Databases in 2025: A Year in Review
Source: https://ottertune.com/blog/2025-databases-review Author: Andy Pavlo Date: 2026-01-06
Summary
Andy Pavlo’s (CMU database professor) annual review of database technology in 2025. Notable for coverage of vector database consolidation and AI-integrated database systems.
Key Claims
- Vector database consolidation: the standalone vector DB boom peaked in 2024; in 2025, major databases (PostgreSQL, MongoDB) added native vector support, making standalone vector DBs harder to justify.
- AI-integrated databases: databases that can run embedding generation and similarity search natively are emerging.
- The “data lakehouse” trend: columnar storage (Parquet, Iceberg) for analytical workloads is standardizing.
- LLM impact on databases: LLMs for query generation are becoming standard in developer tools; NL→SQL is production-ready for simple schemas.
- Agent database access: databases need new paradigms for agent access (different query patterns, authorization models, audit requirements).
Concepts
- Agent Memory — database systems as agent memory substrate
- Coding Agents — agents querying databases in new ways