S3 Tables bring Iceberg V3 tables: positional deletion vectors replace rewrite-heavy delete files, row lineage enables efficient incremental pipelines — with a one-way, atomic, no-rewrite upgrade from V2.
You can now provision S3 general purpose buckets, S3 Tables, and S3 Vectors from the Vercel dashboard — or by describing what you want to build in v0, Vercel's AI builder. It joins existing integrations for Amazon Aurora, DynamoDB, and OpenSearch Serverless, so the whole stack lives in one AWS account with one bill.
The good part is how credentials work: a short-lived OIDC token is exchanged at runtime with AWS Security Token Service for credentials scoped to that resource — so no static keys are stored anywhere. That's the security pattern every new integration should follow: no long-lived secrets sitting in a third-party's dashboard, no IAM user with a permanent key pair.
For frontend-leaning teams already on Vercel, this removes a classic friction point: storage provisioning previously meant hopping consoles, creating IAM users, and pasting keys into env vars. Now storage is provisioned where you build, scoped via short-lived OIDC credentials, and billed in the same AWS account as your other resources.