Vercel Marketplace and v0 now provision S3, S3 Tables and S3 Vectors for a project
Amazon S3, S3 Tables and S3 Vectors can now be created from the Vercel Marketplace and v0, with no AWS console step for policies or credentials. Functions trade short-lived OIDC tokens with AWS STS, so no static keys sit in the project.
The Engineer · Build desk
$100 USD in credits for a new AWS account created through Vercel, valid for six months. AWS says S3 Vectors cuts the cost of uploading, storing and querying vectors by up to 90%; unconfirmed.
| Figure | Value | Note | Claim |
|---|---|---|---|
| Credits for a new AWS account created through Vercel | $100 | 6 | |
| Credits valid for | 6 months | 6 | |
| S3 Vectors cost cut per AWS, up to | 90% | AWS claim, unconfirmed | 18 |
What happened
- In the dashboard the flow is Storage tab, Create, AWS, then a bucket type and Region, and AWS says provisioning takes a few seconds.
- The integration writes AWS_REGION, AWS_ROLE_ARN, AWS_RESOURCE_ARN and a bucket-name variable into the project, and vercel env pull brings them to a local file.
- In v0 the same integration provisions resources, configures connections and generates code from a prompt, Vercel said.
- A new AWS account created through Vercel gets $100 in credits valid for six months, or a team can link an existing account.
Why it matters
- What it limits constraint Vercel says linked accounts support only new buckets created through Vercel, so data already in an existing bucket stays outside this flow.
- What it costs cost The $100 credit lasts six months, so a cost comparison with Vercel Blob made inside that window will not hold once the credit runs out.
- Who has to decide decision Teams now choose per workload: Vercel points plain file uploads to Blob with built-in CDN delivery, and points existing AWS workflows, Iceberg tables and vector search to S3.
- Who's exposed exposure The integration writes the role and its permissions, so the scope of that role is a default nobody on the team has reviewed until someone opens it in the AWS account.
Next story · Build · Sep 30
AWS embeds DuckDB in Aurora PostgreSQL to query Iceberg and Parquet in S3Confirmed2 publishers
The credential path is the part to read closely. The injected variables are a region, a role ARN, a resource ARN and a bucket name [10]. At runtime, awsCredentialsProvider from @vercel/functions/oidc exchanges a short-lived OIDC token with AWS STS for temporary credentials scoped to the resource [2]. Vercel's changelog describes the same arrangement: short-lived OIDC credentials and no stored static AWS keys [3]. According to AWS, developers previously had to go to the AWS Management Console to configure policies and credentials [8]. AWS now says provisioning and connection configuration are handled automatically [16].
Application code stays ordinary. AWS's upload route builds an S3Client from @aws-sdk/client-s3 and sends a PutObjectCommand [4]. Apart from the OIDC import, the credentials argument is the only Vercel-specific code in that route [11]. Vercel's changelog also shows a createS3 helper from @vercel/aws, captioned "An S3 client authenticated through Vercel OIDC, with no static AWS keys." [5]
AWS frames the result as a stack. With the existing Aurora, DynamoDB and OpenSearch Serverless integrations, it says, S3 gives you "a complete application stack inside Vercel", "all on one AWS account with one bill" [12]. For projects that start in the Vercel dashboard, that is a fair description.
The performance and cost claims need the usual reading. AWS says S3 Vectors reduces the cost of uploading, storing and querying vectors "by up to 90%" [18]. "Up to" makes that a ceiling. For it to transfer, your vector count, query rate and current system would have to resemble the comparison, and the post does not say what the baseline was.
S3 Tables are managed Apache Iceberg tables with automated maintenance and compaction, queryable from any Iceberg REST catalog engine, including DuckDB [14]. AWS suggests running DuckDB inside a Vercel Function for "lightweight" queries [17].
What to watch
- The permissions of the role the integration generates, and whether they are limited to the single resource ARN AWS describes.
- Whether Vercel lifts the new-buckets-only limit on linked AWS accounts.
- A measured S3 Vectors comparison on a named baseline, to test AWS's up-to-90% cost figure.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence55
- Adoption
- Not scored
- Hype gap+15
- Incentives60
- Confidence70
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
Amazon S3, including S3 Tables and S3 Vectors, is natively available on the Vercel Marketplace and v0 by Vercel, covering everyday storage, analytics, and AI.
- [2]
At runtime, awsCredentialsProvider from @vercel/functions/oidc exchanges a short-lived OIDC token with AWS STS for temporary credentials scoped to the resource; no static keys are stored anywhere.
- [3]
Authentication for the Amazon S3 Marketplace integration uses short-lived OIDC credentials without storing static AWS keys.
- [4]
The AWS example Next.js upload route creates an S3Client and PutObjectCommand from @aws-sdk/client-s3 and passes credentials: awsCredentialsProvider({ roleArn: process.env.AWS_ROLE_ARN }) imported from @vercel/functions/oidc.
- [5]
The changelog shows a createS3 helper imported from @vercel/aws, captioned 'An S3 client authenticated through Vercel OIDC, with no static AWS keys.'
- [6]
Creating a new AWS account through Vercel gives $100 USD in credits valid for six months, or you can link an existing account and manage your plan and usage from the Vercel dashboard.
- [7]
Linked accounts support only new buckets created through Vercel.
- [8]
Provisioning S3 storage previously required navigating to the AWS Management Console to configure policies and credentials.
- [9]
From the Vercel dashboard, go to the Storage tab, choose Create, select AWS, pick the S3 bucket type, choose a Region and create the bucket; provisioning takes a few seconds.
- [10]
After connecting the bucket to a project, vercel link and vercel env pull .env.local pull the environment variables; the integration injects AWS_REGION, AWS_ROLE_ARN, AWS_RESOURCE_ARN, and a bucket name such as S3_BUCKET_NAME, S3_TABLE_BUCKET_NAME, or S3_VECTOR_BUCKET_NAME.
- [11]
In the AWS example upload route, the only Vercel-specific code is the @vercel/functions/oidc import and the credentials argument that calls awsCredentialsProvider; the client and command classes come from @aws-sdk/client-s3.
- [12]
AWS says that together with existing integrations for Amazon Aurora, Amazon DynamoDB and Amazon OpenSearch Serverless, you get a complete application stack inside Vercel, all on one AWS account with one bill.
- [13]
Vercel Blob provides public and private storage for application file uploads with built-in caching and CDN delivery; Vercel points to the S3 integration for existing AWS workflows, analytics on managed Iceberg tables with S3 Tables, or vector storage and search with S3 Vectors.
- [14]
S3 Tables are fully managed Apache Iceberg tables with automated table maintenance and compaction, directly queryable from any Iceberg REST catalog API-compatible engine, including DuckDB.
- [15]
Amazon S3 is also available in v0, which provisions resources, configures connections, and generates code from your prompt.
- [16]
With the S3 integration, resource provisioning and connection configuration are handled automatically.
- [17]
AWS says you can use DuckDB inside a Vercel Function for lightweight queries against S3 Tables.
- [18]
AWS says S3 Vectors reduces the cost of uploading, storing, and querying vectors by up to 90%.
Sources
2 independent publishers whose own reporting we read for this story.
- aws.amazon.comAmazon S3 is now available on the Vercel Marketplace and v0 by Vercel
2 articles · October 14, 2026
- blog.vercel.comAmazon S3 joins the Vercel Marketplace
1 article · October 13, 2026
Topics and entities
Topics
- Developer platform integrations
- Cloud object storage
- Workload Identity and OIDC/JWT Auth