Software-as-a-service has long been an investor’s dream — high margins, recurring revenues, and sticky customers. Companies like Salesforce, SAP, and ServiceNow grew into giants on the back of that model. But the dream is starting to crack. Over the past two weeks, we co-hosted roundtables with senior business leaders in San Francisco and New York to discuss how AI is reshaping value creation. The threat to SaaS was a recurring theme — and their observations point to three structural forces that enterprise software companies can no longer ignore.

The first is market vulnerability. SaaS margins have been high for decades, propped up by switching costs that keep enterprise customers locked in — whether or not they are satisfied. Many companies pay hefty sums for ERP, CRM, and other business-critical platforms not necessarily because they love the product but because migrating away is painful. That kind of captive market is an invitation to disruption.

The second force is collapsing barriers to entry. Building enterprise-grade software used to require enormous capital and engineering talent. Today, AI coding agents have made it much cheaper and faster. That means more competitors, more alternatives, and ultimately more pressure on the margins that SaaS companies have long taken for granted.

The third — and perhaps most consequential — force is the rethinking of workflows. SaaS companies built their empires by standardizing processes across industries: one CRM for every company, one finance platform for every CFO. But AI is enabling organizations to redesign workflows from the ground up. Several roundtable participants argued that deep vertical expertise — an intimate understanding of healthcare operations, for instance — is becoming more valuable than mastery of a horizontal, one-size-fits-all process. This flips the competitive matrix. Instead of selling a standardized workflow across sectors, the winning play may be offering sector-specific intelligence that adapts to how each industry actually works.

Together, these forces — dissatisfied customers, lower barriers to entry, and shifting value propositions — are likely to intensify competitive pressure across the software industry.

Software companies aren’t alone — AI is reshaping competitive dynamics across entire value chains. The clients of SaaS companies are under pressure too, forced to rethink operations as AI capabilities accelerate.

Despite these concerns, our roundtable participants did not predict the death of SaaS. Foundational software layers will still be needed, and enterprises will resist becoming wholly dependent on any single AI vendor. But margins will likely compress. A key driver will be a shift from input-based pricing — charging per user or per seat —  to output-based pricing, where customers pay for results. Several AI-native companies are already pushing this model. If it takes hold, it could erode the economics that made SaaS so lucrative. There is a catch, however: output-based pricing requires the ability to measure outcomes reliably. For some use cases — AI-powered call centers, for example — the transition may come quickly. For complex domains like legal services or healthcare, defining and tracking output quality is far harder.

Another underappreciated implication of AI is that it destabilizes today’s enterprise software ecosystems. As copilots and agents start to configure and run workflows, the old division of labor — vendors sell, systems integrators implement, consultancies advise — begins to blur. ERP and SaaS vendors are moving into AI-enabled services; integrators and consultancies are productizing vertical agent layers that sit on top of platforms; hyperscalers and model providers are bundling tooling that bypasses parts of the application layer. The fight shifts to new control points: orchestration, privileged data access, and distribution into day-to-day work. Those control points, and the ecosystems built around them, will determine who captures value.

Will software valuations recover? It is impossible to say, and will depend on how SaaS firms play their hand. But February felt like an inflection point. For years, AI was sold on potential. Now it is delivering impact. The technology has become central to geopolitics and national strategy. We may be crossing from the era of “pay attention, this will matter someday” to the era of “pay attention, this matters now.”

The SaaSpocalypse may be an overstatement. But the forces behind it are real, and the software industry’s most comfortable assumptions are no longer safe.

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.