The Software-as-a-Service (SaaS) industry is currently facing an existential question: In a world where AI can talk directly to your data, why do we need software interfaces at all?
Databricks CEO Ali Ghodsi recently provided a definitive answer. Announcing a staggering $5.4 billion revenue run rate (up 65% year-over-year), Ghodsi argued that while SaaS “systems of record” are safe, the traditional SaaS business model—built on specialized user interfaces and “seat-based” licensing—is on the verge of becoming irrelevant.
As of early 2026, the data is clear: Databricks isn’t just surviving the AI transition; it’s thriving on it, with $1.4 billion of its revenue now driven directly by AI products.
The Death of the ‘SaaS Specialist’
For decades, the “moat” or competitive advantage for giants like Salesforce, SAP, and ServiceNow was complexity. Millions of professionals built careers around becoming certified specialists in these specific interfaces.
Ghodsi’s core argument is that Natural Language Interfaces (NLIs) are destroying this moat. When you can ask an AI like Databricks Genie, “Why did revenue spike last Tuesday?” and get an instant, visualized answer, you no longer need a specialist to write SQL or navigate a 15-click reporting dashboard.
What This Means for the Market:
-
Invisible Software: Software is moving from a visible destination (a website you log into) to “plumbing” (a background service).
-
Skill Devaluation: The value of knowing where a button is located is dropping to zero. The value now lies in knowing what questions to ask the data.
-
Usage-Based Surge: Databricks’ growth isn’t coming from selling more “seats,” but from increased data usage as more employees—not just technical ones—can finally interact with their company’s “system of record.”
The Rise of the ‘Toddler’ Database: Lakebase
Perhaps the most telling part of Databricks’ latest update is the success of Lakebase, their serverless Postgres database designed specifically for AI agents.
Ghodsi described Lakebase as a “toddler that’s twice as big” as their traditional data warehouse was at the same age. Why? Because AI agents need a different kind of home. Traditional databases were built for humans to query. Lakebase is built for agents to act. It bridges the gap between transactional data (OLTP) and analytical insights (OLAP), allowing AI agents to not only analyze data but also trigger real-world actions with millisecond latency. This “agent-native” infrastructure is where the next decade of enterprise value will be built.
Strategic Review: Why Not IPO Now?
Despite a massive $134 billion valuation and a newly closed $5 billion funding round, Databricks is staying private. Ghodsi’s reasoning is a masterclass in risk management for 2026.
By securing a $2 billion loan facility alongside their equity raise, Databricks has built a fortress-like balance sheet. Ghodsi is clearly wary of the market volatility seen in previous years and prefers “many years of runway” over the scrutiny and quarterly pressure of being a public company. For Databricks, the goal isn’t an immediate exit; it’s to become the foundational “Data Intelligence Platform” that powers every AI agent in the Fortune 500.
Conclusion: The Future is Agentic
The takeaway for businesses and investors is clear: The UI is no longer the product. SaaS companies that continue to focus on “improving the dashboard” are fighting a losing battle. The winners of the next era will be those like Databricks that focus on governance, data lineage, and agent-ready infrastructure. If your software requires a manual to use, an AI agent is probably already planning to replace it. We are entering the era of “invisible software,” where the only interface that matters is the one that understands what you say.