Mesh LLM: distributed AI computing on iroh
Iroh ships Mesh LLM — distributed AI inference over its peer-to-peer stack (266 HN points).
Why it matters
- First serious peer-to-peer distributed-inference open-source implementation to hit meaningful HN traction.
- Direct signal for Delangue's 'companies done renting AI' thesis and Sunrun's home-data-center pitch.
- Sets a reference architecture for distributed inference beyond hyperscaler datacenters.
- Complements Ollama's $65M revenue-side signal on local inference.