A data marketplace is a place where providers publish data and buyers request it. Databricks Marketplace lets a Databricks customer do that from its own workspace and receive the data as a read-only catalog. This guide explains, from the buyer's side, how a listing becomes data in your workspace, what changes when you do not use Databricks, which parts of a listing to read before you request access, and how to load data that a provider delivers direct.
What Databricks Marketplace is
Databricks's documentation describes Databricks Marketplace as an open exchange where data providers, software vendors and technology partners publish offerings that customers can discover, evaluate and connect with from their workspace. The offerings include datasets, notebooks and AI models. Datasets are typically made available as catalogs of tabular data, and volumes carry non-tabular data.
The two sides have different entry requirements. A provider needs a premium Databricks workspace with Unity Catalog enabled and applies through the Databricks Data Partner Program, or uses a self-service sign-up that is limited to private exchanges. A consumer can browse the Open Marketplace without a workspace, but requesting a data product requires one. A private exchange restricts its listings to member consumers.
How a listing is built
Databricks's page on creating a listing names what a provider fills in, and each item is something a buyer can read before asking for access:
- a name, a short description and a longer description that can include schemas and field names
- categories
- an update frequency, which says how often the data assets are refreshed
- optional advanced attributes such as geographic coverage and time range
- up to ten sample notebooks that show the data in use
- links to documentation, a privacy policy, a licence and terms of service
- a provider profile
- the type of data asset: tables, files, models, notebooks or MCP servers
- an access model: instant, or subject to provider approval
The data itself is attached to the listing through a share, which the same page describes as an OpenSharing object holding tables, views, volumes and AI models that can be secured as a unit. For a buyer with a Unity Catalog workspace, the share arrives as the catalog described next.
How access works for a buyer
Databricks's page on accessing data products lists the requirements: a Premium plan or above, a workspace with Unity Catalog enabled, and the USE MARKETPLACE ASSETS privilege on the metastore, which is on by default. The CREATE CATALOG and USE PROVIDER privileges, or the metastore admin role, also qualify. You can search and filter listings by product type, provider, category and cost, which can be free or paid.
| Access model | What you do | What follows |
|---|---|---|
| Instant | Choose Get instant access and accept the Databricks terms and conditions | A read-only catalog appears in your workspace, and you may rename it first |
| By request | Choose Request access, then give company information and the intended use | The provider reviews the request by email, and the status reads Pending, Fulfilled or Denied under My requests |
The catalog is owned by the person who requested it, who can grant read-only access to colleagues. The page on managing shared data products adds that an installed product shows the provider's information, documentation, terms and sample notebooks, and that you can uninstall it. Read the provider's licence and terms on the listing as well as the Databricks terms you accept. They decide whether you may use the data internally only, embed it in a product or pass it on, which are the three uses a data licence separates.
OpenSharing underneath, and buyers outside Databricks
The protocol that moves the data is OpenSharing. Databricks's announcement of June 2026 says that Delta Sharing is now OpenSharing and calls OpenSharing the next evolution of Delta Sharing, and its documentation describes OpenSharing as the protocol that powers Marketplace. The guide to Delta Sharing explains the protocol, its change of name and what it means for a licence. For a buyer, two consequences follow.
First, you read the data through the share, and the provider decides what the share contains. Databricks's page for recipients adds that recipients can read and copy shared data but cannot change the source.
Second, a buyer without a Unity Catalog workspace still has a route. Databricks's page on using external platforms says you can request a data product, download a credential file from the listing and use it with Microsoft Power BI, Microsoft Excel, pandas, Apache Spark or a Databricks workspace without Unity Catalog. Only the dataset product type is available this way, so notebooks, volumes and models are not.
The credential file can be downloaded once. A later download rotates the credential, and the old one expires after a day or at its original expiry, whichever is sooner. Only two credentials can be active at once. Treat the file as a secret: it is the whole of your access to the share.
What to check on a listing before you request it
For company data, a few questions decide whether a listing is usable, and most of them have a place on the listing or a place to ask.
| Question | Where to look | Why it matters |
|---|---|---|
| How often is it refreshed? | The update frequency | A daily or weekly decision needs a daily or weekly feed |
| How far back does it go? | The time range attribute, if the provider has set one | A back-test needs history, and a monitor does not |
| What are the columns? | The description, the linked documentation and the sample notebooks | A data dictionary should exist before you request access |
| What does it join to? | The description, or ask the provider | Without a stable company identifier and listing identifiers, the data will not join to your records |
| Are published rows revised later? | Ask the provider | Revised history makes a back-test look better than it was |
| What may you do with it? | The licence and terms links | Internal use, embedding and redistribution are separate permissions |
Take an illustrative case. Acme Robotics wants weekly hiring trends for its own sector. A listing with a daily update frequency and a time range of two years fits a trend question but not a test that needs a longer record. A description with no company identifiers means the data cannot be joined to Acme's records until the provider says how. Licence terms that cover internal use leave an embedded dashboard for Acme's customers uncovered. Each finding calls for a different step: request, ask, or negotiate.
The listing fields describe geography and time range. Whether the data covers your own list of companies is something only a sample answers, so ask for one before you commit. Fokals sends sample data for the companies a buyer names on request, with the data dictionary and methodology.
For comparison, here is what Fokals states on the same points. Refresh runs from daily to weekly by dataset: hiring and intent signals daily, intent scores and market series weekly. Each company has one stable Fokals company ID, and a listed company carries its ticker, MIC, ISIN, LEI and share-class FIGI. Daily and weekly tables are written once, after the period closes, and are not revised. Each export carries a manifest naming sources, period, label versions and licence. The data dictionary lists every table and column.
When a provider delivers direct
If a provider is listed with instant access and your team already works in a Unity Catalog workspace, you request it from the listing and receive a read-only catalog. Data that a provider delivers direct arrives another way, and the route is just as short.
Fokals is delivered direct, by REST API and as bulk files in JSON, JSON Lines or CSV, which you load with Databricks's own loaders. The delivery page describes the methods and formats, and Fokals is licensed by written agreement, as the licensing page sets out. Loading is ordinary data engineering in your own workspace, and the guide to loading company data into Databricks shows landing files in a volume and building bronze, silver and gold tables. The guide to governing licensed data with Unity Catalog shows how to give a catalog loaded this way the same grants, tags and lineage as data that arrived through a listing.
Frequently asked questions
Do I need a Databricks account to use Databricks Marketplace?
You can browse the Open Marketplace without a workspace, but requesting a data product needs a Databricks workspace. Once data has been shared with you, a Unity Catalog workspace is not required to read it: a tabular dataset can be used through a credential file in Power BI, Excel, pandas or Apache Spark, according to Databricks. A workspace with Unity Catalog adds governance and auditing.
Is data on Databricks Marketplace free?
Some of it is and some is not. Databricks lets providers publish public data, free sample data and commercial offerings, and the Marketplace search can filter by cost, free or paid. Access is either instant after you accept the terms or subject to provider approval. The price and terms of each product are set by its provider on the listing.
What is the difference between Databricks Marketplace and OpenSharing?
Marketplace is the exchange where listings are found and requested. OpenSharing, which Databricks announced in June 2026 as the next evolution of Delta Sharing, is the protocol that moves the shared data. Databricks documents a listing as being backed by a share, an OpenSharing object holding the tables and other assets a buyer receives.
Can I use Databricks Marketplace data outside Databricks?
Yes, for tabular datasets. Databricks says a buyer can download a credential file from the listing and read the data with Power BI, Excel, pandas, Apache Spark or a workspace without Unity Catalog. Notebooks, volumes and models cannot be used this way. The credential file downloads once, and a second download rotates the credential.
How does Fokals data reach a Databricks workspace?
Fokals is delivered direct, by REST API and as bulk files in JSON, JSON Lines or CSV. You land the files in a volume or pull the API, load them into Delta tables with Databricks's own loaders, and govern the result with the same Unity Catalog grants, tags and lineage as any other catalog. The guides to loading and to governance linked above show each step.
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.