OpenRelay

Category

AI & Machine Learning

Launched

August 7, 2026

Pricing

freemium

Website

openrelay.inc

OpenRelay wants to be the Akamai of AI inference. The company, backed by Y Combinator, calls itself an "Inference Delivery Network" — a phrase that deliberately echoes the content delivery networks that made the web fast and resilient. But instead of caching static files at the edge, OpenRelay routes LLM inference requests across a distributed mesh of GPUs, from consumer RTX 4090s to enterprise H100s. The pitch is simple: one API, every chip, automatic failover. The ambition is large, but the idea is not entirely new. GPU marketplaces like Vast.ai and RunPod have offered access to distributed GPUs for years. What OpenRelay claims to add is a network layer that treats node failures as routine, not exceptional. In a world where centralized clouds treat a single failed node as an outage, OpenRelay's architecture is designed to absorb failures before users notice. Whether that's enough to displace the hyperscalers remains an open question.