OpenAI Opens MRC Networking Protocol for AI Clusters
OpenAI partnered with AMD, Broadcom, Intel, Microsoft, and NVIDIA to release MRC — Multipath Reliable Connection — an open networking protocol designed to make large AI training clusters run faster and more reliably with less wasted GPU time. MRC is already deployed across OpenAI's largest supercomputers, including Oracle Cloud Infrastructure in Abilene, Texas and Microsoft's Fairwater supercomputers. It is now available through the Open Compute Project for industry-wide adoption.
Why It Matters
Opening a core training-infrastructure protocol through the Open Compute Project is a rare industry-alignment move from OpenAI, signaling that interconnect efficiency — not just chip performance — is a binding constraint in the compute race that benefits from shared standardization.