SaaSpocalypse: AI’s Impact on Software Valuation, Retention, and Survival

The term "SaaSpocalypse" has been making the rounds, and for good reason. A confluence of AI developments and shifting investor sentiment has triggered the fastest market recalibration in software since the transition from license/maintenance to SaaS. For founders of software and AI businesses, it is worth taking a step back before drawing conclusions.

Markets and media love drama. What is happening now is better understood as a re-rating of what constitutes a quality software business, not a verdict on SaaS/software as an asset class. Similar resets have occurred before: after the 2008 financial crisis, during the shift from perpetual license to SaaS, and again in 2022 when rising interest rates abruptly refocused investors on profitable growth. This one is moving faster and with more intensity, but the core opportunity in software remains intact.

What Sparked the Disruption

The catalyzing event came in late January and early February with the release of Claude's Sonnet extended thinking model, specifically its application to the legal vertical. For years, large language models had proven useful for individual productivity tasks, but scaling that utility across an enterprise felt clunky and not yet ready for serious B2B deployment. That changed quickly.

On the day of the release, legal technology businesses traded down between 20 and 45 percent. The market was asking a pointed question: if a frontier model can perform core legal workflows out of the box, what is the case for software? Shortly after, similar releases targeted financial services, including banking and wealth management.

The broader implication was clear. The ability of LLMs to replace or augment B2B software is no longer theoretical. It is here now. That shift has fundamentally changed the questions investors ask, what they are willing to pay for, and how they think about risk in software portfolios. Private equity firms are re-examining plans for their existing portfolio companies. Strategic acquirers are asking harder questions about the technical defensibility of the businesses they evaluate. Founders are feeling a change in tenor from conversations that felt familiar just a year ago.

What Makes a Good Software Business Today

For roughly the past decade, the defining characteristics of a quality software business were largely stable. Strong growth, high gross margins, good retention, large addressable market, and favorable unit economics were the metrics that separated premium assets from the rest. That framework is shifting.

Based on conversations with hundreds of investors and strategic buyers, the following criteria are where the market is coalescing around what constitutes a truly competitive company in the current environment.

Vertical Beats Horizontal

Horizontal software businesses face meaningful headwinds. When a product optimizes a well-defined, repeatable workflow without significant nuance, frontier LLMs can perform that same function reasonably well. The value proposition of general-purpose software erodes in that scenario.

Highly verticalized businesses are in a much stronger position. Frontier models are trained to be broad. They skim across a wide range of knowledge rather than going deep in any single domain. A purpose-built ERP for a specialty trade, for example, may contain 14,000 specific part numbers with compatibility rules that reflect decades of accumulated domain knowledge. That depth is not something a general-purpose LLM will replicate in the near term. Verticalized software's edge is not just the product itself. It is the embedded expertise, the workflow specificity, and the institutional knowledge that the software encodes.

Gross Retention Is the New AI Risk Proxy

Gross retention has always mattered. It now matters more, and investors are treating it as a heuristic for AI exposure. The rough benchmark being used as a threshold is 90 percent gross retention. This is not a precise or absolute line, but it is a soft filter for many investors evaluating how defensible a customer base is against AI substitution. A business at 75 percent gross retention is a materially different investment profile than one at 90 percent, and the gap is being evaluated through an AI risk lens.

New bookings are the other critical metric. They signal that demand still exists from prospective customers who have been exposed to all available AI tools and alternatives. When a business is actively closing new logos in this environment, it could be a direct proof point that the value proposition is winning in a competitive market.

System of Record Status

Being a system of record means the software is the authoritative source of truth for the data that flows through it. Systems of record can coexist productively with AI agents. They can surface data to LLMs, respond to queries, and integrate into agentic workflows through MCPs. The key distinction is that the data they hold is the answer, not a suggestion. That structural role is increasingly valued by buyers who are thinking carefully about where software fits in an AI-augmented stack.

Defensible Data and Payments/Fintech Integration

A genuine proprietary data set remains a meaningful differentiator. Frontier models are ultimately designed to generate better outcomes with better data. Businesses that own unique, quality datasets are positioned to produce superior outputs from those models, which could create a compounding advantage.

Payments/Fintech integration is another noted defensive characteristic. Deep payments functionality is still an area where frontier models underperform, which provides a near-term buffer. That said, projecting that limitation out 12 to 24 months is a risky assumption to build a pitch around.

Agentic Capabilities or the Path to Them

Fully agentic AI businesses represent a small portion of the market today given how early the category is. However, investors are now asking whether a software platform has the architecture to distribute AI agents and deliver outcomes on behalf of customers. The broader shift underway is from software as a tool for employee/business efficiency toward software as a system that delivers outcomes** **directly. Platforms built to support that transition may have a structural advantage in the next phase of the market.

Go-to-Market Is More Important Than Ever

As the cost of building software continues to decline, technology differentiation becomes harder to sustain. This is not a reason for pessimism. It is a call to invest in the commercial capabilities of the business.

We believe the companies most likely to win in a world where code is increasingly commoditized are those with real go-to-market discipline: strong sales processes, channel strategy, and customer success practices that protect retention. These capabilities have often been underweighted in software businesses that succeeded on product merit alone. Going forward, they will likely be primary differentiators.

Verticalized businesses have a natural advantage here as well. Deep domain expertise makes the ideal customer more known, sales motion more credible, the product roadmap more defensible, and the customer relationship stickier.

TAM Is Not Shrinking. It Is Shifting.

A common concern is that if software prices compress as coding costs fall, the total addressable market contracts. This misses an important dynamic.

Software largely displaced services businesses by offering a higher-margin, more scalable way to make customers more efficient. In an agentic world, the motion reverses in a useful way. Rather than selling software that enables employees to do more, software and AI companies can sell the outcomes directly. That puts the much larger services TAM back on the table for technology businesses. Tasks and procedures that software companies historically declined to deliver, because executing them required significant headcount, can now be delivered by agents.

The near-term picture requires some nuance. Agents are still maturing, and the economics are still being worked out. But the directional opportunity may be significant, and the companies better positioned to capture it could be those building now for an outcome-driven commercial model.

Concrete Steps for Founders

Given how quickly the landscape has changed, founders evaluating their position should prioritize the following:

Know your retention metrics cold. Gross retention should be calculable by cohort and presentable with precision. This is the first question investors are asking, and the answer needs to be credible and well-supported.

Conduct a genuine AI product audit. A useful exercise is to engage directly with AI development tools and assess how much of your product can be recreated. Founders who have done this honestly often come away with a clearer picture of where their defensibility actually lies and where it does not. That clarity is essential before going to market.

Be precise about AI claims. When discussing AI capabilities with investors or strategic partners, specificity matters. Document exactly how AI is integrated, what it does, and what outcomes it produces. The bar for credibility is high, and overclaiming creates problems during diligence because they will ask for live demos

Get current market intelligence before running a process. The criteria investors use to evaluate software businesses have shifted substantially in a short period of time. What an investor told a founder they were looking for 12 months ago may not reflect how they are evaluating deals today. The market is moving quickly enough that current, direct exposure to buyer conversations is a meaningful advantage in positioning and process management.

Keep Perspective on the Cycle

The Gartner Hype Cycle exists for a reason. Every transformative technology passes through a period of overcorrection followed by a trough of disillusionment before reaching its actual equilibrium. Software will be no exception. Founders who treat the current moment as both a genuine signal and an opportunity to strengthen their businesses may find themselves better positioned on the other side.

The aperture of what qualifies as a quality, competitive software asset is narrowing. But for the businesses that fall within it, the investor and buyer universe competing for those assets appears large and motivated.

This material and the opinions contained herein are for general information only and are not intended to provide specific advice or recommendations for any individual or entity.