turbopuffer: fast search on object storage
Inaugural blog post about the development of turbopuffer, a search engine that uses object storage and SSD caching for cost-effective, low latency search. This post describes into the motivation behind its creation, its unique architecture, and how it significantly reduces costs for large-scale vector searches. Discover how turbopuffer is transforming search infrastructure for companies like Cursor and Suno, offering a scalable and reliable solution.
turbopuffer: fast search on object storage NEW: Instant namespace branching NEW: Branching for instant, copy-on-write namespaces turbopuffer: fast search on object storage Updated: March 05, 2026 • Simon Hørup Eskildsen (Co-founder & CEO) In late 2022 I was helping my friends at Readwise scale their infrastructure ahead of the launch of Readwise Reader (read-it-later app). We wanted to build a highly requested feature: article recommendations and semantic search using vector embeddings. Readwise was paying ~$5k/month for their relational database, but we found that vector search on the same 10
saved by
related reading
- turbopufferturbopuffer.com
- What does Turbopuffer do?technically.dev
- Building a web search engine from scratch in two months with 3 billion neural embeddingsblog.wilsonl.in
- How We Store and Search 30 Billion Facesclearview.ai
- Not All Vector Databases Are Made Equal | Towards Data Sciencetowardsdatascience.com
- Faiss: A library for efficient similarity search - Engineering at Metaengineering.fb.com
- The vector database to build knowledgeable AI | Pineconepinecone.io
- TurboQuant: Redefining AI efficiency with extreme compressionresearch.google
- Rebuilding Linear’s delta sync read pathlinear.app
- Static search trees: 40x faster than binary search · CuriousCodingcuriouscoding.nl
- Announcing ScaNN: Efficient Vector Similarity Searchai.googleblog.com
- Introducing Husky, Datadog’s third-generation event store | Datadogdatadoghq.com