Gartner is now putting a dollar figure on a shift vertical AI SaaS founders have felt coming all year. As of July 1, 2026, Gartner estimates up to $234 billion in enterprise application software spend, about 20% of the category, is exposed to what it calls agentic arbitrage: AI agents that deliver an outcome directly and skip the seat-based login altogether. Gartner Managing VP George Brocklehurst frames the mechanism simply, agentic AI breaks the link between user growth and revenue growth for many enterprise software vendors.
That is not a one-year blip. Gartner projects at least 40% of enterprise SaaS spend will move to usage-, agent-, or outcome-based pricing by 2030, with seat-based revenue's share of vendor income falling from 21% to 15% over the same period, according to Gartner research cited by CIO on June 16, 2026.
The infrastructure is already moving
The transition is not theoretical, it is already priced into products shipping this year. On June 1, 2026, GitHub moved every Copilot tier, Pro+, Business, and Enterprise, off flat and per-request pricing onto usage-based billing through GitHub AI Credits metered by token consumption, per the GitHub Blog.
Salesforce shows what it looks like when a legacy seat-based vendor builds a consumption-priced AI line next to its core product. The company reported combined Agentforce and Data 360 annual recurring revenue of $3.4 billion in fiscal Q1 2027, with Agentforce ARR alone crossing $1.2 billion, up about 205% year over year, according to its May 27, 2026 investor press release.
The billing layer is catching up to the pricing model. Stripe announced on January 14, 2026 that it had completed its acquisition of Metronome, the usage-based billing infrastructure company already used by OpenAI, Anthropic, and Nvidia, for a reported $1 billion.
CEO Patrick Collison said the shift toward usage-based models will be a defining feature of the next decade for the industry.
Vertical AI hasn't picked a side yet
Not every vertical AI leader is rushing to drop seats. Harvey, the legal AI company TechCrunch valued around $8 billion in a November 14, 2025 profile, still sells itself as seat-based, multiplayer software to law firms. CEO Winston Weinberg told TechCrunch he wants outcome-based pricing for narrow tasks that can match human accuracy, and plans to build more consumption-based workflows slowly as the systems improve. He is not there yet, and he is not pretending otherwise.
Customer support already crossed over. Intercom's Fin AI Agent charges $0.99 per resolved outcome, defined as a confirmed or assumed resolution, a procedure handoff, or a disqualification, on top of a $49 monthly base plan that includes 50 resolutions, according to Fin's published pricing.
Sierra goes further. Co-founder and CEO Bret Taylor said Sierra charges its median customer a pre-negotiated rate only when its AI agent autonomously resolves a customer issue, and nothing if the conversation escalates to a human, a pricing structure built entirely around outcomes rather than seats.
The split is instructive. Harvey stays seat-based because a law firm still wants headcount-shaped software. Intercom and Sierra charge for outcomes because a resolved ticket is a countable, dated fact. The founder question this week: does your product produce an outcome specific enough to meter? If it does, the billing infrastructure already exists, Stripe's own Metronome acquisition proves that, and Gartner's numbers say the vendors who wait until 2030 will be defending a shrinking 15% seat-revenue share instead of building toward the 40% that is already moving. Audit your product this week for the one action a customer would pay for by result, then price a pilot around it before a customer, or a competitor, does it for you.