Listen to the Article
Markets punished software stocks this quarter, but the contracts did not. That gap affects valuations, deal velocity, and the choices product leaders make over the next twelve months. The signal is clear. The sector is not collapsing. It is sorting. Software that only wraps a database is at risk. Software that runs a critical workflow is not.
Two Sets Of Numbers
Start with what spooked everyone. The selloff built through a sequence of events: a rare Salesforce revenue miss in late 2025, then January 29, 2026, when OpenAI released Project Operator and triggered structural repricing across the category. On that single day, Microsoft reported $81.3 billion in quarterly revenue, a 17% year-over-year increase and a ninth consecutive earnings beat, and shed $357 billion in market capitalization by the close. ServiceNow beat estimates for the ninth consecutive quarter and fell 11%.
By early February, the S&P 500 Software and Services Industry Group had declined 20% year-to-date and 28% from its October high. JPMorgan and Bank of America analysts both described it in client notes as an “indiscriminate selloff” driven by fear of structural disruption, not by financial performance.
Now look at operating reality. The global SaaS market reached $408.21 billion in 2025 and is forecast to reach $465.03 billion in 2026, according to Precedence Research’s January 2026 report, representing 14% year-over-year growth amid equity market turbulence.
Gartner’s most recent IT spending forecast projects software as the fastest-growing IT spending category in 2026, with 15.2% year-over-year expansion. Stock prices discounted uncertainty about the future. They did not record a collapse in the present.
That divergence is not unusual in software. Markets price where cash flows might move next. Customers pay for what is working today. The tension will resolve as buyers reallocate budgets across tools, not away from software altogether.
Not All SaaS Is Equal
Under the headline growth, a selection is taking place. The split is not between SaaS and AI. It is between replaceable tools and durable systems.
Replaceable tools are thin wrappers on data. They store, retrieve, and present information with a tidy interface. Think simple CRMs, form builders, basic project trackers, and standard dashboards. AI agents are already competent at those tasks. If a workflow looks like create, read, update, and delete with predictable formatting, an agent can handle it without a stand-alone product.
Durable systems do more than present data. They execute and govern complex workflows. They embed policy, compliance rules, and financial logic. They touch revenue recognition, payroll, tax, inventory accounting, quality control, and regulated communications. An answer that is right 6 out of 10 times cannot close the books, file a return, or meet an audit. These products are not being replaced. They are absorbing AI inside the guardrails they already enforce.
Gartner predicts that 40% of enterprise applications will be integrated with task-specific AI agents by the end of 2026, up from less than 5% in 2025, a transformation the firm describes as automating single-purpose workflows while increasing the value of platforms that coordinate multi-step processes across systems. The common thread among resilient products is responsibility. They carry it, prove it, and price it.
Pricing Will Decouple From Seats
The per-seat model strains when one person with AI support does the work of five. Seats stop mapping to value. Expect pricing architecture to keep shifting toward usage and outcomes.
The trend is already visible. OpenView’s research shows that more than 60% of SaaS companies now offer some form of usage-based billing, up from 27% in 2018, with Chargebee’s 2025 State of Subscriptions Report finding hybrid model adoption at 43% today and projected to reach 61% by the end of 2026.5 This is not a fundamental break in the SaaS model. It is a routine refactor of how revenue maps to utility.
As AI augments work, pricing will migrate to clear meters: transactions processed, workflow runs completed, documents classified, events ingested, or dollars under management. Per-seat can still apply to role-based capabilities. The core meter should tie to the economic unit customers defend in budget reviews.
The AI-Native Reality Check
There is also a reason the narrative that AI will eat SaaS keeps stalling. Many AI-native products have weak unit economics at scale. Gross margins are compressed by inference costs, frequent model updates, and heavier customer success requirements. ICONIQ’s January 2026 data shows AI-native gross margins averaging 52%, up from 41% in 2024, while Bessemer Venture Partners found that scaling AI companies average 25% gross margin at the growth stage, compared to the 75 to 80% that traditional B2B SaaS companies typically achieve at scale.6 The structural gap is narrowing but remains material.
Low retention is another warning sign. A sizable portion of booked AI revenue looks like pilot budgets, which are easy to cut when the next quarter’s priorities shift. That does not mean AI is a fad. It means buyers are still testing where it belongs in production and what reliability threshold they need before signing multi-year deals.
The takeaway is not to dismiss AI. It is to ground product and pricing in outcomes that customers can measure and defend. When results are tied to audited processes or contractual service levels, retention and margins improve because buyers are paying for more than a demo.
How Durable SaaS Wins Now
Resilient vendors are already changing how they build, package, and sell.
Anchor on high-stakes workflows. Pick processes where errors have financial, legal, or safety consequences. Encode the steps, exceptions, and approvals. Make the system of record the single source of truth.
Make responsibility visible. Publish explicit service levels for accuracy, timeliness, and auditability. Provide evidence packs that customers can hand to finance, risk, and regulators. Responsibility is the moat that thin tools do not have.
Put AI inside guardrails. Use AI for classification, extraction, recommendations, or drafting. Always bind agents to policy, role permissions, and data boundaries. Log decisions. Give admins the ability to replay and review outcomes.
Redesign packaging and pricing. Move the primary value meter to a unit customers care about. Keep role-based add-ons for admin and advanced analytics. Introduce committed-use discounts tied to outcomes, not only seats.
Prove dollar impact with finance-grade metrics. Point to cycle-time reduction on a defined workflow. Show error-rate deltas against a baseline. Quantify rework cost avoided or late-fee exposure eliminated. Tie results to the general ledger where possible.
Track the right KPIs internally. Monitor net revenue retention, gross margin after model and infrastructure costs, blended CAC payback, and the ratio of seats to usage. Watch the attachment rate of modules that carry responsibility, because those modules predict durability.
Prepare for heavier procurement. Expect deeper reviews on data residency, model provenance, third-party subprocessor lists, and indemnities. Provide clear documentation up front. Reduce friction with templated answers that legal and security teams accept.
The Signal In The Selloff
Investors compressed multiples on tools that look easy for agents to replicate. They will pay up for platforms that encode policy and compliance into day-to-day operations. That split will widen as AI performance improves, because better models raise the bar for presentation-only software while increasing the value of systems that make high-stakes work safer and faster.
A go-to-market shift follows. Messaging must move from features to process integrity. The economic buyer tilts toward operations, finance, and risk leaders who care about reliability. Partnerships should favor integrators who own the surrounding business process, not just cloud marketplaces. Category design should reflect the outcome, not the interface.
Why The Thesis Matters
Buyers are not abandoning subscriptions. They are eliminating purchases that do not change how work gets done. Products that organize information will be consolidated or absorbed by agents that come bundled with larger platforms. Products that execute, verify, and document the work will keep winning budget.
This has valuation implications. Revenue tied to responsible workflows tends to carry higher net revenue retention, stickier multi-year terms, and pricing power as complexity grows. Revenue from presentation layers faces deflation as agent bundles expand. Even when growth rates look similar for a few quarters, the underlying quality of revenue is not the same. CFOs, boards, and acquirers will price that distinction in.
A More Useful Frame Than The Headlines
The SaaS market is growing. Enterprise software budgets are expanding. Buyers are drawing a sharper line between tools that tidy the bookshelf and systems that run the library. Sectors like telecom, where AI is already reshaping how operators structure procurement, compliance workflows, and vendor accountability, illustrate exactly how quickly that line is being redrawn in practice.
The pace of this selection depends on two variables. First, how quickly agent reliability crosses the thresholds that procurement, audit, and regulators will accept. Second, how fast pricing models move from seats to usage without breaking sales productivity or customer predictability. Both remain unresolved. That is precisely where the strategic risk sits: not in whether durable software survives, but in whether vendors move fast enough to prove process ownership before procurement teams decide the category is commoditized. The vendors who stall on that question will not be acquired at a premium. They will be replaced by the platform that already answered it.
