Databricks gave its AI coworker a spreadsheet. It's the fifth startup it's bought this year.

The deal, in plain terms

On 24 September, Databricks said it had bought Row Zero, a Seattle startup that builds a fast, cloud-scale spreadsheet, and folded it into Genie, the AI coworker built into its data platform. Terms were not disclosed, which is usually the sign of a small deal, and this is a small company: Row Zero was founded in 2021 by two former Amazon Web Services engineers, Breck Fresen and Nick End, and had raised about $10 million in 2025 at a reported $40 million valuation. The pitch is simple enough that it barely needs a press release. Business teams live in spreadsheets, Databricks' AI agent lives next to the data, so connect the two.

The product Databricks is describing is a "governed spreadsheet": an Excel-and-Sheets-compatible interface, formulas and pivots included, wired directly to live enterprise data with the access controls, audit trails and security of Databricks' Unity Catalog underneath. Every cell reads from governed sources, and, the company says, every action an AI agent takes inside it stays interpretable. That is a genuinely reasonable idea, and worth saying plainly before the skepticism: the spreadsheet is where most business data work actually happens, and giving an agent a governed one to operate in is more useful than another chat box.

The quote that gives away the strategy

Databricks CEO Ali Ghodsi was refreshingly blunt about the logic. "Spreadsheets are one such interface that every business analyst loves. So it makes a lot of sense to intermarry Business Intelligence (BI), Agents, and Spreadsheets," he told TechCrunch. Read that as the whole enterprise-AI thesis in one sentence: the winning move is not a smarter model, it is putting the agent inside the tool people already refuse to leave.

Row Zero's founders tell the same story from the other side. "We built Row Zero because we love spreadsheets but recognized they needed to seamlessly connect to enterprise-scale data with governance and security at the core," said Fresen, now joining Databricks. His co-founder Nick End framed Databricks as "one of the best-positioned companies to lead that shift." Founders talk their acquirer up; that is the genre. But the underlying observation, that the interface is the battleground, is correct.

The louder signal: a shopping list

The more revealing part of this deal is not the deal. It is what Ghodsi said around it. Databricks "intend[s] to do many more acquisitions like this in the future," and Row Zero is already its fifth purchase of 2026, following Quotient AI, SiftD.ai, Panther and Electric. This is a roll-up in progress. The company is not so much building its agentic platform feature by feature as buying it startup by startup, absorbing teams and slotting their tools into Genie.

That pattern is the actual news for anyone paying attention to enterprise AI. The large data and cloud platforms have concluded that the fastest path to a credible "AI coworker" runs through acquisitions, and they have the balance sheets to shop. For founders building enterprise-AI tooling, that sets up a stark binary: you are either a plausible acquisition target for a platform assembling its agent, or you are competing head-on with a company that can buy the capability you spent years building.

What is real, and what is packaging

Strip the announcement down and two things are true at once. The real part: a governed spreadsheet connected to live enterprise data, with auditable agent actions, is a sensible and even overdue piece of enterprise AI, and Databricks is well placed to ship it because it already owns the governance layer. The packaging part: "AI coworker" is a product name doing a lot of narrative work, the acquisition is a modest tuck-in of a small team, and nothing here is a technical breakthrough. It is distribution and integration, not invention.

None of that makes it a bad deal. Buying a strong spreadsheet team to give your agent a native surface, at what was almost certainly a small price, is close to textbook. It just should be read as what it is: an incremental, smart, unglamorous move in a consolidation wave, not the arrival of a new kind of intelligence.

Our take

The Row Zero acquisition is the enterprise-AI story in miniature: the interface matters more than the model, governance is the moat, and the platforms are buying rather than building their way to an agentic future. The spreadsheet-for-agents idea is genuinely good and Databricks is a natural home for it. But the headline capability is a feature, the deal is a tuck-in, and the real disclosure is Ghodsi's promise of "many more" to come. Watch the shopping list, not the spreadsheet. That is where this year's enterprise-AI map is being drawn, one quiet acquisition at a time.