July 31, 2026
Gartner's July 2026 report puts a hard number on something a lot of us have been saying anecdotally for a year: $234 billion in enterprise SaaS spending -- about 20% of all enterprise application SaaS spend -- is at risk from what Gartner calls 'agentic arbitrage' between now and 2030, according to [beri.net's coverage of the report](https://www.beri.net/article/234-billion-saas-risk-agentic-ai-bypasses-enterprise-stack). That's not a rounding error. That's a fifth of a category most CFOs treat as a fixed cost, suddenly up for renegotiation.
The logic behind the number is straightforward. Per-seat licensing was built for a world where a human clicks through a dashboard, and the vendor charges per person who has to click. Once an AI agent is doing that work end-to-end -- pulling data through an API, updating a record, closing a ticket -- there's no seat to charge for. The agent doesn't need a UI at all. As Gartner Managing VP George Brocklehurst put it in the report: 'You are no longer buying software primarily for people. You are increasingly buying it for agents.'
The report doesn't just theorize this -- it points to vendors already repricing around it. GitHub moved off flat-rate pricing to usage-based billing in June 2026. Zendesk rolled out pricing tied to resolved tickets rather than agent seats. Workday launched 'Flex Credits' as a consumption-based alternative sitting outside its traditional per-seat licensing. Three very different categories -- dev tools, support, HR/finance -- all arriving at the same conclusion within months of each other. That's not a coincidence, that's a category responding to the same pressure at once.
Here's where we think the conversation should go next, and it's not where Gartner stops. If GitHub, Zendesk, and Workday -- three of the most entrenched vendors in enterprise software -- are all scrambling to retrofit their pricing models because per-seat licensing stopped making sense, that's not just a pricing story. It's an admission that the SaaS middleman itself is the thing under pressure, not just its invoice format.
Think about what outcome-based or consumption-based pricing actually is from the buyer's side: you're now paying per action, per resolved ticket, per credit consumed, for a workflow that an agent runs against an API. At that point, ask the obvious question: why route that workflow through a third party's software at all? If an agent is calling an API to resolve a ticket, update a CRM record, or reconcile an invoice, a team can build that exact workflow directly with a tool like ViibeStack's AI app builder and own it outright, instead of paying a vendor's newly-invented consumption fee for the privilege of routing the same task through their infrastructure.
This is the same pattern we've written about with the SaaSpocalypse playing out at Curative and in our build-vs-buy framework: the cost of building a purpose-fit internal tool has dropped faster than the cost of licensing someone else's general-purpose one. Gartner's $234 billion figure is really a measurement of how much of that gap has opened up industry-wide. When Zendesk reprices around resolved tickets, that's Zendesk admitting the ticket-resolution workflow, not the dashboard, is the valuable part -- which is exactly the kind of workflow you can replicate with a purpose-built helpdesk instead of paying by the resolution.
If you're budgeting for next year's SaaS renewals right now, don't just brace for a repriced invoice from your existing vendors -- audit which of those tools are being used primarily by agents or automated workflows rather than humans clicking around a UI. For any tool where that's true, you're the exact buyer Gartner is describing, and you have a real alternative to accepting whatever consumption model the vendor lands on: build the workflow directly. Start with the systems where the human UI was always the thin part of the value -- ticket routing, CRM updates, approval chains, reporting -- and check our internal tools software or templates for a faster starting point than a blank canvas.
The vendors aren't wrong to reprice -- Gartner's data says they have to. But repricing is still renting, just at a different rate. The businesses that come out ahead over the next four years won't be the ones that negotiate the best consumption terms with GitHub, Zendesk, or Workday. They'll be the ones that noticed the same shift Gartner did and used it as the moment to stop renting the workflow entirely.
Sources