The leverage question turns on whose budget is under pressure, and on that the interview is specific. Gilg's account is that nobody is ripping SAP out, but that token spending is now competing for the same money [8]. "Customers are seeing token costs go through the roof already," he said, adding that "budgets won't necessarily increase. It will still have to come from somewhere. Therefore, the scrutiny will become much, much bigger" [9]. Read as a sales position, that is confidence. Read as a procurement input, it says the buyer's AI bill has to be funded by squeezing an incumbent line, and in most large estates the incumbent line is the ERP contract.
On the repricing itself, the record is one clause. Gilg lists changing pricing models alongside AI reinvention and data investment as how SAP intends to end up among the winners [6], and Business Insider's account of that change ends mid-sentence at a move away from some subscription models, with no product, date, or unit of measure attached [7]. That is not something a buyer can negotiate against. What the material supports is an executive naming price as one of three levers, which is a different object from a published change in how SAP charges.
The share price is the firmer number, and it points the other way from the leverage thesis. The same report has SAP down about 20 percent over the past year [4] and up roughly 40 percent since the late-July earnings report [5]. Taking both as of the same day, the pre-rally trough sat about 43 percent below the year-earlier level, since 0.80 divided by 1.40 is 0.57 [16]. The rebound is real, and it is a rebound off a deep hole, which means the market has already paid SAP for the AI story Gilg is telling. A vendor that has just been re-rated does not usually arrive at renewal talks in a discounting mood.
The recovery reflects investors repricing SAP's narrative rather than SAP repricing its software, and a president of customer success discussing pricing models is not a rate card. The two claims in this interview do not rest on the same class of evidence: the 40 percent move is measurable [5]; the pricing overhaul, in the material available, is an assertion [7]. Where they meet is Gilg's concession that customers have not yet seen the benefit of AI [10], which is a vendor stating the fact that weakens its own pricing power.
The tradeoff a consumption model puts in front of a finance function is predictability against optionality. A subscription line is a number defended once a year; a metered line is a number defended every month, falling when usage falls and rising when the agents work as advertised. SAP's own defence of packaged software is that mission-critical buyers want auditability, governance, and maintenance rather than vibe-coded replacements [15], and it says its edge is data from more than 10,000 customers [13]. Both of those are switching-cost arguments, and that switching-cost moat gives SAP little incentive to hand over a cheaper meter.
This quarter's decision and next quarter's consequence are separate. This quarter, a buyer can hold term length and cap metered growth, using the single thing the vendor has put on the record against itself: the outcomes are not in yet [10]. The longer question is who owns the ontology layer, where Gilg places SAP against Anthropic, OpenAI and Palantir [14] while SAP also partners with Palantir to move data between systems [12]. A meter agreed this quarter is read every quarter afterwards, which is why the term is worth more argument than the rate.