“At OpenAI, we utilize an unsharded architecture with one writer and multiple readers, demonstrating that PostgreSQL can scale gracefully under massive read loads.”
OpenAI runs most of its critical systems on one unsharded Postgres primary and dozens of read replicas. It works because reads dominate. The single writer is a single point of failure, so they push writes elsewhere and refuse to add new services to it. One SEV0 in nine months, and Microsoft gets an Azure case study out of it.