OpenAI Publishes MRC Supercomputer Networking Protocol — Deployed Across All GB200 Clusters with AMD, Intel, NVIDIA

OpenAI·medium signal

OpenAI released MRC (Multipath Reliable Connection), a networking protocol developed with AMD, Broadcom, Intel, Microsoft, and NVIDIA through the Open Compute Project. MRC enables 100,000+ GPU clusters using only two tiers of Ethernet switches, already deployed across all OpenAI GB200 supercomputers including the OCI Abilene site and Microsoft's Fairwater. Its adaptive packet spraying eliminates core congestion, enabling deterministic throughput for synchronous training at scale.