SaaS Isn’t Dead. It’s Splitting

How AI is reshaping software economics, compressing multiples, and separating durable platforms from vulnerable workflow layers.

By Laura Sánchez-Quiñones, Partner

Software multiples have compressed materially in the last few weeks. That part is obvious.

What’s less obvious is why.

Earnings haven’t collapsed. Demand hasn’t disappeared. Yet valuations have reset. The easy explanation — “SaaS is dead” — makes for a strong headline. But it doesn’t hold up under scrutiny.

What’s happening is more structural. Software isn’t disappearing, it’s splitting.

And for investors, that distinction matters.

The real divide inside software

For years, software was treated as a single category. In practice, it never was.

At one end sit systems of record (ERP, payroll, CRM, financial reporting, healthcare infrastructure, etc.). These systems hold a company’s operational truth. They are embedded, integrated, and painful to replace. Compliance and reliability aren’t optional.

On top of them sit workflow and insight layers: dashboards, automation tools, reporting overlays, summarization, recommendations, and routing.

AI applies unevenly across this stack.

Large language models are extremely good at pattern recognition and lightweight automation. That creates pressure on products whose core value is drafting, summarizing, or simplifying tasks.

But companies are not ripping out their ERP systems (at least yet). They’re adding intelligence on top of them.

Even leading AI-native firms still rely on incumbent systems of record. That tells you something important about where real durability sits. It reflects regulatory, integration, and audit constraints that are difficult to bypass.

This is where the bifurcation begins.

Why the market is repricing software

Traditional SaaS economics were built on per-seat pricing. More employees meant more licenses. Revenue was recurring and relatively predictable. That predictability justified revenue-based valuation frameworks.

Agentic AI disrupts that structure.

If AI executes workflows through APIs, companies may need fewer human seats. Pricing shifts toward hybrid models such as subscription plus usage, compute, workflow volume, or outcomes.

Operationally, that can be powerful. Financially, it introduces variability.

Usage-based revenue is harder to model than contracted ARR (annual recurring revenue). Forecast ranges widen. When visibility declines, markets demand a higher risk premium. Multiples compress.

Even modest NRR (net recurring revenue) compression — for example, from 120% toward 105–110% — materially alters valuation math over time.

At the same time, valuation frameworks are shifting. Revenue growth alone carries less weight. Cash flow durability, margin resilience, and operational leverage matter more.

The repricing reflects uncertainty about future durability, not a collapse in present demand.

Where Advantage Concentrates

As AI lowers the cost of building features, surface-level differentiation weakens.

For instance, a polished interface is less defensible if the interface becomes conversational. Standalone workflow tools are more exposed if agents can coordinate tasks across systems.

In this environment, durable advantage tends to concentrate around companies that:

• Control distribution.
• Own deeply embedded workflows.
• Sit on proprietary enterprise data — not for model training, but as the operational system of truth.
• Maintain governance and compliance infrastructure.
• Integrate best-in-class models rather than compete at the model layer itself.

As the model layer commoditizes, the value shifts upward to workflow ownership and data authority.

Importantly, AI monetization is still early. Experimentation is widespread, but clear, repeatable ROI remains uneven. The narrative is running ahead of the financial data.

That gap contributes to market volatility.

Importantly, AI could also expand overall software spend by automating workflows that were previously manual.

The debate is not about whether software grows, but where growth accrues and how durable it is.

Adoption will likely follow a familiar pattern

Enterprise transitions rarely happen overnight. Take cloud computing, for example, which moved gradually from peripheral use cases to mission-critical systems. Security concerns and migration friction slowed adoption before the model became dominant.

AI appears to be following a similar path but at a faster pace. Early traction is strongest in lower-risk workflows: marketing, customer support augmentation, internal analytics. Core financial and operational systems remain anchored in legacy platforms.

Adoption may move faster than prior cycles, but migration friction still exists.

This suggests dispersion, not collapse.

Private markets will not be spared

The same dynamics have sharper consequences in private markets.

In public markets, slower expansion primarily manifests as multiple compression. In private markets — where leverage, refinancing cycles, and exit timing matter — the same operational compression can translate into materially higher balance-sheet risk.

Many software assets acquired during the ZIRP era were financed with elevated leverage under assumptions of sustained growth and stable margins. AI-driven disruption does not need to eliminate companies to create stress. Margin compression and NRR deceleration alone can impair highly levered capital structures.

The SaaS reset is a valuation story and also, increasingly, a capital structure story.

Implications for underwriting

The right question is not whether SaaS survives, but rather which categories retain pricing power and durable cash flow.

Several adjustments follow:

• Lower seat-driven expansion.
• Underwrite usage volatility where pricing transitions from licenses to consumption.
• Prioritize platforms anchored in systems of record and embedded workflows.
• Focus on measurable economic outcomes from AI, not narrative positioning.
• Expect consolidation as strategic buyers acquire workflow depth and data assets rather than standalone point features.

AI does not need to eliminate companies to create pressure. Margin compression and slower expansion are sufficient to reshape valuations.

The opportunity lies in distinguishing structurally advantaged platforms from feature layers that are easier to replicate.

Dispersion, not extinction

Software is not disappearing. It is reorganizing around data authority, workflow control, distribution, and earnings durability.

Capital will not exit the category but rather will concentrate.

The next phase of software investing will reward selectivity. Broad exposure is unlikely to be sufficient. Category-level underwriting matters more.

The “SaaS is dead” narrative misses the nuance. What we’re seeing is a sorting mechanism.

SaaS is not dead.

But durability, not growth alone, now sets the bar.

Interested in learning more? A technical paper with further analysis can be accessed here.