Databricks hits 5.4 billion dollars run rate as LLM interfaces reshape SaaS moats

Databricks is using its latest growth numbers to make a point B2B software leaders can’t ignore: LLM-based interfaces are not automatically “killing SaaS,” but they are changing what creates defensibility.

Revenue growth and the shift from SaaS UI to LLM UX

Databricks reported it has reached a 5.4 billion dollars revenue run rate, up 65 percent year-over-year, and said more than 1.4 billion dollars of that run rate comes from AI products, according to the company’s announcement (Databricks press release).

CEO Ali Ghodsi framed the update as a response to the narrative that AI will hollow out SaaS. His argument: the bigger disruption is the interface layer. As natural-language UX becomes standard, end users may no longer need years of training on a vendor’s screens and workflows. For marketers and operators, that translates to faster onboarding and more self-serve analytics—but also weaker “UI-as-moa`t” lock-in across the stack.

A concrete example is Databricks’ LLM interface, Genie, which lets non-specialists ask business questions in plain language (e.g., diagnosing spikes in warehouse usage and revenue) instead of writing SQL or commissioning custom reports. This is the same UI replacement pattern discussed in industry commentary (Bucco Capital on X).

Systems of record stay put, but agents change buying criteria

Ghodsi’s view is that enterprises are unlikely to rip out systems of record (where critical operational data lives) just because LLMs exist. The more immediate threat is that model makers and AI agents will demand software that is easier to query, automate, and control via APIs and plug-ins.

Databricks is positioning newer products for that world, including Lakebase, which it says is built for agent-driven workloads. Separately, the company also closed a previously announced 5 billion dollars funding round at a 134 billion dollars valuation and added a 2 billion dollars loan facility (TechCrunch funding report).

For B2B teams, the practical takeaway is to evaluate vendors on “agent-readiness”: natural-language access, permissioning, auditability, and integrations—not just dashboards.