Does AI Spell Doom for SaaS?

When Klarna CEO Sebastian Siemiatkowski mentioned during a quarterly investor call that his company was planning to use AI to shut down Salesforce and Workday, he had no idea of the storm it would create. Klarna, a buy-now-pay-later platform, is reported on the cusp of filing for a $1 billion-plus IPO so investors are watching closely to get a sense of how AI is impacting the cost of operating a business, with a focus on business software and SaaS.

Not surprisingly, the bold statement quickly morphed into headlines suggesting that the Swedish fintech giant had replaced major SaaS providers like Workday and Salesforce with AI. The narrative gained such momentum that it appeared to cause a rift between the companies, with Salesforce CEO Marc Benioff publicly questioning the decision at a press conference, asking how Klarna was doing it without humans.

Months later, Siemiatkowski admitted feeling "tremendously embarrassed" by the fallout and clarified what really happened: Klarna hadn't replaced SaaS with AI at all. Instead, the company used AI to analyze its data ecosystem, identifying what information held real value and what was duplicated or conflicting across various platforms. This analysis revealed that Klarna's knowledge was fragmented across multiple SaaS solutions, leading the company to consolidate systems and eliminate silos. Klarna eliminated Salesforce CRM usage while holding onto the Slack chat application that was deeply embedded in internal processes. Klarna did roll its own CRM out of a combination of internal and external tools.

Even if Klarna didn’t end up using AI to entirely replace all SaaS, this episode highlights a growing question for businesses: Is AI driving a fundamental decoupling of data from traditional SaaS platforms? And if so, will SaaS survive in its current form or will it need to deliver more value, more flexibility and connectivity, or both?

The Great Fragmentation Problem

Klarna’s case illustrates a clear challenge many enterprises face today. The firm eagerly embraced AI and claims it saw real benefits — a 25% reduction in marketing spend and the creation of an AI-powered customer support system that obviated the need to hire hundreds of additional support reps. However, this came about because they recognized a big problem with fragmented data.

As companies have enthusiastically adopted specialized SaaS solutions for every business function, they've inadvertently created islands of critical business information. Knowledge about identity, goals, performance, relationships, workforce, and insights becomes scattered across dozens of platforms, making it difficult to derive comprehensive value. These siloes complicate cross-team collaboration and drive up costs by requiring more per-seat payments.

But what makes this fragmentation the most problematic in the AI era is that artificial intelligence thrives on connected, contextual data. When information exists in separate silos with different structures and access methods, AI systems struggle to generate meaningful insights. Siemiatkowski noted that Klarna's solution wasn't to replace SaaS with AI but to develop an internal tech stack to bring data together and then better deploy and partition that data out to a smaller pool of SaaS platforms. He also opined that SaaS platforms with broader capabilities — ironically, like Salesforce, would benefit more from this AI-driven consolidation. The upshot? Enterprises could leverage AI to first decouple and assess data value before then using AI in SaaS platforms to drive better results.

Microsoft's Vision of an AI-Transformed SaaS Landscape

Microsoft CEO Satya Nadella offers an even more radical perspective on this evolution. Speaking on the B2G podcast, Nadella suggested that the "notion that business applications exist" could "collapse" in the agentic AI era.

Nadella argued that at a fundamental level, SaaS applications are essentially CRUD (create, read, update, delete) databases with business logic layered on top. The Microsoft CEO predicts that this business logic—the rules that govern how applications work—will increasingly be handled by AI agents rather than being hardcoded into individual applications.

"They're going to update multiple databases, and all the logic will be in the AI tier, so to speak," Nadella explained. "Once the AI tier becomes the place where all the logic is, then people will start replacing the back ends." This vision suggests that rather than having separate applications for CRM, marketing automation, ERP, and so on—each with its own database and business rules—we might see AI agents handling tasks across all these domains, pulling and pushing data where needed. In other words, SaaS becomes a set of workflows and UX running atop data that the customer controls and meters. This is a big shift. SaaS companies today still charge not just by the seat but by the record, profiting on what is essentially data storage and access.

Consolidation vs. Elimination: The Future of SaaS

Despite these transformative predictions, both Siemiatkowski and Nadella stop short of declaring the death of SaaS. Instead, they envision a landscape with fewer, more comprehensive platforms that serve as knowledge hubs. Siemiatkowski specifically complimented Salesforce for expanding beyond CRM, suggesting it's well-positioned to become the "hub of knowledge that modern companies will seek." However, he also warned that many enterprise SaaS providers suffer from a "fallacy" by catering to too many demands and becoming little more than "glorified databases." The Klarna CEO argued that the real value lies in "opinionated software" that reflects experience and a clear understanding of what drives results. This suggests that SaaS platforms that offer genuine expertise and proven methodologies—rather than just storage and retrieval of information—will continue to thrive even as AI transforms the landscape.

Mere Decoupling or Full AI Transformation?

The real question is whether what we are hearing from Klarna and Microsoft represents a mere decoupling of data from different SaaS front-ends or is it a prelude to a full AI transformation of software that is conceived, built, and delivered. In the decoupling camp, we may be seeing the emergence of a necessary reintegration layer of fragmented business knowledge to power more intelligent systems and better take advantage of AI embedded in our existing SaaS system. In that sense, the decoupling makes the front end even more valuable. As Nadella's comments suggest, this could indeed transform what we understand as SaaS today. Business applications might become less about predefined workflows and more about providing structured environments where AI agents can operate effectively.

For a more extreme view of the AI Transformation, consider venture capitalist Tomas Tunguz, who is using Claude Code to write Android apps in languages he has never learned and deploying them to his phone for discrete tasks. “Over the weekend, I created my first Android app, not knowing a single line of Kotlin, the language in which Android is written. It cost me a little less than $5 to build a single-purpose app that listens to my voice, extracts one or a series of tasks, and then sends them to Asana. It’s a feature I don’t expect any task manager to build. I didn’t search the App Store, but perhaps such an app exists already. But I prefer mine because it’s built to my specification & took less than a few minutes.”

That’s a radical view of the future, where users build their own features to interact with SaaS platforms — or any other platform. Tunguz is still using a SaaS product, Asana, but he is adding features himself, without needing to go to the SaaS company. This paints a broader vision of AI Transformation, where SaaS remains in place but is potentially partially devalued by users who can tweak and customize the tool to their liking, reducing it to a de facto database with some elements of a front-end.

​​Conclusion: Navigating the AI-SaaS Transformation

The AI revolution isn't eliminating SaaS. It's transforming it. For business leaders in all fields that use SaaS (which is, most fields today), this transformation requires practical action, not just observation.

1. Start by reassessing your SaaS value equation. Premium pricing for basic data storage and simple workflow automation no longer makes sense. Demand more. Look for partners who help you maintain control of your data while delivering AI-enhanced insights that drive measurable outcomes.

2. Build your data strategy before expanding AI investments. The "decoupling" we've discussed isn't automatic. It requires intentional architecture. Consider creating a unified data foundation that connects information across systems while maintaining appropriate security boundaries.

3. Be ruthlessly selective about SaaS consolidation. Fewer platforms make sense but choose wisely. The most valuable partners combine AI capabilities with deep domain expertise—Siemiatkowski's "opinionated software" that embodies proven methodologies and industry knowledge.

4. Explore the build-your-own-features approach. AI tools are democratizing development. Your teams may now create custom extensions to core platforms without vendor delays. Or they might expedite customizations your primary development team can make. This flexibility can deliver competitive advantages while reducing dependency. Quality control must be maintained but in this new LLM world, a prompt can be a basic programmer

5. Extends this logic to all related apps including those on-prem business apps which have existed for years. Any companies evaluating or planning migration of legacy on-prem apps to SaaS should run through the same process described here before committing to SaaS apps.

6. Prepare for a hybrid reality. Some functions will remain in traditional SaaS while others shift to AI-agent workflows. This balanced approach lets you benefit from AI advances while maintaining enterprise-grade security, compliance, and reliability.

The winners won't be those who simply adopt AI or cling to traditional SaaS. They'll be leaders who integrate both approaches strategically. They'll maintain data control while leveraging increasingly intelligent tools. They'll create organizations where knowledge flows freely, powering innovation and driving growth in ways we're only beginning to imagine.

The future of business software may not be a binary choice between AI and SaaS but rather a thoughtful integration that preserves the best aspects of both approaches. The companies that understand this—whether they're SaaS providers or their customers—will be best positioned to thrive in the AI-first era.

Alex Salkever is a partner at Techquity and a former BusinessWeek technology editor, as well as an advisor to startups and large companies on the impacts of technology change and artificial intelligence. He is the author of four award-winning business books including “Driver in the Driverless Car” and “Your Happiness Was Hacked”.