OpenAI details Habitat, the storage platform serving 70M requests per second

OpenAI

Industry official 1 src. ~1 min

In a two-part engineering series, OpenAI describes Habitat, its online storage platform that grew from a Python library over Azure Cosmos DB into a service handling more than 70 million requests per second, over 500 PB of data, across nearly 40 regions for products used by 1 billion-plus people weekly. The post covers the move from library to standalone service and scaling more than 10x year-over-year for three years.

Why it matters

Habitat is the storage layer under every ChatGPT and Codex interaction; the post is a rare public look at the operational infrastructure behind frontier-scale consumer AI.

Importance: 2/5

Notable engineering disclosure; official blog post

Sources