SAP Business AI 2026: Usage, Units & Reality

Similar to many software vendors, SAP has directed their efforts to embedding Artificial Intelligence within their services as the standard. Originally, AI functionality was offered as a surface level addition for isolated business cases in automation and analytics. This has progressed through SAP’s Business AI Commercial model: comprised of 3 individual value creation and monetisation streams.   

Base AI, Premium AI & AI Foundation are distinguished by the degree of functionality they enable, and pricing coupled towards the perceived return on investment possible with each.  

SAP Business AI Model 

How many AI units will I need?  

SAP’s AI Service List shows the AI Unit conversion of each block of eligible AI Service which provides a basis to volume requirement. Some of the conversion factors are aligned with the tiered pricing structure of individual SAP solutions. 

For example, the SAP Cash Application FI-AR AI service (Measured by Transactions) equates to 110 AI Units for up until 20 blocks. At over 400 blocks purchased, the AI Unit Conversion Factor drops to 26 per unit. However, the majority of eligible AI Services are converted to the same number of AI Units at all usage quantities. Such as AI-assisted Skill Identification from Resume for which each Object processed, correlates to 0.1 AI Units – regardless of usage.  

Here is a partial list of the AI Services List:  

*Adapted from the SAP AI Services List – full conditions can be found in the AI Services List and all elements of the AI Services List are subject to change.

*Joule entitlement is bundled with SuccessFactors Users, RISE with SAP S/4HANA Cloud, Private Edition & SAP IBP but additional messages can also be purchased.

An important distinction to make is (if available in dual format) if the AI service is  embedded into the solution or if a custom version, developed and deployed on SAP BTP is required.  

The distinction between the two can be critical during service selection because choosing both models can result in double-licensing, which SAP is typically unwilling to remediate once the Bill of Materials are finalised. 

Using SAP Document AI as an example: embedded Document AI follows a user-based licensing model as it sits directly into the solution and will count and consume each document “Request” as 0.1 AI Units. Alternatively, Custom Document AI will be consumed by document requests under either the BPTEA or PAYG models after document blocks or requests are processed.  

Important clarifications for actions constituting to actual measured use?  

• For some AI Services the AI Unit Conversion Factor for this AI Service is based on productive use only. – this means anything performed in a development system will not be counted as use.  
• There could be a minimum usage period for the AI Service (e.g. 6 months) 
• Pre-requisites of RISE or GROW deals may be applicable to certain services.  

Estimating AI Cost: SAP AI Pricing Estimator Tool

To aid in the uncertainty of cost and required volume projection of AI Units, SAP provides an AI Pricing Estimator Tool (accessible via the SAP Discovery Centre). This tool gives the user the ability to plug in as many AI Packages, Agents and Features as required which then (subject to volume tiering) gives an estimated gross pricing breakdown, expected AI Unit consumption for each month and the perceived value add-on (ROI) of implementation.  

What counts as a metric used?  

As previously mentioned, like with previous ECC licensing, there is some nuance behind what will trigger SAP AI functionality. The most important thing is to have understanding of the corresponding AI service metrics and which discrete user actions will invoke the consumption model. For example, usage may be counted per prediction call, per document processed, per recommendation generated, or per natural-language interaction. Two services both within the same business function labelled “AI” may therefore behave very differently from a consumption standpoint, even if they appear similar. 

Key Considerations:  

• Process Integration Depth – AI embedded directly into standard transactions or automated jobs will generate greater usage than optional user interface functionality. 
• User behaviour – Only using the AI functionality when necessary will reduce usage, so internal process controls and governance are very important.  
• Governance and Controls – role restrictions and authorisation objects are crucial in limiting the AI functionality triggers to those who are permitted.  

Without competence of these distinct metrics and logic, companies face misjudging allotted license volumes which could result in unexpected costs.   

As the SAP Business AI Model becomes embedded across S/4HANA and BTP the newest challenge is the projection, control, monitoring and cost of AI. Layered pricing models and various consumption metrics mean organisations will depend on the clear visibility of their estate going forward.  

At JNC, we support our clients with bespoke S/4HANA migration strategies and detailed cost modelling to forecast in the financial impact of AI for up to 10 years. We validate your entitlements and align AI adoption with commercial objects to ensure your investment remains predictable, well-governed and cost effective.