Pinecone's serverless vector database is generally available on AWS

Pinecone, the leading vector database for building fast and accurate generative AI applications, has announced the general availability of its serverless architecture on AWS. Since Pinecone announced the database's public preview announcement, over 20,000 companies have leveraged its capabilities to build fast, accurate, cost-effective generative AI applications, indexing over 12 billion embeddings on the novel architecture. Building on this success, and the serverless general availability on AWS, Pinecone also launched the Private Endpoints for AWS PrivateLink public preview aimed at users on the Enterprise plan.

The new architecture separates reads, writes, and storage, reducing costs significantly for many workloads. It features industry-first vector clustering on top of object storage, providing low-latency vector search over practically unlimited records. Innovative indexing and retrieval algorithms ensure fast and memory-efficient vector search without sacrificing retrieval quality. Finally, on-demand retrieval is powered by a multi-tenant compute layer, enabling a serverless experience that frees developers from concerns about infrastructure. The compute layer also allows companies to only pay for what they use according to Pinecone's usage-based billing.

Private Endpoints for AWS PrivateLink lets users connect from their VPC to their Pinecone index without public internet exposure. This reduces exposure risk, ensures only specific VPCs can access Pinecone, and secures data traffic over Amazon's network. Azure Private Link, GCP Private Service Connect, and role-based access control are upcoming.

Four months after the Serverless public preview launch, many industry-leading customers serving millions of customers can attest to Pinecone's efficiency, accuracy, and cost-effectiveness. Gong, Notion, New Relic, TaskUs, You.com, and Shortwave already use Pinecone serverless for the large-scale, critical workloads comprising billions of vectors needed to serve their customers.

Pinecone serverless in AWS includes a new Global Control Plane API; SDKs for Python, Node, and Java; and Pulumi, Terraform, and Spark integrations. Those looking to get started can start building with Pinecone free of charge using Pinecone's sample notebooks. Users of pod-based indexes can migrate to Pinecone's serverless indexes for free.