What About the SaaSpocalypse?

Software stocks are recovering, SaaS revenues are growing, and the predicted collapse never arrived. Was the market wrong about AI's impact on software? Not entirely.

Earlier this year, the software industry appeared to be facing an existential crisis. The narrative was compelling. AI could write code, build applications, automate workflows and increasingly perform tasks that previously required expensive software subscriptions. If software could suddenly be built at a fraction of the cost, why would businesses continue paying a premium for it?

Investors reacted accordingly. Software stocks sold off, valuations compressed, and the term "SaaSpocalypse" quickly entered the vocabulary. In May, we argued that the market was drawing the wrong conclusion. AI wasn't making software obsolete. It was fundamentally changing where value in software resides.

Five months later, the market is beginning to reflect that distinction.

The SaaSpocalypse That Never Quite Happened

The numbers tell an interesting story. According to Hargreaves Lansdown, the average software stock in its S&P 500 sample declined 27% from the beginning of 2026 to its April low. Since then, the sector has recovered approximately 94% of those losses. 

Meanwhile, the underlying business of selling software has proven remarkably resilient. Dealroom's September analysis found that aggregate SaaS payment growth accelerated to 31.5% year-on-year, compared with 25.3% at the end of 2025, based on Stripe's payment data.

The anticipated collapse in demand simply hasn't materialized. Yet something equally important has happened beneath the surface. While share prices have recovered, valuations have not returned to their previous levels.

Dealroom found that the median listed business-software revenue multiple stood at 2.34x in August, compared with 2.93x a year earlier. And Hargreaves Lansdown's broader software sample showed that forward earnings multiples remained substantially below their starting point for the year. With other words: the market has regained confidence in software's earnings, but not necessarily in its old valuation model. That distinction matters.

The Market Was Wrong About the Timing, Not the Direction

The original SaaSpocalypse narrative confused two very different developments: the falling cost of building software and the declining value of owning software. They are not the same thing.

AI has dramatically reduced the effort required to develop applications. But developing an application is only one part of delivering enterprise software. Deploying it reliably, integrating it into existing systems, maintaining security, managing regulatory requirements and supporting business-critical processes remain difficult.

A company might be able to reproduce the interface of a CRM system in a weekend. Reproducing the integrations, customer history, operational processes, security model and organizational adoption behind it is an entirely different proposition. That explains why enterprise customers haven't suddenly abandoned their existing platforms.

It also explains why the impact of AI is becoming increasingly uneven.

Three Categories, Three Different Futures

In our original analysis, we argued that AI would divide the software industry into three distinct categories. The developments of recent months have reinforced that thesis.

1. Infrastructure: The Compounding Layer

Every new AI application and autonomous agent requires infrastructure. Compute, storage, databases, identity management, security and integration capabilities become more important as the number of applications and automated processes increases.

Rather than replacing this layer, AI expands demand for it. The market has increasingly recognized this dynamic. Infrastructure, cybersecurity and data platforms have been among the strongest beneficiaries of the software recovery. Their value is not primarily derived from how difficult their interfaces are to reproduce, but from the scale, reliability and critical capabilities they provide.

2. Commodity SaaS: The Structural Squeeze

The situation is very different for software products built around relatively simple functionality. Basic dashboards, scheduling tools, form builders and lightweight workflow applications are becoming easier to reproduce or replace.

Customers may not immediately abandon these products. Switching still involves effort, and an established solution can remain more convenient than building an alternative. But the economics are changing.

When comparable functionality becomes cheaper to develop, barriers to entry fall and competitive pressure increases. Over time, this affects pricing power, margins and ultimately valuation multiples.

The SaaSpocalypse hasn't eliminated commodity SaaS. It has exposed how much of its historical value depended on the difficulty of building alternatives.

3. Mission-Critical Software: The Enduring Foundation

At the other end of the spectrum are the systems businesses depend on to operate. ERP, CRM, financial administration, supply chain management and other mission-critical applications represent considerably more than collections of features. They contain years of accumulated business logic, transactional history, integrations, permissions and organizational processes.

Replacing them isn't simply a development exercise. It is an operational transformation carrying significant cost and risk. This gives established enterprise platforms a degree of resilience that the original SaaSpocalypse narrative largely overlooked. But resilience should not be confused with immunity. The greatest threat to traditional enterprise software may not be replacement by AI, but a fundamental change in how users interact with it.

From System of Record to System of Agency

For decades, enterprise software has been designed around human interaction. Employees navigate interfaces, enter information, interpret reports, initiate workflows and make decisions. Software provides the environment in which people execute business processes. AI is reversing that relationship.

Instead of employees navigating multiple applications to complete a task, intelligent agents can increasingly interact with those applications on their behalf. Consider a sales process. Rather than manually updating CRM records, checking inventory in ERP, preparing a quotation and coordinating follow-up activities, an AI agent could orchestrate much of that process across multiple systems, subject to appropriate permissions and human oversight.

The underlying applications remain essential. They continue to hold the authoritative data, enforce business rules and execute transactions. But the primary interface gradually shifts away from the application itself. Systems of record remain the foundation. Systems of agency become the layer through which work gets done.

This creates a fascinating strategic question for incumbent software vendors. If users spend less time inside an application's interface, does the vendor still command the same pricing power? And who captures the economic value created by an intelligent agent operating across several platforms? These questions are arguably more consequential than whether AI can reproduce a software product's features.

The Next Valuation Reset Is About Value Capture

Historically, SaaS valuations have been supported by predictable recurring revenue, high gross margins, strong retention and the ability to expand subscriptions as customers grow. Those characteristics haven't disappeared, but AI introduces new uncertainties.

Seat-based pricing becomes more complicated when a single agent can perform work previously distributed across multiple users. Feature-based differentiation weakens when capabilities can be replicated more easily. And the boundary between software, automation and services becomes increasingly blurred.

At the same time, new sources of value emerge. Software providers that control mission-critical data, orchestrate complex processes, integrate across enterprise environments and deliver measurable business outcomes may become more valuable.

The question shifts from how many people use a product to how much work the product enables or performs. This does not mean every software vendor will move to outcome-based pricing, nor that traditional subscriptions will disappear. But it does mean the relationship between software usage, business value and monetization is being reconsidered.

For investors, that changes the valuation equation. Revenue growth and retention remain important, but they need to be evaluated alongside a company's ability to defend its position and capture value in an AI-driven operating model.

So, Is the SaaSpocalypse Over?

As a stock market panic, largely yes. As a structural transformation of the software industry, no.

The recovery has demonstrated that enterprise software is considerably more resilient than the initial narrative suggested. Businesses aren't abandoning their systems, and AI isn't eliminating the need for reliable, integrated applications. But the recovery should not be mistaken for a return to business as usual.

The cost of creating software continues to decline. Intelligent agents are changing how work gets executed. And the competitive advantage of owning an application is gradually shifting away from its features toward its data, integration, operational importance and ability to deliver outcomes.

The industry is not facing a uniform collapse. It is undergoing a redistribution of value.

At HIPER, we believe the next generation of enterprise technology will combine the reliability of established ERP and CRM platforms with the flexibility and intelligence of AI-driven execution. The opportunity isn't simply to replace existing systems or add another AI feature to them. It is to transform how businesses operate on top of those systems. The SaaSpocalypse may be over. The reinvention of enterprise software has only just begun.