For months, a specific fear has circulated in Silicon Valley: that artificial intelligence will wipe out traditional software companies by writing its own code and building its own tools. Databricks, a massive data infrastructure company, just released financial numbers that suggest the opposite is happening right now. But while their bank account is full, their CEO is warning that the era of human expertise in specific software tools is ending.
Key Takeaways
- Databricks reached a $5.4 billion revenue run rate, growing 65% year-over-year.
- The company closed a $5 billion funding round at a $134 billion valuation.
- AI products accounted for more than $1.4 billion of the total revenue run rate.
Databricks is primarily a “data warehouse” provider. This means they sell digital storage space where large companies keep all their records—sales figures, customer logs, and inventory lists—so they can analyze them later. It is unglamorous, backend plumbing for the internet.
The company announced it has raised $5 billion in new funding. This values the company at $134 billion. More importantly, their revenue from AI-specific products has crossed $1.4 billion. CEO Ali Ghodsi argues that AI isn’t killing the software business; it is just increasing how much people use the software they already have.
The big deal
The numbers are large, but the real story here is about job security and skills. For the last twenty years, people built entire careers on knowing how to use difficult software. You could get a high-paying job just by being a “Salesforce specialist” or an expert in SAP. That knowledge was a protective moat around your career.
Ghodsi suggests that moat is drying up. As AI interfaces take over, you no longer need to study a manual to use complex business software. You just talk to it. This makes the software “invisible,” similar to plumbing. You don’t need to know how the pipes fit together to wash your hands; you just turn the tap.
This shifts value away from knowing how to use a tool and toward knowing what to ask it. The software becomes easier to use, but the “power user” loses their special status.
How it works
Databricks is using large language models (LLMs) to change how humans interact with databases. They call their interface “Genie.”
Think of a massive university library. In the old system, if you wanted to find a specific fact, you had to understand the card catalog, know the Dewey Decimal System, and physically hunt through the stacks. If you didn’t know the system, you couldn’t find the book. Databricks is replacing the card catalog with a librarian who speaks plain English. You simply ask the librarian, “Why did revenue spike last Tuesday?” and the librarian runs into the stacks, finds the right ledger, and reads you the answer.
In the technical world, this replaces complex coding languages (like SQL) with natural conversation. A manager can ask why warehouse usage went up on a specific day without asking a data scientist to write a custom report.
The catch
While this makes data easier to access, it creates a new vulnerability for established software companies. The CEO notes that the main reason companies stick with old, clunky software is that their employees spent years learning how to use it. It is hard to switch when your whole team is trained on one specific interface.
If the interface becomes plain English, that “lock-in” disappears. It becomes much easier for a company to switch to a competitor because their employees don’t need to learn a new system—they just keep speaking English. This opens the door for new, AI-native competitors to steal customers from established giants.
There is also a scale issue. While Databricks’ new database for AI agents is growing fast, the CEO admits it is still a “toddler” compared to their traditional business.
What now?
Despite the massive influx of cash, Databricks is not planning to go public immediately. Ghodsi stated that now is not a good time for an IPO. The company is hoarding cash to ensure it has years of runway in case the economy crashes again.
If you work in data analysis or enterprise software, stop relying on your knowledge of button clicks and menus to prove your value. Start focusing on the underlying data strategy. The interface is becoming a commodity.
Watch for the release of more “agent-based” databases, which are designed for AI programs to talk to each other without human involvement at all.
