'The SaaS apocalypse is overrated': How Workday and other software provders plan to survive AI

The market is still overpricing the binary "SaaS dies" narrative and underpricing the migration of value from the interface to the control plane. In the next 1-3 quarters, the more important effect is not lost revenue per se but slower seat expansion, weaker new-logo conversion, and a higher cost of retention as AI layers sit between the user and the app; that is a multiple problem before it becomes a top-line problem. Names with durable workflow ownership, identity, auditability, and embedded data access should defend better than thin front-end SaaS.

Within the group, WDAY and SNOW look structurally better positioned than FRSH if they can own permissions, logs, and cross-domain workflows, because those are the pieces agents cannot easily abstract away. FRSH is more exposed to agent-native substitution in support, where the user outcome is easier to replicate and the switching cost is lower; if AI reduces ticket-handling friction, pricing power in service software can erode faster than vendors imply. AMZN and GOOGL are quiet beneficiaries as the tax collectors of inference, orchestration, and data gravity even if the app layer is partially disintermediated.

The contrarian miss is that this may be less about existential displacement and more about a margin transfer from legacy app vendors to infrastructure/model providers. That argues for relative value, not a broad short SaaS basket: the immediate risk is 1-2 earnings prints where management commentary on AI attach does not translate into monetization, while the 6-18 month upside is reserved for vendors that can re-price around usage or platform control. What would falsify the thesis: WDAY/SNOW showing sustained AI monetization that offsets seat pressure, or FRSH proving it can keep ARPU flat despite agent adoption; absent that, the market likely still has room to compress weaker interface-only names.