A quarter of the SaaS market, reclassified
Research firm Gartner published a forecast on 1 July 2026 putting a number on something enterprise software buyers have been sensing for a while: $234 billion in enterprise application software spending, roughly a fifth of the entire business-applications software-as-a-service (SaaS) market, is at risk of being redirected by 2030 as AI agents take over work that used to require a human logging into an application.
Gartner's name for the underlying mechanism is "agentic arbitrage": an AI agent completes a task by acting across several backend systems directly, rather than a person clicking through each application's own interface. George Brocklehurst, a managing vice president at Gartner and the lead analyst behind the forecast, put it plainly: agentic systems "deliver outcomes directly, bypassing traditional user experience (UX)-heavy applications and making the software invisible." His broader point is structural, not just about interfaces - this "breaks the link between user growth and revenue growth for many enterprise software vendors," the metric the SaaS business model has been built on for two decades.
Not a collapse, but a redirection
Gartner is reviving an old industry term, "Saaspocalypse," to describe the shift, but Brocklehurst is careful to temper it: "This is less an apocalypse and more of a metamorphosis." The numbers back that framing up more than the label suggests - overall enterprise application software spending is still projected to grow at roughly 12 percent a year through 2030, even as the composition of that spending changes underneath the total. The money is not leaving software; it is moving away from paying for seats and toward paying for completed outcomes.
Part of what is fueling the reassessment is Gartner's own skepticism about the current market: the firm estimates that about 70 percent of AI agents being sold today are "agent-washing," meaning products marketed as autonomous agents that do not actually have the independent decision-making capability the label implies. That matters for buyers trying to tell which vendors are positioned for the shift and which are just relabeling existing chatbots.
How buyers are already thinking differently
Brocklehurst frames the change as buyers moving away from thinking in software categories at all: "A CRM is not your sales process," he said, arguing that enterprises now want to buy a working outcome, like a completed sales cycle or a resolved support ticket, rather than a customer relationship management (CRM) license and the internal effort to operate it. Brocklehurst also points to institutional memory as a competitive line: "Better outcomes from AI require systems that can retain deep institutional memory and customer context over time," meaning vendors that can accumulate and reuse a customer's history have an advantage that a stateless point tool cannot match.
Gartner frames this as a fork in outcomes for vendors, not a uniform threat: companies that build genuine agentic, cross-domain workflow capability face what the firm calls a substantial revenue opportunity, while vendors that stay UX-centric and don't adapt face what it calls an existential threat. One category Gartner expects to benefit regardless of which vendors win: systems integrators and IT services firms, since making agentic workflows actually work across a company's existing systems takes new integration expertise that most in-house IT teams do not have on staff today.
Pricing is already moving
This is not purely a forecast about the future - three concrete examples of enterprise vendors restructuring how they charge are already underway. GitHub moved to token- and usage-based billing in June 2026. Zendesk rolled out a restructured outcome-based pricing model. Workday introduced a consumption-based pricing system it calls Flex Credits. Different mechanics in each case, but the same underlying pattern: subscription pricing tied to how many people log in is giving way to pricing tied to how much work actually gets done, which is precisely the shift agentic arbitrage would predict if Gartner's forecast holds.
What it means in practice
For IT and procurement teams, the practical move is to start asking vendors during renewal cycles how their pricing would change if usage shifted from named-seat logins to agent-driven, API-level activity, since a contract signed today on a per-seat basis may not reflect how the tool is actually used in two years. For vendors and software teams, Gartner's 70 percent "agent-washing" estimate is a useful gut check to run on your own roadmap before running it on a competitor's marketing: if your product's agentic capability would not survive a demo where a customer tries to remove the human from the loop entirely, that gap is exactly where a competitor - or a customer's own internal AI tooling - has room to arbitrage you.
