Databricks’ Row Zero Acquisition: Separate Live Analysis From Public Releases
What Databricks’ Row Zero acquisition means for spreadsheet sharing, plus checks for permissions, refresh behavior and reproducible public datasets.

Databricks announced its acquisition of Row Zero on September 24, 2026. The deal brings a warehouse-connected spreadsheet product into Databricks’ plans for Genie, its AI coworker. The official announcement describes a planned native spreadsheet integration; TechCrunch reported that financial terms were not disclosed.
If you use spreadsheets to support a report, research project or public dataset, the practical task is to distinguish a live analysis workbook from an approved public release. Before sharing either, establish who can open it, what data they can see and whether the values will change after publication.
What the acquisition brings together
Row Zero already supports warehouse connections, automatically updating spreadsheets, write-back of results and real-time collaboration. Its documentation also describes publishing a query as a reusable data source for teammates. These are documented Row Zero capabilities—not proof that every planned Genie integration is already available. Source: Row Zero overview.
Row Zero’s current product page advertises billion-row scale, Excel-compatible formulas and enterprise controls including SSO, SCIM, row-level security and restrictions on data export. Treat those as product capabilities to verify against your organization’s plan and configuration, not permissions automatically attached to every workbook. Source: Row Zero.
For data publishers, the important distinction is not spreadsheet size. It is the boundary between governed access to changing warehouse data and a release that outside readers can inspect and reproduce.
“Publish” does not always mean public
In Row Zero, publishing a data source means sharing a query with teammates so they can import its results into connected tables. Those sources can be refreshed or scheduled to update automatically.
The connection identity matters:
- OAuth: the query runs as the end user through their connection; warehouse row-level security is enforced, and the user needs database access.
- Service account: the query runs through the connection used to create the source; people in the organization with whom the source is shared can run it.
That difference belongs in your access review. Do not assume a shared service-account query applies each reader’s personal warehouse permissions. Source: Row Zero shared data sources.
Workbook links are a separate mechanism: review workbook access as well as query permissions. As documented on October 3, 2026, Free and Pro individual accounts can create public links that require no login. Business link sharing defaults to workspace-only access; Enterprise policies can restrict or disable link sharing. A Viewer cannot modify the workbook, but read-only access is not the same as private access. Source: Row Zero sharing documentation.
Audit the sharing workflow before changing tools
Use the acquisition as a prompt to review existing links—not as evidence that they have stopped working.
| Check | Record before sharing |
|---|---|
| Audience | Public readers, named collaborators or workspace members |
| Data scope | Approved columns, rows, aggregates and reporting period |
| Connection identity | End-user OAuth or a service account |
| Update behavior | Fixed snapshot, manual refresh or scheduled refresh |
| Release owner | Person responsible for approval and corrections |
Test intended public links while signed out. For restricted workbooks, test with an account outside the intended group. Review every sheet and field reachable through the shared workbook, rather than only the chart or table you intend readers to use.
Never upload sensitive information to a public publishing service. Prepare a separate, explicitly approved public dataset instead of treating a filtered internal workbook as the publication artifact.
Make the public release reproducible
For a fixed report, prepare an approved snapshot and keep its provenance alongside the published table. A minimal release record could use this structure:
| Field | What to document |
|---|---|
release_id |
Identifier for this published version |
data_as_of |
Cutoff date for the observations |
extracted_at |
Extraction timestamp, including time zone |
source |
Source dataset and query or method reference |
row_count |
Number of published observations, excluding the header |
transformations |
Filters, aggregation and missing-value treatment |
This is a suggested schema, not a Row Zero feature. Check row counts and totals against the approved source result, and preserve stable row identifiers where readers need to compare releases.
If you choose a live public dataset instead, disclose its refresh schedule and last successful update. Keep the dated snapshot behind any fixed claim in an article: a live spreadsheet may be useful for exploration, but readers also need to know which version supported the published analysis.