For decades, the software-as-a-service model has been the envy of every industry — predictable revenue, massive scale, and gross margins that routinely reach between 70 and 90 percent. That enviable position is now under siege, not from a rival software paradigm, but from artificial intelligence tools capable of replacing the very workflows that SaaS platforms were built to support.
The structural threat became tangible when Klarna publicly announced in late 2024 that it had abandoned Salesforce's flagship CRM product in favor of a proprietary, AI-driven system developed in-house. What was once a headline-grabbing exception is rapidly becoming a credible option for a growing number of enterprises — and capital markets have taken notice.
In early February, an investor sell-off erased nearly $1 trillion in market value from software and services stocks, with further losses arriving later that same month. Analysts have taken to calling the phenomenon the SaaSpocalypse, while at least one market observer has coined the term FOBO investing — shorthand for fear of becoming obsolete. The anxiety is neither irrational nor unfounded, though its full implications remain contested.
At the center of the disruption sits a deceptively simple shift in how enterprises think about building versus purchasing software. Historically, buying an enterprise platform from Salesforce, Workday, or a comparable vendor was the path of least resistance. The economics of custom development made in-house alternatives impractical for all but the largest technology firms. That calculus is changing fast.
"The barriers to entry for creating software are so low now thanks to coding agents, that the build versus buy decision is shifting toward build in so many cases," said Lex Zhao, an investor at One Way Ventures.
Zhao's observation was prompted by a founder who texted him with a pointed update: the company was replacing its entire customer service team with Claude Code, an AI tool capable of writing and deploying software autonomously. The message carried a broader implication — that platforms like Salesforce are no longer the automatic default for enterprise software needs.
The threat to SaaS extends well beyond the build-versus-buy debate. The entire pricing architecture of the industry rests on a per-seat licensing model, whereby vendors charge based on the number of employees actively using the software. When AI agents can execute the same workflows that once required dozens of human users, that revenue foundation begins to erode in fundamental ways.
"SaaS has long been regarded as one of the most attractive business models due to its highly predictable recurring revenue, immense scalability, and 70-90% gross margins," noted Abdul Abdirahman, an investor at the venture firm F-Prime.
The disruption is compounded by the speed at which AI tools are replicating not just core SaaS functionality but also the value-added features vendors rely upon to expand revenue from existing accounts. Tools such as Claude Code and OpenAI's Codex can now approximate capabilities that SaaS companies spent years developing and monetizing. This dynamic hands enterprises a powerful negotiating instrument: the credible threat to build their own solution.
"Even if they do not take the build route, this creates downward pressure on contracts that SaaS vendors can secure during renewals," Abdirahman added.
Public markets have become unusually sensitive to each new AI product announcement. The pattern is perhaps best illustrated by Anthropic's recent release cadence. When the company launched Claude Code for cybersecurity applications, shares in cybersecurity-adjacent software firms declined. When it introduced legal tools through Claude Cowork AI, the iShares Expanded Tech-Software Sector ETF — a basket that includes companies such as LegalZoom and RELX — also dropped in response. Each product launch functions as a stress test for existing SaaS valuations.
Part of this sensitivity reflects a longer-standing vulnerability. SaaS companies performed much of their growth during the zero-interest-rate era, when cheap capital allowed aggressive expansion. As borrowing costs have risen, the cost of doing business has followed, exposing valuations that many investors now regard as inflated even before accounting for AI disruption.
The deeper issue is one of terminal value — the long-run estimate of a company's worth that underpins how analysts price software stocks. When it is genuinely unclear whether enterprises will rely on SaaS platforms to the same degree in one year or five, every new AI capability announcement triggers a reassessment.
"This may be the first time in history that the terminal value of software is being fundamentally questioned, materially reshaping how SaaS companies are underwritten going forward," Abdirahman said.
Meanwhile, a new generation of AI-native startups is scaling at a pace that legacy SaaS vendors cannot easily match. Software has become easier and cheaper to build, and therefore easier to replicate, as Yoni Rechtman, a partner at Slow Ventures, pointed out. That dynamic favors nimble entrants while putting legacy incumbents — who spent years and considerable capital constructing their technology stacks — at a structural disadvantage.
The emerging business models that might replace per-seat pricing are still taking shape, and the market has not yet accumulated enough evidence to judge their long-term viability. Some AI companies are adopting consumption-based pricing, charging customers according to how much AI capacity they use, measured in tokens. Others are pursuing outcome-based pricing, in which fees are tied directly to the measurable performance of the AI system.
The outcome-based approach has found an early and notable champion in Sierra, the AI startup led by former Salesforce CEO Bret Taylor. Sierra offers AI-powered customer service agents and prices them according to results delivered — positioning the company as a direct, if indirect, challenger to Salesforce's model. The strategy appears to be gaining traction: in November, Sierra hit $100 million in annual recurring revenue in less than two years.
The IPO market reflects these tensions clearly. A Crunchbase report released in early 2026 found that, while initial public offering activity is beginning to recover in certain sectors, there are no venture-backed SaaS filings on the immediate horizon and none expected imminently. Large private SaaS companies such as Canva and Rippling face a difficult combination of pressures: a narrow IPO window, elevated expectations tied to AI advancement, and the turbulent stock performance of their already-public peers.
"Nobody wants to be subjected to the volatility of public markets when sentiment can send companies into downward tailspins," Rechtman said, adding that he expects many of these companies to remain private for considerably longer.
Some mid-size SaaS companies have encountered difficulty even raising extension rounds in private markets, according to Aaron Holiday, a managing partner at 645 Ventures, as private investors share the same concerns driving the public sell-off. The window for both entry and exit has narrowed considerably across the sector.
All eyes are now on whether OpenAI and Anthropic — both reportedly contemplating IPOs, possibly later this year — will provide public markets with a clearer picture of what AI-native company finances actually look like. Their filings, when they arrive, will likely serve as a critical reference point for repricing the entire software landscape.
Despite the alarm, veteran investors are careful to distinguish between structural change and market overreaction. Holiday describes the current moment not as the death of SaaS but as an old snake shedding its skin — a painful but ultimately regenerative process. He argues that enterprises will continue to require software that satisfies compliance requirements, supports audits, manages complex workflows, and offers the kind of operational durability that experimental AI tools have not yet demonstrated.
"The most important thing to understand about the SaaS pullback is that it is simultaneously a real structural shift and potentially a market overreaction," Abdirahman observed, noting that investors typically "sell first and ask questions later."
The most probable outcome, as Holiday and others suggest, is a convergence of old and new — enterprises retaining SaaS infrastructure where compliance and durability demand it, while adopting AI-native tools for functions where speed and cost efficiency take priority. What will not survive, investors argue, is the assumption that the SaaS model is impervious to disruption.
"Durable shareholder value isn't built on hype," Holiday said. "It's built on fundamentals, retention, margins, real budgets, and defensibility."
