If you sat through an enterprise software renewal this year, you probably noticed something odd: the vendor no longer has one price. Salesforce now sells Agentforce three separate ways at once, at two dollars per resolved conversation, at ten cents per “action” drawn from a credit bundle, and as a flat per-user license north of $125 a month. Three pricing philosophies, offered simultaneously, to the same buyer, for adjacent parts of the same product. That is not a company that has found the answer. It is a company hedging against the possibility that nobody has.
The prevailing story in boardrooms and analyst notes goes something like this: AI agents let one person do the work of many, so per-seat licensing, the model that has funded enterprise software for two decades, is finished, and usage- or outcome-based pricing will replace it in an orderly handover. It’s a tidy narrative. It is also, on the evidence, wrong about the timing, and the timing is exactly what CEOs and CFOs are mispricing.
The transition everyone assumes is already happening isn’t
Deloitte’s own technology forecasting puts it plainly: by 2030, at most 40 percent of enterprise SaaS spend will have shifted toward usage, agent, or outcome-based pricing. Read that the other way and the more interesting number appears: six in ten dollars of enterprise software spend will still sit on some other model four years from now. Gartner’s research tells a similar story from the services side, where only 13 percent of technology service agreements and 19 percent of buyers currently use outcome-based pricing at all, and Gartner expects that figure to stay under a quarter of contracts through 2031. Tom Coshow, the Gartner analyst who tracks this, put it more bluntly than most vendors would like: “what we see is that the increase in outcome-based pricing is more buzz than reality.”
Meanwhile the part of the market that has changed is changing chaotically rather than converging. Credit-based pricing, the flexible middle ground between seats and outcomes, grew 126 percent year over year across B2B vendors in 2025, and hybrid pricing structures, models that stack two or more schemes on one contract, jumped from 27 to 41 percent of vendors in the same period. HubSpot, Figma and Adobe have all bolted credit systems onto existing seat licenses rather than replacing them. That is not a market settling on a new answer. It is a market that has stopped trusting its old answer and hasn’t agreed on a new one, which is a very different, and more dangerous, thing for a buyer to walk into.
The reason it’s stuck is structural, not a temporary awkwardness that a clever product manager will solve next quarter. Outcome-based pricing requires two things almost no enterprise relationship currently has: a metric both sides trust as causally linked to the vendor’s work, and a vendor willing to absorb the downside when that metric disappoints. Paul Fisher, CIO at Seton Hall University, notes that outcome contracts “add complication to contract negotiation as you need to be very specific on what the desired outcomes are,” and Coshow’s sharper test for buyers is the one that actually matters: if the vendor isn’t taking on the risk, why are you bothering with outcome-based pricing at all? Most of what is marketed as outcome-based pricing today is consumption pricing wearing a nicer label, transparency and planning dressed up as risk transfer.
The compression is happening anyway, with or without a new contract to hold it
Here is the part the pricing-model debate obscures: value is already draining out of the old model, in real time, whether or not a new one exists to capture it. Nowhere is that clearer than in Indian IT services, an industry built almost entirely on a linear relationship between headcount and revenue, and one now watching that relationship come apart in the reported numbers. Indian IT revenue grew 6.1 percent in FY26 while headcount grew only 2.3 percent, a gap between the two lines that, for an industry historically priced by the person-hour, should not exist.
The mechanism is visible in the earnings calls, not the press releases. HCLTech’s CEO, C Vijayakumar, has said teams now need roughly 25 to 30 percent more effort to earn the revenue they used to earn for the same work. TCS’s CEO, K Krithivasan, has confirmed the company typically passes through 10 to 15 percent of AI-driven productivity savings to the client at the point of signing, essentially discounting the work before any new contract structure exists to price the replacement value. Wipro has flagged lower margins on some deals even as it leans on AI to “deliver our fixed-price programs better.” None of this required a single outcome-based contract to happen. It happened inside the old model, through renewal pressure and informal productivity pass-throughs, faster than the new model could be built to absorb it.
The new model, meanwhile, remains genuinely small. Outcome-based contracts run at roughly 6 to 7 percent of revenue at Coforge, and Infosys’s CEO Salil Parekh describes client interest as real but not yet “a large part of our activity.” AI-specific services across the whole $315 billion Indian IT sector amount to perhaps $10 to $12 billion, and much of that is one or two-quarter modernization work, such as Infosys converting COBOL systems into microservices at 60 percent lower cost and in 60 percent less time, rather than the multi-year annuity relationships the industry was built to sell. Every one of those reference projects also resets the client’s expectation of what the next contract should cost, which compounds the pricing pressure on everyone else’s renewal before anyone has designed a mechanism to capture the new value being created.
Why the gap, not the winner, is the thing to manage
This is the piece the “seats versus outcomes” framing misses entirely. The risk to margins, valuations and, in India’s case, an entire employment engine built around campus hiring, is not that one pricing model will beat the other. It’s that the old model is being drained faster than the new one is being built to replace it, and that gap has no natural floor. A SaaS vendor losing seats to agentic consolidation and a services firm passing through productivity savings at signing are experiencing the identical problem from opposite sides of the same contract: value is leaving the relationship before either party has agreed on how to re-price what remains.
For a CEO or CFO, the practical response isn’t to bet the balance sheet on guessing which pricing model wins, because the evidence says nobody is winning yet. It’s to stop treating “outcome-based” as a synonym for de-risked, and start asking Coshow’s question of every vendor who uses the phrase: what exactly are you putting at risk here? It’s to decouple workforce and campus-hiring plans, especially for organisations leaning on Indian GCCs and IT services partners, from the assumption that revenue growth still tracks headcount growth, because the two have already diverged in the data. And it’s to renegotiate on cadence rather than category: shorter terms and more frequent repricing checkpoints will protect a budget through this window better than a confident multi-year bet on either the seat license or the outcome contract, because instability, not the wrong model choice, is the actual condition everyone is operating under.
The vendors and services firms that emerge stronger from this window won’t be the ones with the cleverest pricing sheet. They’ll be the first to build measurement good enough that a customer will let them share genuine downside risk, which is the only thing that turns “outcome-based” from a marketing phrase into a mechanism anyone can rely on. Until that exists at scale, every enterprise buyer is negotiating twice over: once on what the software or service actually does, and again on what anyone can prove it was worth. That second negotiation, far more than the artificial intelligence sitting underneath it, is where the advantage of the next few years will actually be won.
