Databricks CEO: AI Won't Kill SaaS, but It Will Make It Obsolete

# AI won't kill SaaS, but it will make it obsolete — and here is why that makes sense

Cloud software is not going away. But its reign is ending. That is the forecast from Ali Ghodsi, CEO of Databricks, and his reasoning deserves attention at a time when every startup is rushing to wrap existing software in a layer of artificial intelligence. Ghodsi rejects the popular narrative that powerful language models will simply "rebuild" traditional SaaS applications on the fly. The reality will be far more interesting, and more dangerous for today's incumbents.

The core of his vision is simple but destructive: a new generation of companies will not bolt AI onto finished products like Salesforce or Workday. Instead, they will build software from scratch, with artificial intelligence at the foundation of the architecture. That is not the same as adding a chatbot to the side of an app. It means rethinking the very logic of how people interact with enterprise software. When AI is built into a product's DNA rather than stuck on top, everything changes: speed, interface intuitiveness, automation capability, price. Old players, weighed down by the legacy of the previous decade, will prove sluggish in this race.

The history of technology offers many examples of this kind of upheaval. When the web appeared, companies tried to simply move their desktop applications into the browser, and the result was clumsy. Those who started with a clean slate (Google Docs instead of MS Office in the cloud) won. The same story is repeating now, one level up. Companies like Notion or Linear already show how simple, elegant AI assistants can fit into work better than complex systems built two decades ago. Databricks, incidentally, positions itself as a platform on which it is convenient to build exactly these kinds of new products.

It is important to understand that Ghodsi is not talking about the literal death of the SaaS model. Cloud software will remain cloud software. Subscriptions will continue to exist. The question is a different one: who will own this space? Today the cloud is dominated by companies founded in the 2000s, when AI was an exotic curiosity. They have already invested billions in infrastructure, contracts and reputation. But if AI really does change the rules of the game, the power of those assets loses its value. A young startup with the right architecture, built on up-to-date models, may prove more useful than the giant legacy of Microsoft or Oracle.

This creates a window of opportunity for new companies and a danger for the established order. Uber did not kill the taxi; it redefined the market and took its dominant share. AI may do the same to SaaS, only faster. Not through direct replacement, but through fresh competitors that will show enterprise software can be more intuitive, more efficient and cheaper if it is rebuilt around the capabilities of machine learning.

For the industry, this means an upcoming revaluation: those who navigate the transformation successfully will survive, but many latecomers will face customer migration. For us, the users, this is good news. Competition always squeezes prices and raises quality. Cloud software will not go away, but who builds it and how it works will look very different.