Pinecone launched a serverless vector database, Pinecone said on 21 May 2024, arguing that accurate, fast and scalable retrieval should be accessible to organisations large and small.
Vector databases became default infrastructure for retrieval-augmented generation. Serverless tiers lower the ops burden for teams that do not want to size pods for spiky AI workloads.
Pinecone was already a named player in managed vectors; serverless is the packaging shift that competes with cloud-native alternatives and open-source stacks on total cost.
Architects should benchmark recall quality, multitenancy isolation and price at their true query volume — serverless list prices often move once traffic leaves the demo tier.