Databricks CEO Ali Ghodsi says SaaS will persist, but agentic AI is poised to make traditional software interfaces and usage models increasingly obsolete.
For years, SaaS was the end state of enterprise software. According to Databricks’ CEO, it may soon be just a transitional phase.
The chief executive of Databricks said that while SaaS as a business model is not “dead,” advances in AI — particularly agentic systems — will increasingly make traditional software workflows irrelevant. The shift, he argued, will not be about pricing or hosting, but about how software is used at all.
The comment reflects a growing belief among infrastructure leaders that AI is changing software at a structural level.
From interfaces to outcomes
Traditional SaaS depends on users navigating dashboards, configuring workflows, and interpreting results. Agentic AI challenges that paradigm by collapsing those steps.
Instead of interacting with software, users increasingly describe goals — and AI agents execute tasks across systems automatically.
In that model, SaaS tools still exist, but they recede into the background, functioning as execution layers rather than destinations.
Why infrastructure companies see this first
Databricks sits below the application layer, powering data pipelines, analytics, and machine learning across enterprises. From that vantage point, the company sees how AI is being embedded into workflows rather than bolted onto products.
Agentic systems rely heavily on clean, governed data and scalable infrastructure — areas where Databricks benefits as software interfaces abstract away.
That positioning explains why infrastructure firms are often more blunt about SaaS disruption than application vendors.
SaaS economics face pressure
If AI agents replace manual interaction, traditional SaaS metrics — seat licenses, usage tiers, UI-driven engagement — begin to weaken.
Value shifts from access to outcomes. Pricing models based on users and clicks may give way to task completion, throughput, or business impact.
That transition threatens many SaaS incumbents while favoring platforms that sit closer to execution and data.
Not extinction, but absorption
Crucially, the argument is not that SaaS companies disappear — but that they become components rather than products.
The winners will be those that expose APIs, integrate cleanly into agentic workflows, and allow AI systems to act on behalf of users.
Software that resists that shift risks becoming friction rather than leverage.
A familiar pattern in tech history
The transition echoes earlier shifts: from on-premise software to SaaS, from command lines to GUIs, from desktop apps to mobile-first design.
Each wave didn’t eliminate software — it redefined where value lived.
If Databricks’ CEO is right, the next redefinition is already underway.
SaaS may survive. But the way humans experience software may not.
