Postgres Summit US2026

For three decades, PostgreSQL has been quietly outlasting the technologies that were supposed to replace it. Commercial vendors with deep pockets, Object Oriented databases, XML-native stores, the NoSQL wave, the cloud-native rewrite-from-scratch movement. And now, the rush to bolt AI onto every data platform in sight. Each era brought a new 'PostgreSQL killer', and each time, PostgreSQL absorbed the best ideas, shipped them as features or extensions, and emerged more capable than before.

This talk traces that arc of how an open-source project from Berkeley became one of the most popular databases in the world, not by chasing trends, but by combining a disciplined extensibility model, a stable core, and a fiercely independent community. We'll look at the architectural decisions that made PostgreSQL adaptable including the type system, the extension framework, foreign data wrappers, logical replication, the planner and how they let the project answer each successive challenge without abandoning its foundations.

Along the way, we'll demo features that exemplify this resilience such as JSONB and how it quietly defused the NoSQL argument, logical replication enabling a multitude of use cases from upgrades and migration patterns to multi master systems that were once the exclusive domain of expensive commercial databases, pgvector and the in-database AI story, and a look at where extensions are taking the platform next. Attendees will leave with a clearer picture of why PostgreSQL has lasted, and a practical sense of the features that make it a credible default for the next decade, not just the last one.