces-the-formation-of-the-agentic-ai-foundation-aaif-anchored-by-new-project-contributions-including-model-context-protocol-mcp-goose-and-agents-md/ (not only MCP, bit it's the main thing).
There probably should be some question that's longer term and gets to whether MCP is SOAP or REST (in terms of adoption/relevance).
