The SaaSpocalypse and What Survives It

Amna ManzoorPosted on May 25, 2026

15-16 Min Read Time

Key Takeaways

• The February 2026 SaaS selloff was not a market correction. Investors repriced software categories where AI agents can now perform the same work that companies were paying human seat licenses to handle.
• Per-seat pricing is under direct pressure in categories like customer support, sales engagement, and generic project management, where agent automation is already operational at scale.
• Software with deep industry-specific data, regulatory complexity, or embedded operational infrastructure is growing faster than horizontal platforms and retaining customers at three times the rate.
• The pricing shift is happening now. 43% of SaaS companies already use hybrid pricing models in 2026, and 60% of large IT contracts are expected to include outcome-linked clauses by year-end.
• Engineering teams are now building for two audiences simultaneously: the humans using their products today and the agents that will interact with them soon. That requires a different architecture than most teams are currently working from.

The Great Repricing of Enterprise Software

In the span of a few weeks in early 2026, the enterprise software sector lost roughly $2 trillion in market capitalization. Thomson Reuters posted its largest single-day decline on record. LegalZoom dropped nearly 20%. Atlassian fell 35%. Salesforce shed 28%. By late May 2026, the iShares Expanded Tech-Software ETF was still down roughly 12% year-to-date, trading well below its 52‑week highs

The press called it the SaaSpocalypse.

This was not a rotation driven by interest rates or macro fear. It was a reclassification. Wall Street looked at the pace of agentic AI progress and concluded that a large share of enterprise software was structurally overvalued. Software sold on a per-seat, per-month basis for work that AI agents can now perform autonomously. If an AI agent can do the work of ten people, a company has little reason to pay for ten seats.

Anthropic CEO Dario Amodei put it plainly at a recent enterprise briefing: some SaaS companies will go bankrupt. The ones that do not adapt will not survive the decade.

Why Per-Seat Pricing Is the Specific Problem

The SaaS business model spent two decades resting on a single assumption: more employees means more software revenue. As companies grew, they added seats and expanded contracts. It was a reliable engine that funded some of the highest valuations in technology history. It also made an unspoken bet that humans would always be the ones doing the work inside these tools.

Agentic AI broke that bet directly. AI agents reduce the number of human seats needed for a given workload, particularly in categories where per-seat counts were historically highest: customer support, sales engagement, marketing automation, document processing, and legal research. These are repetitive, stepwise workflows that agents now handle without a human in the loop.

The numbers from 2026 make the direction clear. Gartner projects that 40% of enterprise applications will feature AI agents by the end of 2026, up from less than 5% in 2025. IDC forecasts that 70% of software vendors will move away from pure per-seat models by 2028. Zylo's 2026 SaaS Management Index, drawn from $75 billion in tracked enterprise SaaS spend, shows spending on AI-native applications jumping 108% year over year, and 393% among large enterprises. Gartner also predicts that by 2026, 60% of large IT services contracts will include outcome-linked clauses, what the industry is now calling AI clawback provisions.

The bill is coming due inside renewal calls and contract negotiations right now.

Which Software Categories Are Genuinely at Risk

The SaaSpocalypse narrative flattened what is actually a more nuanced picture. How exposed a product is depends on a simple question: does it make money because humans are doing things inside it, or because it holds data and processes that teams cannot easily walk away from?

The Most Exposed Workflows

Products whose value maps directly to how many humans are active are under immediate pressure.

Customer support platforms are a clear example. If an agent can open a ticket, draft a response, log a CRM entry, and route an escalation without a human touching a keyboard, the software underneath that workflow loses its economic justification. Intercom has already moved to a $0.99-per-resolved-ticket pricing model, an explicit acknowledgment that charging for human seats when agents are doing the resolving no longer makes sense.

Generic CRM and horizontal project management tools face similar pressure. Their core job was helping humans track, update, and coordinate work; agents now do all three. Atlassian and Salesforce saw some of the steepest declines in the February selloff partly because their workflows are exactly the ones agents automate best.

The Categories With Durable Structural Protection

The companies holding up well share a different profile. They own systems of record, proprietary data, and operational infrastructure that are hard to replicate and hard to leave.

Vertical software built for industries with serious compliance requirements sits in a different position. Healthcare, legal, and financial services have regulatory complexity that agents cannot simply route around. A platform built for radiology practices, trained on millions of scans from that specific clinical context, produces outputs that a general-purpose AI tool cannot match. A legal platform built on decades of case data and citation patterns carries the same advantage. Vertical SaaS is growing at 18–22% CAGR in 2026 versus 12–15% for horizontal platforms, and retains customers at three times the rate, according to analysis across 200 companies.

Developer tools and infrastructure software are accelerating engineers rather than replacing them, which pushes usage up. Cybersecurity platforms operate in adversarial environments where live human judgment and real-time context still matter. And the core operational systems that enterprises run on ERP platforms, financial ledgers, and databases, face a different dynamic entirely. Agents still need somewhere reliable to read from and write to, and those systems hold that position.

The SaaSpocalypse Is Separating Fragile Software From Durable Platforms

What is happening in 2026 is a separation between fragile, seat-dependent software and durable platforms strengthened by AI-driven workflows.

Pressure on Seat-Based Software Models

Products tied closely to human seat counts are under immediate pressure from agentic AI. Dario Amodei's warning was not theoretical, Monday.com has already replaced 100 sales development representatives with AI agents. Enterprise buyers are entering renewal negotiations with leverage they did not have two years ago. EY research shows value-to-cost satisfaction for per-seat software has fallen below 40%, and companies now have a credible alternative to expanding headcount and software seats in parallel.

The Breakdown of the Seat Expansion Logic

For two decades, growth in employee count translated directly into growth in software revenue. Agentic AI weakens that relationship because a growing share of operational work can now be completed autonomously. When one agent handles workflows previously distributed across multiple employees, the logic behind aggressive seat expansion becomes harder to sustain. Investors reacted sharply in the February selloff because markets are repricing software categories where revenue growth depends heavily on increasing human usage.

Why the “Software Is Dead” Narrative Is Misleading

At the same time, the broader “software is dead” narrative misses what is happening inside the strongest platforms. NVIDIA CEO Jensen Huang has called the idea that software is becoming irrelevant one of the most illogical narratives in tech. A March 2026 analysis of 40 SaaS earnings calls pointed in the same direction: companies with deep, industry-specific data and embedded operational workflows reported AI as an accelerant to their business.

Why Agents Increase the Value of Strong Platforms

Agents rely on structured systems, reliable datasets, operational context, and trusted environments to function effectively. Platforms that hold those layers are becoming more important because they provide the infrastructure agents depend on to make decisions and execute tasks safely. Mary Meeker and the Bond Capital team have echoed the same theme: platforms with rich contextual data and deeply integrated workflows are gaining influence as AI adoption expands.

What Will Actually Survive the Transition

The companies likely to emerge stronger from this transition are not resisting AI. They are treating it as a long-term product and architecture shift. They are building systems that humans and AI agents can operate through together, with data and infrastructure as the foundation of durable competitive advantage.

How the Pricing Model Is Actually Changing

The companies surviving this shift are not waiting for the market to settle. They are changing how they charge for their products entirely, and the early moves are instructive.

What the Early Movers Are Doing

Salesforce launched its Agentic Enterprise License Agreement, offering fixed-price access to Agentforce, Data 360, and MuleSoft rather than per-seat billing. In Q4 fiscal 2026, Salesforce delivered 2.4 billion Agentic Work Units, discrete tasks completed autonomously by agents. The metric has shifted from who logs in to what gets done.

ServiceNow has repositioned itself as a coordination layer sitting above departmental workflows, with a shared system connecting siloed teams. The bet is that as AI agents become more common, the infrastructure needed to coordinate them becomes more valuable, not less.

Across the market, 43% of SaaS companies already use some hybrid model in 2026, combining a fixed base fee with variable consumption pricing, and that figure is projected to reach 61% by year-end. 31% of enterprise software firms expect outcome-based pricing to become their primary model by mid-2029.

What This Means for Buyers

For engineering and product leaders on the buying side, every renewal is now a different kind of conversation.

The software stack most companies are running today was assembled for a world where humans drove every workflow. If agents are now handling a growing share of that work, the contracts signed last year may not reflect what is actually being delivered. Vendors going through this pricing shift often want early customers willing to try new models in exchange for better terms. That is a different conversation than asking for a discount and it is worth having at the next renewal. 

What This Means for How You Build

The SaaSpocalypse looks like a vendor pricing problem on the surface. Underneath, it is a product and architecture question for every engineering team building something today.

Building for Two Audiences at Once

The question for product teams is what your product looks like when the users are increasingly agents rather than humans.

Agents do not click buttons; they call APIs. They do not scan dashboards; they query structured outputs. They do not read documentation the way a new hire would; they parse schema. Gartner is precise on this point: products built with composable microservices, API-first, cloud-native, headless architectures will establish a significant competitive advantage through 2028. Engineering teams today are building for two audiences at once – the humans using their products now and the agents that will interact with or operate around them soon.

That is a different design problem than the one most teams are currently working from.

The Data Layer as the New Competitive Advantage

The teams that will be in the strongest position are building products where the data they hold becomes more valuable over time, not less.

Companies that build curated datasets, enrichment APIs, and specialized knowledge bases are becoming the infrastructure that AI agents depend on. By 2028, Gartner projects that 90% of B2B buying will be AI-agent-intermediated, pushing over $15 trillion in B2B spend through AI agent exchanges. The products that sit in the path of those agents, holding the data and context they need to operate, will have advantages that workflow-based point solutions cannot replicate.

The companies most likely to win three years from now are the ones reading this shift early and building their product and data strategy around it before it becomes obvious.

Frequently Asked Questions

1. What caused the SaaS market selloff in early 2026?

Wall Street repriced software categories where AI agents can now perform the same work that per-seat licenses were being charged for. It was a structural reclassification, not a reaction to interest rates or macro conditions.

2. Which SaaS companies are most at risk from AI agents?

Companies generating revenue primarily from human seat counts in repetitive, step-by-step workflows. Generic CRM, customer support platforms, sales engagement tools, and horizontal project management software are the most exposed categories right now.
 

3. Which software categories are likely to survive and grow through this transition?

Vertical software built for regulated industries like healthcare, legal, and financial services is holding up well, as are developer tools, cybersecurity platforms, and core operational systems. These are categories agents depend on rather than replace.
 

4. How is SaaS pricing actually changing in 2026?

The market is shifting from pure per-seat billing toward hybrid models combining a fixed base fee with variable consumption or outcome-based pricing. 43% of SaaS companies already use some form of hybrid pricing in 2026, projected to reach 61% by year-end.
 

5. What should engineering teams do differently as AI agents become more prevalent?

Design products that work for agents as users, not just humans. Agents call APIs and parse schema rather than clicking through interfaces. Products built with API-first, composable, cloud-native architectures will have a structural advantage through 2028 and beyond.