A few months back, when Satya Nadella suggested that SaaS was dead, or at least dying, plenty of people dismissed it as provocation. Bold, maybe. Wrong, definitely. Except here we are. And the stock market, which has a habit of cutting through comfortable narratives, seems to be nodding along.

Consider this. Some of the biggest SaaS vendors on the planet have seen their stocks fall between 35 percent and 50 percent over the last year. Not because they missed earnings. Not because growth collapsed. In many cases, they beat estimates and reported solid fundamentals. Yet the punishment has been swift and persistent. Even Microsoft, which derives a significant share of its revenue from SaaS, is slightly down despite a few short-lived rallies. Oracle, another SaaS-heavy enterprise, is up a modest four percent, which barely registers compared to broader market gains.

Two recent pieces from Salesforce Ben highlight this disconnect. One explores why Salesforce remains in a slowdown despite strong results. The other looks at ServiceNow losing roughly half its market value in a year even while delivering good earnings. The conclusion is unsettling. The market is not questioning execution. It is questioning the SaaS model itself.

One explanation is simple. Valuations got far ahead of fundamentals. For years, SaaS companies traded at multiples that ignored earnings discipline, cash flow realities, and reasonable growth expectations. Story mattered more than substance. Now the market is repricing risk, rediscovering gravity, and forcing companies to earn their valuations again. If that is all this is, then so be it. Welcome back to fundamentals.

But that explanation only gets us part of the way there.

The bigger issue, the one you cannot ignore if you spend any real time in enterprise technology, is AI.

AI has put SaaS into a kind of suspended animation. Not because SaaS stopped delivering value, but because AI has absorbed nearly all the oxygen in the room. Budgets, roadmaps, and executive attention are bending toward AI. Agentic AI in particular has become the focal point. If a project is not AI-related, many vendors are hesitating to invest, to build, or even to promote it. I see this every day at Techstrong. The conversations, the pitches, the sponsorship discussions, and the strategic priorities all orbit AI. Traditional SaaS initiatives are being delayed, reduced, or quietly pushed aside.

To better understand whether this was anecdotal or systemic, I looked at what my colleagues at Futurum Group have been researching on this topic. Their analysis points to a clear AI crowd-out effect in enterprise planning cycles. Across recent enterprise AI adoption research, discretionary software spend is being reallocated toward AI infrastructure, model access, and data engineering, particularly in fiscal year 2025 budgets. In practical terms, that means money moving away from incremental SaaS modules and seat expansions and toward GPU and accelerator capacity, vector databases, data quality pipelines, and agent orchestration layers. Total software budgets may be flat or slightly up, but the composition of that spend is shifting away from traditional SaaS expansion.

This is not just a spending shift. It is a belief shift.

CIOs and boards are increasingly asking whether it makes sense to commit additional dollars to SaaS platforms when autonomous agents promise to gather data, analyze it, present insights in a preferred format, and then take action on a user’s behalf. If you believe that future is even moderately close, you think twice before signing another long-term SaaS contract.

I asked my Futurum analyst colleagues what their buyer research showed, and the pause is real. In enterprise interviews conducted for recent AI planning notes, more than half of large organizations reported delaying noncritical SaaS renewals or module upsells specifically to free up funding for generative AI pilots and agentic proofs of concept. At the same time, there is growing scrutiny around what many buyers now call copilot taxes. These are the per-user price increases attached to AI assistants embedded inside SaaS suites. Many enterprises are weighing those costs against building AI capabilities at the platform layer and surfacing outcomes through agents that can operate across systems.

That tradeoff is driving early experimentation with agent frameworks that sit above the application layer. In those architectures, value accrues to data access, orchestration, and governance rather than to any single application interface.

Here is the irony in all of this. The SaaS vendors getting hit hardest by the market are often the same ones leading the charge into agentic AI. Salesforce and ServiceNow are not laggards. They are pioneers. They are embedding agents into workflows, rethinking automation, and positioning themselves as platforms for autonomous work. On paper, they should be getting credit for that leadership.

So why are they not?

One explanation is that the market does not fully believe the agentic AI story yet, at least not enough to reward incumbents. A more uncomfortable possibility is that investors believe true agentic AI ultimately threatens the SaaS model itself. If agents can operate across systems and abstract away interfaces and workflows, the value of the underlying SaaS platform diminishes. That leaves SaaS vendors facing an uncomfortable choice. They can lead the transition and risk cannibalizing their core business, or they can protect the core and risk becoming irrelevant.

My Futurum colleagues describe this as platform inversion risk. Value shifts away from application-specific front ends and toward cross-system agent platforms along with the data and control layers beneath them. In analysis of the enterprise AI stack, durable margin pools are more likely to accrue to the data plane, the control plane, and the orchestration plane. In that world, application interfaces matter less, which compresses the traditional SaaS premium on proprietary workflows. SaaS incumbents can still win, but only if they compete on clean APIs, open agent tooling, data adjacency, safety, and extensibility rather than on seats and screens.

Markets are fickle. Trying to extract perfect logic from stock movements is often a fool’s errand, something closer to reading goat innards than disciplined analysis. Still, the signal here is hard to ignore. The market is saying that SaaS, as we have known it, is no longer the growth engine it once was. And it is not convinced that AI, at least in its current agentic form, has fully earned its keep.

Shimmy’s Take

Sooner rather than later, traditional SaaS is toast.

When I have agents that can gather the information I need, perform the analysis, present it in the way I prefer, and then take action autonomously on my behalf, I do not need an expensive SaaS platform packed with dashboards and licenses. I need outcomes. Agents, if they live up to even part of their promise, deliver outcomes more directly than SaaS ever did.

That said, there is a long road from here to there. Agentic AI today is powerful, but it is also immature, uneven, and often overhyped. Enterprises do not flip switches overnight, and mission-critical systems are not replaced on a whim. The SaaS players understand this. They are smart, well capitalized, and deeply embedded. They will find ways to insert themselves into the agentic future, if not to dramatically improve our lives, then at least to keep the gravy train moving.

So is SaaS dead? Not yet. But the stock market is telling us it is no longer sacred. And once a business model loses its aura of inevitability, the clock starts ticking.