While everyone builds chatbots, SAP just closed a €1bn-plus bet on AI that understands your spreadsheets

This is strategically more important than financially large: SAP is buying distribution into the system of record, which is where structured AI can actually convert into budget. For enterprise buyers, a model that sits inside finance, procurement, and supply-chain workflows can justify price uplift and retention improvement far more easily than a standalone chatbot. The first-order effect is better wallet share in the installed base; the second-order effect is higher switching costs, which should support longer-duration recurring revenue and modest multiple support.

The main winners are SAP’s own cloud stack and, indirectly, the hyperscalers that host inference; the losers are point-solution vendors whose value proposition is “AI for enterprise data” without owning the transaction layer. That creates pressure on data-prep, workflow automation, and copilot vendors that need customers to export data out of SAP before they can monetize. The risk is integration: if the product remains a lab-to-production story rather than a usable feature set, the market will treat this as another tuck-in with little P&L impact.

Catalysts are slower than the headline implies. Over the next 1-2 earnings cycles, watch for evidence in deal size, renewal commentary, and whether AI features are attached to higher-priced tiers rather than bundled away. Over 6-18 months, the thesis only works if SAP can show incremental gross margin or at least no margin dilution while improving cloud backlog quality. The contrarian view is that the market may be underestimating how defensible embedded structured-data AI is: in enterprise software, context beats model quality. What would falsify this is flat AI attach rates, no uplift in net retention, or weaker-than-expected cloud conversion despite the new capabilities.