August 24, 2026
For weeks we've been making a case, one vendor at a time, that per-seat SaaS pricing has a structural problem: it charges growing teams more precisely as AI absorbs more of the work those teams used to need extra headcount -- and extra seats -- to do. We made that argument looking at HubSpot, Asana, Zendesk, Trello, Calendly, Jobber, and Bullhorn. Each time, the mechanics were slightly different, but the pattern was identical: the software gets cheaper to run per unit of value delivered, while the bill keeps climbing with headcount. On August 18, 2026, BigGo Finance reported on research from Bain & Company that says, in effect, the same thing -- except now it's about the two biggest seat-license vendors on earth, not a mid-market tool. Bain analyzed more than 30 SaaS vendors and found that roughly 65% have already adopted hybrid pricing models that blend AI-usage charges into traditional per-seat pricing. That's not a hypothetical shift. That's most of the industry already admitting the old model doesn't hold on its own.
The specific figures in the BigGo Finance report are worth sitting with. At Salesforce, the customers using AI agents the most heavily are already spending 1.5x more year-over-year -- not because they added more people, but because usage of AI is the thing driving the spend now. Meanwhile, CEO Marc Benioff has kept engineering headcount flat at roughly 15,000 for two years, while AI drove the productivity gains that let the company grow anyway. That's the seat-license logic breaking in real time, inside the company that arguably built the modern version of it: growth used to require proportionally more people, which required proportionally more seats. Now it doesn't. At ServiceNow, the number is even blunter: roughly half of net-new business revenue already comes from pricing that isn't seat-based at all. That's not a future risk to the seat model. That's ServiceNow's own current revenue mix telling you the model is already half-replaced by something else, inside one of the companies with the most incentive to keep it alive.
Todd Ahlsten, CIO of Parnassus Investments, put the institutional read on this plainly, telling BigGo Finance: "The seat license model is just going to be under a lot of pressure." He added: "I just think the long-term bankability of that is going to be challenging." That's a Wall Street CIO, not a challenger vendor with something to sell, saying the quiet part about Salesforce and ServiceNow specifically. Here's the practical logic for anyone running a software budget right now: the companies with the deepest pricing-power incentive to keep the per-seat model alive forever are the ones already blending in usage-based charges. If Salesforce and ServiceNow -- who invented and perfected this pricing category -- are hedging away from pure per-seat billing because they see the pressure coming, that's a far stronger signal than any outside critique. It means the "just pay per seat forever" assumption baked into most software budgets is already out of date, not hypothetically, but in the actual revenue mix of the largest players in the category.
It's tempting to file this under generic AI hype, but that undersells what's actually happening. The Bain data isn't a forecast about what AI might do to software pricing someday -- it's a description of pricing structures more than 30 vendors have already changed. ServiceNow's 50% figure isn't a prediction either; it's the current state of their books. This is a repricing that has already started, visible in the numbers of the vendors with the most reason to resist it. The deeper issue is that per-seat pricing was always a proxy. It assumed a fixed, roughly linear relationship between the number of people on a team and the amount of value a piece of software delivered them. AI breaks that link cleanly: a five-person team using AI agents can now do work that used to require ten seats, and under a pure per-seat model, they'd never see that efficiency reflected in their bill -- only the vendor would.
We didn't need Bain's research to build ViibeStack around this idea -- but it's good confirmation. The whole premise of an AI app builder that builds the specific tool a team actually needs, rather than selling a generic seat license and hoping teams grow into it, was a bet that software pricing needed to track the business, not headcount. That's the same logic we laid out in Buy vs. Build vs. ViibeStack: the honest match for where software economics were headed was never "charge more as the team gets bigger," it was "charge for the value delivered, however many people happen to be doing the work." If you're heading into a renewal conversation with a seat-based vendor right now, this data gives you real leverage, not just a feeling. The vendors themselves are telling you, in their own revenue mix, that the seat-scales-with-value assumption is already breaking down. That's worth bringing to the table before you sign another multi-year seat commitment for software that's doing proportionally less of the differentiating work every year AI gets better.
Sources