Everyone shipped AI, not everyone is winning from it
A new survey from Revenera, a software monetization platform, puts a number on a gap enterprise buyers have been sensing anecdotally all year: nearly every software provider now ships some form of AI capability, but almost half of the 415 surveyed leaders at software producing firms report low or limited customer adoption of those features. The survey ran in June and July, capturing a market where AI functionality has become close to table stakes in a sales pitch, regardless of whether customers are actually using what got built.
That gap between shipping and adoption matters directly to the enterprise technology leaders on the other side of these purchases. A feature a vendor has invested heavily in, and now expects to monetize, is not the same thing as a feature a buyer's teams have adopted into daily work. Enterprises renewing contracts this year are effectively being asked to help underwrite vendor AI investment whether or not their own teams are using the resulting features, and most procurement processes still have no clean way to separate the two when a renewal quote arrives.
The profitability split vendors are not advertising
The survey's profitability numbers are more uneven than most vendor marketing suggests. Just over half of the providers surveyed say offering AI is improving their profitability, which is the number that tends to appear in investor updates and sales decks. Nearly one in five say the opposite: offering AI has actually reduced their profitability, a figure that rarely makes it into a public facing narrative but matters enormously to how sustainably a vendor can keep supporting the AI features it is selling.
For an enterprise buyer, that split is a real signal about vendor stability. A provider whose AI investment is quietly eroding its margins is under more pressure to either raise prices sharply, cut support quality, or get acquired than one where AI is genuinely additive to the business. Due diligence on a strategic software vendor in 2027 should include asking directly which side of that profitability line the vendor sits on, not assuming the answer from the pitch alone, because a vendor losing money on its own AI bet has every incentive to recover that margin from the next renewal.
Pricing confusion is the customer complaint vendors keep hearing
Nicole Segerer, SVP and general manager of Revenera, describes the core customer frustration in blunt terms: customers do not want to deal with very bumpy pricing, and understanding exactly what a consumption based model means for any given vendor takes real work. Subscription pricing still leads revenue for 66 percent of surveyed providers, with one time perpetual licensing a distant second at 44 percent, but AI features are increasingly layered on top with usage based pricing that does not map cleanly onto either model.
That mismatch is exactly the friction enterprise procurement teams have been describing informally for most of 2026: a predictable annual subscription number suddenly becomes unpredictable once AI usage based add ons enter the invoice. Segerer's framing suggests vendors themselves recognize this as a problem rather than a feature, which gives enterprise buyers real leverage to push for pricing structures with hard caps or predictable tiers rather than accepting open ended consumption pricing by default. A vendor willing to acknowledge the complaint publicly is a vendor more likely to negotiate on it directly, which is worth testing at the next contract conversation.
The subscription seat threat hiding in the data
Perhaps the most consequential finding for enterprise software strategy is forward looking rather than descriptive. Seven percent of providers already report selling fewer software subscriptions specifically because of AI, and nearly a quarter expect wider AI agent adoption to reduce demand for subscription seats going forward. That is a vendor base openly anticipating that the per seat pricing model underpinning most enterprise software contracts is about to come under real pressure as agents, not individual human users, start doing more of the work.
For a CIO negotiating multi year software agreements right now, that expectation is worth raising directly in the negotiation rather than waiting for the vendor to restructure pricing unilaterally later. A vendor that privately expects seat based pricing to shrink has an incentive to lock enterprises into current per seat terms before that shift becomes undeniable, which argues for shorter contract terms or explicit renegotiation clauses tied to agent adoption rather than long commitments locked to today's assumptions about how many human seats a workflow actually needs.
What belongs in the next renewal conversation
Put together, the Revenera data gives enterprise buyers a specific checklist for the next software renewal: ask whether the vendor's AI features are seeing real internal adoption or sitting unused, ask which side of the profitability split the vendor's AI investment falls on, and push for pricing structures that do not expose the enterprise to open ended consumption costs. None of those questions require deep technical diligence, they require treating the vendor relationship with the same financial scrutiny applied to any other strategic contract, and asking a vendor to answer them on the record tends to be revealing on its own.
The broader lesson for 2027 planning is that AI pricing has not settled into a stable model anywhere in the software industry, including among the vendors selling it most aggressively. Enterprises that negotiate as if the current pricing structures are fixed will likely find themselves renegotiating from a weaker position once vendors move to capture the agent driven shift this survey shows many of them already expect. Treating this year's contract terms as temporary, rather than as the new baseline, is the more realistic posture for any CIO planning software spend into 2028 and beyond.
