Every CTO I talk to has a version of the same complaint: too many tools, too many handoffs, and not enough to show for it. Satya Nadella declared “SaaS is dead” on a podcast, and now everyone has an opinion. The interesting question isn’t whether he’s right. It’s what enterprise leaders actually do about it.
What’s really changing is the default operating model for enterprise software: less “add another app,” more SaaS rationalization, more orchestration across systems, and more selective building of workflow layers where off-the-shelf tools don’t fit. Getting this right requires a clearer enterprise software strategy, one focused not just on what tools you buy, but how you design and support them over time.
If SaaS were collapsing, you’d expect spending to follow. Instead, Gartner forecasts Cloud Application Services (SaaS) spending rising from $250.8B in 2024 to $299.1B in 2025 (about 19.2% growth).
So, the pressure isn’t on SaaS delivery. It’s on the app-first experience: dozens of interfaces, duplicated data, constant context switching, and licensing models that don’t map cleanly to how work gets done.
The frustration is real and measurable. The average enterprise runs somewhere between 106 and 275 SaaS applications depending on who you ask, spends roughly $49M a year on them, and still has nearly 70% of employees complaining about too much context switching. The problem is not any individual tool. It is that the portfolio grew without a coherent model for how work actually flows through the business.
Why the Build vs. Buy Decision Has Shifted in 2026
Generative AI is changing the economics of building, especially for internal workflow software.
McKinsey expects generative AI to accelerate vendor switching and argues that churn will rise as the growing ease of software development pushes more enterprises to shift from buying to building genAI solutions. They estimate a two to four percentage-point shift in spending allocation over the next three to four years, which they frame as roughly $35B to $40B.
In plain terms: “buy” is no longer the default answer for every workflow problem, particularly when the real cost sits in integration, governance, and adoption, not in the license itself.
But here’s the counterargument worth naming directly: we tried building everything before, and it didn’t go well. Enterprises spent decades unwinding custom stacks that became expensive, brittle, and impossible to staff. SaaS proliferation happened for a reason. What’s different now is not that building has become easy. It’s that the scope of what should be built has narrowed and sharpened. Generative AI lowers the cost of creating software. It does nothing for the ongoing cost of owning it. The argument for building in 2026 is not “build more.” It’s “build where it’s rational”, which is a much more defensible position.
A lot of the “SaaS is dead” debate is really an interface debate: if AI agents can do routine work, do humans still need to navigate SaaS screens all day?
CIO summarizes the current state well: experts are split. Some believe agents will remove much of the GUI layer for routine tasks while backend SaaS functionality remains essential for reliability and governance. Others argue the interface layer may erode first, with value shifting toward clean APIs, schemas, and orchestration.
That uncertainty is actually useful for leaders. It suggests you should invest in what remains valuable across outcomes: data access, integration quality, identity controls, auditability, and workflow design.
The unhelpful version of this conversation is “SaaS vs custom.” The helpful version is deciding where each approach is rational.
MIT Sloan offers a simple framework for acquiring genAI capabilities: buy, boost, or build. The same logic works well for enterprise workflows more broadly.
Buy when:
Boost when:
Build when:
In these cases, investing in focused, well-architected custom software solutions can reduce complexity rather than add to it.
Callibrity’s viewpoint is not “build more.” It’s “design better work.” Sometimes that means buying less. Sometimes it means boosting what you already have. Sometimes it means building the workflow layer so the business can move faster without adding another disconnected tool.
If you’re deciding how to respond to the “SaaS is dead” moment, these moves are practical regardless of where the market lands:
Organizations that approach SaaS consolidation and workflow design strategically tend to see measurable improvements in cycle time, cost control, and developer productivity. See how leading enterprises are rethinking their SaaS strategy
SaaS isn’t dead. But the era of “SaaS everything” without ruthless attention to integration, governance, and workflow design is ending. The winners will buy what should be bought, boost what should be made usable, and build where differentiation lives.