DATABRICKS JUST BOUGHT A SPREADSHEET STARTUP TO MAKE AI WORK WITH NUMBERS
Databricks just bought Row Zero, a startup built around an extremely old idea: the spreadsheet. The reason is surprisingly modern. Databricks wants AI agents to work with huge amounts of business data without forcing employees to abandon the spreadsheet interface they already understand.
THE SPREADSHEET WON
On September 24, 2026, Databricks announced that it had acquired Row Zero, a Seattle startup building a cloud spreadsheet designed for enormous datasets.
Financial terms were not disclosed.
The strategic reason was much more interesting than the purchase price.
Databricks wants to combine three things that traditionally lived in separate worlds:
Business data
↓
AI agents
↓
Spreadsheets
The company says Row Zero will become part of Genie, its AI coworker for business teams.
And there is a reason Databricks thinks spreadsheets still matter after four decades.
Almost everyone in business already knows how to use one.
DATabrICKS HAD A PROBLEM
Databricks stores and processes enormous quantities of enterprise data.
Its AI assistant, Genie, can already let employees ask questions in normal language.
For example:
“Why did our profit margin fall last quarter?”
or
“Which sales opportunities are most likely to close?”
Genie can analyze the underlying data and produce an answer.
But eventually the employee often wants to work with the numbers themselves.
They want to:
→ filter rows
→ change assumptions
→ build formulas
→ create a pivot
→ compare scenarios
→ visualize the result
→ collaborate with colleagues
And what do they usually open?
A spreadsheet.
DATabricks' OWN FINANCE TEAM FOUND THE ANSWER
The acquisition apparently came from Databricks' own finance department.
Its finance team had been using Row Zero because the spreadsheet could handle more than one million live rows without the limitations of conventional spreadsheet workflows. They combined it with Databricks' Genie internally.
Then the executive team noticed.
That led to a rather unusual acquisition logic:
Databricks employees use startup
↓
startup solves a real internal problem
↓
Databricks realizes customers have the same problem
↓
Databricks buys startup
CEO Ali Ghodsi described the logic as combining business intelligence, agents and spreadsheets because spreadsheets remain a familiar interface for business analysts.
The company didn't discover Row Zero by studying a competitor.
It discovered it by watching its own employees work.
ROW ZERO WAS BUILT FOR SPREADSHEETS THAT ARE TOO BIG
Row Zero was founded in 2021 by former AWS and Tableau engineers Breck Fresen and Nick End.
The startup's central idea was that traditional spreadsheet software wasn't designed for modern enterprise-scale datasets.
Row Zero built a cloud spreadsheet capable of working with extremely large datasets while maintaining familiar spreadsheet concepts such as formulas, pivots and keyboard shortcuts.
That combination is important.
It isn't trying to teach a finance employee an entirely new way of working.
It says:
Keep the spreadsheet.
Make the spreadsheet powerful enough for modern data.
THE REAL PROBLEM ISN'T EXCEL
The problem Databricks is trying to solve isn't that spreadsheets are bad.
It's that spreadsheets are too useful.
Employees constantly export information from corporate systems into Excel or Google Sheets because spreadsheets are flexible.
But once the data leaves the company's governed data environment, things become messy.
You can end up with:
Database
↓
CSV export
↓
Excel file
↓
email attachment
↓
another copy
↓
someone changes a formula
↓
nobody knows which version is correct
Databricks calls these kinds of uncontrolled spreadsheet systems “spreadmarts.”
The spreadsheet is convenient.
The data trail is not.
DATABRICKS WANTS THE SPREADSHEET WITHOUT THE DATA ESCAPE
Row Zero connects directly to live, governed data.
That means an employee can work with information in spreadsheet form without necessarily creating another uncontrolled copy of the underlying corporate dataset.
The structure becomes:
Company data
↓
Databricks
↓
Row Zero spreadsheet
↓
Human + AI
Instead of:
Company data
↓
download
↓
spreadsheet copy
↓
email
↓
unknown version
That is the real reason the acquisition matters.
AI CAN NOW WORK INSIDE THE GRID
Databricks isn't only giving humans a better spreadsheet.
It wants AI agents to work through the spreadsheet too.
Genie can understand a business question.
Row Zero provides the environment where the answer can be manipulated.
So an employee might go from:
“Why did our gross margin change?”
to:
“Show me the underlying data.”
Then:
“Create a scenario where supplier costs increase 8%.”
Then:
“Compare that with a 5% price increase.”
Then:
“Build a chart I can send to finance.”
The AI isn't just returning text.
It is helping manipulate the business model.
THIS IS WHY NUMBERS ARE DIFFERENT FROM CHAT
Generative AI is extremely good at producing language.
But business decisions often depend on things that cannot simply be approximated with plausible-sounding text.
Revenue.
Margins.
Inventory.
Cash flow.
Headcount.
Forecasts.
Pricing.
Sales pipelines.
Financial models.
One wrong number can completely change a decision.
A spreadsheet provides something AI chat interfaces often don't:
visible structure.
You can inspect the rows.
See the formulas.
Change an assumption.
Check the output.
That makes the spreadsheet a kind of control surface for AI.
DATABRICKS WANTS AI THAT CAN SHOW ITS WORK
This is one of the most interesting parts of the deal.
Databricks says Row Zero's agent actions remain interpretable and auditable through the spreadsheet's formulas and processing model.
That matters inside companies.
Imagine an AI tells a CFO:
“Operating margin will fall 2.4 percentage points.”
The CFO doesn't necessarily want:
“Trust me.”
They want:
“Show me how you calculated it.”
A spreadsheet can potentially expose:
→ the source data
→ the formulas
→ the assumptions
→ the transformation
→ the final result
AI becomes easier to inspect because it is operating inside a familiar analytical environment.
THE AI DOESN'T HAVE TO REPLACE THE SPREADSHEET
That is the clever part.
Databricks isn't saying:
“Stop using spreadsheets.”
It is effectively saying:
“What if the spreadsheet itself became AI-native?”
That is a much easier proposition for businesses.
Employees don't need to abandon decades of habits.
Instead:
Excel-style workflow
live enterprise data
AI agent
=
new interface for business analysis
ROW ZERO CAN HANDLE BILLIONS OF ROWS
Databricks says Row Zero's processing engine can work with billions of rows at interactive speeds.
That is dramatically different from the typical spreadsheet use case.
Most people think of spreadsheets as:
1,000 rows
10,000 rows
maybe 100,000 rows
Row Zero is designed for situations where the spreadsheet is effectively sitting on top of an enterprise data warehouse.
That makes it less like a giant Excel file and more like a visual interface to a database.
THIS COULD MAKE DATABRICKS MORE USEFUL TO FINANCE TEAMS
Databricks has historically been associated with data engineers, analysts and technical teams.
But the people making business decisions aren't always technical.
A CFO doesn't necessarily want to learn how a data warehouse works.
A sales manager doesn't necessarily want to write SQL.
A marketing executive doesn't necessarily want to build a data pipeline.
They already know spreadsheets.
So Databricks can now say:
Your data stays in our governed environment.
Your AI stays there too.
And you can work with it through a spreadsheet.
That lowers the interface barrier.
THE SPREADSHEET BECOMES THE BRIDGE
The acquisition creates a bridge between two groups.
Technical side
→ databases
→ data warehouses
→ governance
→ permissions
→ AI agents
Business side
→ rows
→ columns
→ formulas
→ pivots
→ charts
The spreadsheet sits between them.
That's why something that looks old-fashioned suddenly becomes strategically useful for AI.
SECURITY IS PART OF THE PRODUCT
Enterprise customers don't only care about whether an AI can calculate something.
They care about:
Who can see the data?
Where did it come from?
Can someone export it?
Who changed it?
Can we audit the action?
Databricks says Row Zero queries respect user permissions, data can automatically refresh from authoritative live sources, exports can be restricted, and interactions can be audited.
That turns the spreadsheet from a loose file into something closer to a governed business application.
DATABRICKS IS ALSO BUYING A HABIT
There is another reason this acquisition is clever.
Databricks isn't only buying software.
It is buying a behavior that already exists inside companies.
People already open spreadsheets when they need to answer questions.
They already build models there.
They already send them to colleagues.
They already understand formulas.
They already trust the grid.
So instead of trying to convince millions of workers to adopt an unfamiliar AI interface, Databricks can put AI inside an interface they already use.
THE COMPANY IS ON AN ACQUISITION STREAK
Row Zero isn't Databricks' only acquisition in 2026.
The company has also acquired:
→ Quotient AI
→ SiftD.ai
→ Panther
→ Electric
The targets span AI-agent evaluation, interactive notebooks, security and database technology.
Databricks CEO Ali Ghodsi told TechCrunch that the company intends to pursue more acquisitions like these.
The pattern is revealing.
Databricks isn't simply buying companies that make its core database product bigger.
It is assembling pieces around the idea of AI agents doing work inside enterprises.
DATabrICKS WANTS GENIE TO BECOME A COWORKER
The company calls Genie an AI coworker.
That description is important.
A chatbot answers questions.
A coworker is expected to:
→ inspect information
→ analyze it
→ create something
→ modify something
→ collaborate
→ help make decisions
Row Zero gives Genie a familiar workspace for that kind of activity.
The AI doesn't have to stop when it has produced an answer.
It can move into the spreadsheet and continue working.
THE BIGGER FIGHT IS OVER THE WORK INTERFACE
Microsoft has Excel.
Google has Sheets.
Salesforce has CRM data.
Databricks has enterprise data.
Every major software company is now trying to determine where AI should actually do work.
The winner may not be the company with the most impressive chatbot.
It may be the company that owns the place where employees already perform their daily tasks.
For finance teams, that place is still very often the spreadsheet.
MAACAT PERSPECTIVE
Databricks didn't buy Row Zero because spreadsheets are the future.
It bought Row Zero because spreadsheets never stopped being the present.
The AI revolution created a strange problem:
Companies have enormous amounts of machine-readable data, incredibly powerful AI agents and millions of employees who still want to see the answer in rows and columns.
Databricks is connecting those worlds:
live enterprise data
↓
AI reasoning
↓
spreadsheet interface
↓
human decision
The spreadsheet isn't being replaced by AI.
For now, AI is moving into the spreadsheet.
