For more than ten years SAP's pitch has been simple. Move to the cloud, or get left behind. AI was the big carrot.
Now SAP is putting that carrot back on the table. It is offering it to customers who never made the jump.
The reason is that AI is forcing the whole software industry to rewrite how it sells.
SAP's old model charged customers per user, or per seat. AI agents are now doing the work that used to need a person at a screen.
That breaks the link between users and value. CEO Christian Klein laid it out in a Bloomberg interview earlier this year.
He said it would be foolish to keep charging by subscription. The reason is that AI is taking over so many tasks.
The company is moving to consumption-based pricing. Customers will pay for what the AI actually does, not how many badges scan in.
That puts SAP in the same camp as cloud infrastructure firms. Costs scale with use, not seat counts.
A huge chunk of SAP's revenue still sits with customers running older on-premises systems. That includes S/4HANA on-prem and ECC.
SAP has spent years pushing those customers toward the cloud. It did that by making cloud-only features the only way to access new tools.
Opening AI access to those customers is a clear shift. SAP would rather collect AI fees from a non-cloud customer than lose them to rivals.
It also lowers the political cost of staying on-prem. Customers who wanted AI but did not want to rip out their old systems now have a third path.
SAP shares have lost about 20% of their value this year. Investors are watching the pricing shift closely.
The new model removes the planning power that made SaaS so valuable on public markets. Klein is also building out new "forward deployed engineering" teams to work with customers hands-on.
SAP also agreed to buy Reltio, a master data firm, and Dremio, a data lakehouse. Both deals are meant to feed SAP's AI agents the clean data they need.
Without good data, the AI agents cannot run a workflow.
The new pricing model also adds risk for customers. They have to forecast AI use up front to budget for it.
The real question is whether SAP customers actually move more work onto AI agents. Consumption pricing only works if usage goes up.
If usage stays flat, the model that replaces SaaS could end up smaller than the one it killed.
SAP also has to keep up with rivals like Oracle and Workday. All of them are pitching the same AI story to the same shrinking pool of buyers.
For investors, the next signal will be SAP's earnings calls. Klein has to show that AI use is rising fast enough to offset the move away from per-seat fees.
Without that growth, the new model could just shrink the pie. The bull case is that AI tools drive far more usage per customer over time.
The bear case is that customers cap their AI use to keep bills in check.
