AWS Data Exchange and Databricks Marketplace both let a team find third-party data from inside a cloud platform it already uses, and their documentation describes two different arrangements. On AWS Data Exchange a data product is a subscription with a price, a term and a contract, billed by AWS, and the data arrives in one of five forms. On Databricks Marketplace a listing is a catalogue entry for data shared through an open protocol, and for a commercial listing the documentation has the transaction run through the provider's own communications and payment platforms. This comparison follows a buyer through both, so that you know what you would hold, pay and sign in each.
A marketplace is built for three things: data that a provider publishes there, data read in place, as a read-only catalog or a datashare, with no load of your own, and, on AWS, a charge on your AWS bill. Fokals is delivered direct, by REST API and as bulk files, which you load into your own bucket or workspace with the platform's own loader. It brings firmographic, technographic, hiring and intent data on public and private companies, with security identifiers on every listed company, and the worked example below joins it to a marketplace table.
What each one is
AWS describes AWS Data Exchange as a service for sharing data entitlements between organisations and managing them. It has two units of exchange. A data grant is created by a sender to give a receiver access to a data set: the request goes to the receiver's AWS account and the receiver accepts it. A data product is published by a provider, listed in the AWS Marketplace catalogue after AWS has reviewed it against its guidelines and terms, and taken up by subscribers. A provider of products must be registered as an AWS Marketplace seller.
Databricks describes Databricks Marketplace as an open exchange: data providers, software vendors and technology partners publish there, and Databricks customers find and reach the offerings from inside a workspace. Listings include datasets, AI models, notebooks, apps and Model Context Protocol (MCP) servers. Data is exchanged through OpenSharing, which Databricks announced in June 2026 as the next evolution of Delta Sharing, the open source protocol it pioneered in 2021. The guide to Delta Sharing covers the protocol and the move to OpenSharing.
Both are a data marketplace of the kind built into a cloud platform. They differ in how much of the purchase the platform carries.
The two side by side
| AWS Data Exchange | Databricks Marketplace | |
|---|---|---|
| What is listed | Data products, each made of one or more data sets | Datasets, notebooks, AI models, apps and MCP servers |
| How data arrives | Files to export, an API, an Amazon Redshift datashare, access to the provider's Amazon S3 objects, or AWS Lake Formation permissions (preview) | A read-only catalog in your workspace, or a credential file for other platforms |
| Where you browse | The AWS Marketplace catalogue | The Open Marketplace website, or the Marketplace in a workspace |
| What you need to get data | An AWS account | A Databricks workspace to make the request; the Premium plan or above and Unity Catalog to receive a catalog |
| Provider approval | A subscription request stating your identity and intended use, reviewed by the provider | None for instantly available listings; the provider approves a request for the rest |
| Terms you accept | The offer: price, duration, payment schedule, refund policy and data subscription agreement | The Databricks terms and conditions, accepted with the request |
| Who bills | AWS, on the AWS bill | The provider, on its own payment platform; customers can also put part of a spending commitment towards eligible partner products |
| Private route | Private offers, Bring Your Own Subscription offers and data grants | Private exchanges, where a provider shares its listings with member consumers only |
What lands in your account
On AWS Data Exchange the answer depends on the type of data set, and AWS's page on data sets names five.
- Files. The subscriber gets a copy of the data set as an entitled data set and exports the assets to its own Amazon S3 bucket or downloads them to a local computer. AWS's page on product subscriptions says new revisions can be exported automatically to your Amazon S3 buckets as the provider publishes them, within the limit that page states.
- API. The subscriber calls an endpoint managed by AWS Data Exchange, which passes the call to the provider's Amazon API Gateway API.
- Amazon Redshift. The subscriber gets read-only access to query the provider's data in Amazon Redshift without extracting or loading it.
- Amazon S3 data access. The subscriber uses the provider's own S3 objects directly, with no copy to manage.
- AWS Lake Formation data permissions. A preview: the subscriber queries the provider's tagged databases, tables or columns through services such as Amazon Athena.
Only the first of these puts a file in your hands. The other four give access to data the provider holds, for as long as the subscription runs.
On Databricks Marketplace there are two outcomes. In a workspace enabled for Unity Catalog the data product appears as a read-only catalog, addressed like any other table as catalog, schema and table, and it can hold volumes of non-tabular data, AI models and notebooks as well as tables. Without such a workspace, the consumer downloads a credential file and reads the data with an OpenSharing connector. Databricks's page on external platforms lists Microsoft Power BI, Microsoft Excel, pandas and Apache Spark, and says that only tabular data sets are available this way. The request is still made from a Databricks workspace, for which a free trial is offered.
That page adds a detail worth planning for. The credential file can be downloaded once. A further download rotates to a new credential, the old one expires after one day or on its original expiry date, whichever is sooner, and only two credentials can be active at a time. Decide who holds the file before anyone clicks.
Paying, and what you sign
AWS states that every product is subscription-based. A subscriber accepts the offer's price, duration, payment schedule, data subscription agreement and refund policy, and an active subscription appears on the AWS bill as part of AWS Marketplace charges. The services you use on the data, such as Amazon S3 or Amazon Athena, are billed apart. The page on offers says durations run from 1 to 36 months, prices are in US dollars, and the provider controls the legal terms: it can use AWS's standard data subscription agreement, edit it or upload its own.
Two points in those pages reward a careful reading. If you turn on auto-renewal and the offer's terms have changed by the renewal date, the new terms apply, including a new price and a new agreement. And when a subscription to a Files data set ends you keep the files you exported, while AWS tells you to check whether your agreement requires you to delete them.
Databricks takes a different line. Its consumer page says that some listings are free and instantly available once you accept the terms. Others are available by request, typically because a commercial transaction is involved, and for those the page states that the transaction runs through the provider's own communications and payment platforms, and that the Marketplace itself handles no commercial transactions. When the transaction is complete, the provider makes the data product available in your workspace. Databricks's overview also describes a programme under which customers can put part of a spending commitment towards eligible partner products bought through the Marketplace.
So the same purchase reads differently in a procurement file. On AWS there is one more line on an existing bill and an agreement accepted with the offer. On Databricks, for a commercial listing, the documentation sends you to the provider's own communications and payment platforms.
A worked check on either platform
Whichever route delivers a table, test it the same way: join it to your own list on an identifier and count. The names here are illustrative. In Databricks the vendor's table sits under the catalog created from the listing, and in Amazon Athena it is a table you define over the exported files.
-- Share of your holdings found in the vendor's table
select
count(*) as holdings,
count(v.isin) as matched,
round(100.0 * count(v.isin) / count(*), 1) as matched_pct
from research.holdings h
left join (select distinct isin from vendor_trial.companies) v
on v.isin = h.isin;Data delivered direct joins in the same place once you have loaded it. Fokals rows carry a stable company ID and, for a listed company, its ISIN, LEI, FIGI, ticker and MIC, so a marketplace table and a table you loaded meet on an identifier. The second query sets a vendor's company list beside Hiring Activity, the daily count of open, new and closed postings per company.
-- A marketplace table beside a table you loaded yourself
select v.isin, f.day, f.open_postings, f.new_postings
from vendor_trial.companies v
join fokals.company_hiring_daily f on f.isin = v.isin
where f.day >= date '2026-09-01';The guides to loading company data into Databricks and querying company data on Amazon S3 with Athena show the load on each side, and the delivery page sets out what there is to load.
Which fits, by what you need
- Your procurement runs through AWS. AWS Data Exchange puts the charge on the AWS bill and the contract in the offer.
- You need files that you hold. A Files product on AWS Data Exchange exports to your own bucket, within the terms of its agreement.
- You work in Databricks with Unity Catalog. A Marketplace listing arrives as a read-only catalog under Unity Catalog, with notebooks and models where the provider includes them.
- Your analysts work outside both platforms. A tabular Databricks listing can be read from Power BI, pandas or Apache Spark with a credential file. Files exported from AWS Data Exchange can be read by anything that reads your bucket.
- You want to try before any contract. Look for an instantly available listing on Databricks, or for an AWS product whose data sets include a data dictionary and samples, which AWS lets you view before subscribing.
- The provider delivers direct. License the data by agreement and load it with the platform's own loader. The two platforms then hold it as they hold any table of your own.
The guides to AWS Data Exchange and Databricks Marketplace treat each on its own, and Snowflake Marketplace vs AWS Data Exchange sets AWS beside a third model.
What you still check on both
Each marketplace lists the providers that chose to publish there, and the product details are written by the provider. How the data was collected, whether it covers the companies you track and whether the licence allows your use are settled with a sample, a document and a contract, on a marketplace or off it.
Frequently asked questions
What is the difference between AWS Data Exchange and Databricks Marketplace?
On AWS Data Exchange a data product is a subscription: you accept an offer with a price, a duration and a data subscription agreement, AWS bills you, and the data arrives as files, an API or access in place. Databricks Marketplace lists data shared through OpenSharing: free listings open at once as a read-only catalog, and for commercial ones Databricks's documentation has the transaction run through the provider's own communications and payment platforms.
Who do I pay for a paid listing on Databricks Marketplace?
Databricks's documentation sends you to the provider. It says that listings which need approval typically involve a commercial transaction, that any transaction that follows uses the provider's own communications and payment platforms, and that no commercial transactions are handled directly on the Marketplace. Once the transaction is complete, the provider makes the data product available in your workspace. Databricks also describes a programme that lets customers put part of a spending commitment towards eligible partner products.
Do I need a Databricks workspace to use Databricks Marketplace data?
You need one to request access, even if you will read the data elsewhere. To receive a data product as a catalog you need a workspace enabled for Unity Catalog on the Premium plan or above. Without that, tabular data sets can be read on other platforms, such as Power BI, pandas or Apache Spark, with a credential file downloaded from the listing.
What types of data delivery does AWS Data Exchange support?
Five, according to its user guide: files that the subscriber exports to Amazon S3 or downloads, an API reached through an endpoint managed by AWS Data Exchange, read-only access to an Amazon Redshift datashare, direct access to the provider's Amazon S3 objects, and AWS Lake Formation data permissions, which AWS labels a preview. A product can contain more than one data set.
Can I receive data on AWS Data Exchange without subscribing to a product?
Yes, through a data grant. AWS describes a data grant as created by a sender to give a receiver access to a data set, for a period the sender sets. The grant request is sent to the receiver's AWS account, and the receiver accepts it to gain access. A data product, by contrast, is the unit of exchange that a provider publishes and that AWS lists in the AWS Marketplace catalogue.
How do I use Fokals data on AWS or in Databricks?
Fokals is delivered direct, by REST API and as bulk files in JSON, JSON Lines or CSV, under a written agreement. On AWS you place the files in your own Amazon S3 bucket and define a table over them in Amazon Athena. In Databricks you load them into your own workspace with the platform's own loader. Rows carry a stable company ID and, for a listed company, its ISIN, LEI, FIGI, ticker and MIC, so the loaded tables join to a marketplace table on a shared identifier.
The queries and code on this page are examples to adapt. Test them in your own environment before you rely on them.
What this page says about the products it names was checked against their public documentation on 4 October 2026. Product and company names are trademarks of their owners. Fokals is not affiliated with them or endorsed by them.