OpenAI Launches MRC Network Protocol With AMD, Intel, NVIDIA; Supports 100,000+ GPUs

According to OpenAI's announcement on May 6, the company partnered with AMD, Broadcom, Intel, Microsoft, and NVIDIA to launch Multipath Reliable Connection (MRC), an open network protocol for large-scale AI training cluster GPU interconnection. The protocol splits single data transmissions across hundreds of pathways to reduce core congestion and enables microsecond-level bypass of link and switch failures.

OpenAI has already deployed MRC in its Stargate supercomputer (built with OCI) and Microsoft's Fairwater supercomputer, allowing connection of over 100,000 GPUs with just two switch layers while reducing power consumption and hardware requirements. The MRC specification has been released to the industry through the Open Compute Project.