Agentic arbitrage could impact one-fifth of enterprise SaaS spend by 2030

• Legacy SaaS market share will be cannibalised by incumbents and taken by new entrants delivering horizontal agentic platforms.

A quiet revolution is unfolding inside the enterprise. It does not announce itself with flashing dashboards or yet another SaaS subscription. Instead, it moves through systems, across workflows, and past interfaces — invisible, autonomous, and increasingly in control.

According to new research from Gartner, Inc., this revolution has a price tag attached: up to $234 billion in enterprise application spending will be exposed to what analysts are calling agentic arbitrage between now and 2030. By the end of the decade, that figure will represent roughly 20 per cent of all enterprise SaaS expenditure.

It is a number large enough to reset the conversation about what enterprise software actually is — and what it is about to become.

What agentic arbitrage actually means

The term agentic arbitrage describes a deceptively simple dynamic. When AI agents are capable of completing tasks across multiple systems — logging into platforms, extracting data, making decisions, triggering workflows — the human user no longer needs to interact with each individual software interface.

The agent does the clicking, the navigating, and the orchestrating. The software beneath becomes, in effect, invisible.

“Agentic AI changes the economics of software,” explains George Brocklehurst, Managing Vice President at Gartner. “Agentic systems deliver outcomes directly, bypassing traditional user experience-heavy applications and making the software invisible. This breaks the link between user growth and revenue growth for many enterprise software vendors.”

That broken link is the heart of the disruption. For decades, SaaS vendors have operated on a straightforward premise: more users mean more seats, and more seats mean more revenue. Agentic AI dismantles that premise. When a single AI agent can perform the work of dozens of licensed users, the traditional per-seat pricing model loses its foundation. The question is no longer how many people need access, but what outcomes can be delivered autonomously .

The SaaSpocalypse, redefined

The term “Saaspocalypse” has circulated in industry circles for years, often deployed as a dramatic shorthand for the disaggregation of the legacy SaaS market. But Gartner’s analysis suggests the metaphor needs updating.

This is less an apocalypse and more of a metamorphosis. The SaaS industry is not being destroyed — it is being reshaped into something that looks fundamentally different from the subscription-based, dashboard-centric model that has dominated enterprise IT for the past two decades.

“This shift will lead to a redefinition of ‘Saaspocalypse,'” Brocklehurst notes. “SaaS will not be destroyed; it will emerge in a different form.” That emerging form carries both existential threats and extraordinary opportunities, depending on where a company sits in the value chain and how quickly it can adapt .

The metamorphosis is already visible in market signals. In early 2026, the software sector experienced turbulence that erased hundreds of billions in market capitalisation, with investors beginning to price in the risk that AI agents pose to conventional SaaS business models .

The question driving the repricing is stark: if an AI agent can orchestrate work across platforms without ever opening a traditional application, what is the durable value of that application’s user interface?

The buyer’s new calculus

Enterprise buyers are shifting their expectations accordingly. The era of purchasing more tools, more dashboards, and more feature-rich interfaces is giving way to a more demanding calculus. Organisations increasingly evaluate technology not by what it displays but by what it delivers.

“Enterprise buyers will deemphasise buying more new tools or dashboards,” says Brocklehurst. “They want better outcomes, and adding more AI features often creates more cost, not better outcomes.”

The distinction is critical. Simply layering AI capabilities on top of existing interfaces does not address the underlying shift. What matters is whether a system can retain deep institutional memory and customer context over time — the kind of persistent, contextual intelligence that allows agents to act with genuine autonomy and judgment .

This emphasis on institutional memory marks a departure from the data-centric mindset that has dominated enterprise technology. Data, after all, is abundant. What is scarce — and what agentic systems are uniquely positioned to capture — is knowledge: the contextual understanding of how an organisation operates, what its customers need, and how decisions ripple across interconnected workflows.

The strategic implications split along familiar lines, but with an unfamiliar urgency. Incumbent software vendors face a dilemma that is at once existential and deeply practical. Their existing revenue models, built on seat licenses and interface-driven value, are the very structures being undermined.

To survive the metamorphosis, they must move from selling access to software toward selling outcomes enabled by software. That requires embedding agentic capabilities directly at the point of execution — not as an add-on feature, but as the core value proposition.

“As organisations increasingly use agentic AI systems, the user interface is no longer a differentiation,” Brocklehurst observes. “Legacy SaaS market share will be cannibalised by incumbents and taken by new entrants delivering horizontal agentic platforms.” The warning is pointed: defend the dashboard at your own peril .

For AI-native startups and service providers, the opportunity is substantial. These companies can position themselves as the agentic layer that sits across enterprise systems, orchestrating cross-domain workflows without asking users to learn yet another interface.

Their value proposition centres on measurable outcomes rather than feature lists, and they can assist organisations in redesigning workflows around AI-native patterns rather than bolting AI onto legacy processes.

“Ultimately, they can capture not just existing spend, but incremental budget unlocked through ROI upside,” Brocklehurst notes. This points to a crucial dynamic: the $234 billion at risk is not necessarily money that disappears from the technology ecosystem. Much of it will be redirected — toward platforms, services, and solutions that enable agentic workflows, and toward the measurable business results those workflows generate.

What comes next

The transformation is still in its early stages, but its contours are clear. Today, delivering agentic solutions that provide autonomous end-to-end workflow execution and cross-system orchestration typically requires heavy services engagement. That dependency will diminish over time as platforms mature and as organisations build the internal capabilities to design, deploy, and govern agentic systems at scale.