The 3 forces quietly dismantling the business model that made enterprise software fabulously profitable

S&P software stocks fell about 20% in February as investors reassessed the enterprise SaaS model under pressure from AI-driven disruption. The article argues AI is lowering barriers to entry, weakening switching-cost moats, and pushing pricing from per-seat to output-based models, which could compress software margins. It does not predict the death of SaaS, but it frames February as an inflection point for the sector.

Analysis

The selloff is less about near-term demand elasticity and more about a structural repricing of software’s moat premium. AI lowers the cost of replicating features faster than customers can re-underwrite vendor risk, so the market is likely discounting a multi-year compression in gross margin, retention quality, and ACV growth rather than an immediate revenue collapse. That means the weakest links are the names with the most generic workflows, the highest valuation multiples, and the most exposed seat-based pricing models.

Second-order, the bigger vulnerability is not just apps but the ecosystem around them. If agents increasingly sit between users and software, the value capture shifts away from the application layer toward orchestration, data rights, and distribution into the daily workflow. That creates pressure on incumbent vendors while potentially benefiting infrastructure, vertical AI wrappers, and large cloud providers that can bundle tooling and control access points. It also raises the odds of margin leakage in services-heavy implementations, since consulting and SI attach rates are easier for customers to disintermediate when configuration becomes automated.

The near-term catalyst path likely runs through guidance, not earnings. Over the next 1-3 quarters, the market will punish any language suggesting slower renewals, higher concessions, or lower net expansion, because those are the first visible signs of pricing power erosion. The key reversal risk is that AI adoption itself can lift productivity enough to offset pricing pressure, especially in domains where outcome measurement is fuzzy and seat-based billing survives longer than expected.

Consensus may be overestimating the speed of full-stack disruption and underestimating incumbents’ ability to use distribution, data, and trust to defend core accounts. The more likely intermediate outcome is not SaaS death but a bifurcation: horizontal platforms rerate lower while vertical and mission-critical workflow owners hold up better. That argues for relative-value positioning rather than outright sector capitulation.

Key Decisions for Investors

• Short CRM vs long MSFT or GOOGL over 3-6 months: use a pair to express application-layer margin compression while keeping exposure to AI beneficiaries with distribution and infrastructure control; target 10-15% relative downside in CRM if guidance weakens.
• Short NOW into any post-earnings strength with a 1-2 quarter horizon: NOW is more exposed to workflow automation substitution risk, and downside accelerates if management signals slower seat growth or more discounting; manage risk with a tight 8-10% stop.
• Avoid buying the first dip in SAP/CRM until at least one quarter of evidence on renewal math: the setup is a classic multiple trap where lower growth estimates can outrun valuation support; wait for implied downside to be reflected in revised guidance.
• Long vertical AI enablers over horizontal SaaS via a basket trade: favor names that own proprietary workflows/data over generic application stacks; this is a 12-24 month theme with asymmetric upside if agentic automation reduces implementation friction.
• Consider put spreads on an equal-weight software ETF for 2-4 months: the sector tape is vulnerable to a sequence of negative guideposts, and spreads limit theta burn if the selloff overshoots and stabilizes.