SAP BDC, the new Business Data Complexity

AI and agentic AI need a consistent database

The success of AI depends on a consistent database, which makes data management and archiving a strategic field of action. SAP has laid the foundation for the harmonization of SAP data and its integration with third-party sources with the Business Data Cloud (SAP BDC). In order for the BDC to develop its impact as a standardized data platform in companies, it must be continuously developed further to strengthen its role as a transparent and connectable data layer.

Currently, however, the SAP BDC regulations, instructions and tools are still completely inconsistent and imprecise, meaning that this further development of ERP data will hardly take place. Participants at the DSAG Technology Days 2026 in Hamburg reported an ironic reinterpretation of SAP BDC by DSAG: Business Data Cloud became Business Data Complexity!

As reported to E3 magazine, Michael Bloch, Head of Licenses, Contracts and Support at the German-speaking SAP User Group (DSAG), presented numerous stumbling blocks and contradictions in the licensing of the BDC universe in a presentation. The consolidated data universe desired by existing SAP customers appears to be more of a multiverse. According to some DSAG members, Michael Bloch has announced that he will enter into negotiations with SAP about the problems and inconsistencies of BDC licensing.

Expiry date of credit and capacity units

Existing SAP customers and partners are familiar with the annual expiry of BTP credit units. SAP has now tightened this regulation for the Business Data Cloud: purchased and required capacity units already expire on a monthly basis.

DSAG board member Michael Bloch also criticized the administrative effort involved in calculating license fees, according to the presentation participants: Almost every aspect of SAP BDC has its own metric and its own pricing scheme with numerous exceptions and extensions. DSAG members said that under the given circumstances, it is almost impossible to determine the expected license costs in advance. Add to the complexity of an S/4 Hana landscape the Business Data Complexity!

The often still fragmented IT landscapes at companies stand in the way of this. „This can be remedied by more ready-made, documented data products with user control over the domains plus practice-oriented and predefined training and governance packages. Companies also need a clear cataloguing of data products and a mapping between customer-specific structures and standard products,“ explained Stefan Nogly, DSAG Chief Technology Officer, in Hamburg at the Technology Days 2026. The mapping of modern data governance ensures that data silos, redundant structures and different interpretations of terms and key figures are resolved. This is where SAP BDC can provide effective support so that AI can be used to create benefits through AI agents, modern analytics functions and intelligent processes.

SAP's pace of innovation in the areas of cloud, AI and analytics is high and poses considerable challenges for existing SAP customers. Many are still in the process of introducing new cloud technologies and S/4 conversion with Clean Core and need further guidance, otherwise the technical distance between SAP and existing customers will grow.

Clarity about future IT architectures

In order to make optimum use of the strategic leverage of AI, existing SAP customers also need transparency about their future IT architectures. In reality, companies very often invest in their own AI solutions alongside their SAP strategy. On the one hand, this is a sign that some companies are still in an experimental phase and, on the other, that SAP's entry barriers are too high for existing customers, see Business Data Complexity.

This is also reflected in current investment behavior: According to the DSAG Investment Report 2026, the member companies surveyed are still using few AI use cases productively in the AI sector. One reason for this is that the companies surveyed do not yet have any SAP cloud applications in use. „That's why companies now need clarity about future architectures and application scenarios. At the same time, however, they are also required to build up their own AI expertise in a targeted, structured and sustainable manner in order to keep pace with the rapid developments,“ said Stefan Nogly, DSAG Chief Technology Officer, in Hamburg.

Let's take a sober and analytical look behind the scenes of the Walldorf marketing machinery. SAP is boldly promoting the SAP-RPT-1 (Relational Pre-trained Transformer) model as the first foundation model designed natively for relational and structured business data - i.e. the bare tables of the ERP system. But how does an existing SAP customer actually get this tool to run on their own SAP Business Technology Platform (BTP)?

AI with SAP BTP and SAP-RPT-1

The implementation is less like a simple plug-and-play and more like building a well thought-out architecture that requires strategic advance decisions, specific licenses (business data complexity) and robust data pipelines.

Before even a single line of code is written, SAP forces the existing customer into the premium segment of BTP. RPT-1 is not available in the basic or free tier plans. You must have an instance of SAP AI Core under the Extended Plan. In addition, the SAP AI Launchpad (usually in the Standard Plan) is required as a graphical control center to maintain an overview. For the IT manager, this means Users must engage with the consumption-based license models (CPEA or BTPEA) in the form of cloud credits or AI units, the actual costs of which must be monitored precisely during operation.

In order for the RPT-1 model to work, the ERP data must flow to it. As RPT-1 acts as a „GPT for tables“ and does not break down number sequences into error-prone text tokens, but understands them as mathematical values, clean data provision is essential. In practice, this means that users have to set up data pipelines. For example, they use OData services to extract relevant table content from SAP ECC or S/4 and feed it to the RPT-1 model on the BTP. This is where the topic of business data complexity with OData services comes up again. However, it is this integration work that turns the theoretical AI function into operational business value.

SAP is positioning RPT-1 as a technical liberation to avoid the notorious hallucinations of classic language models (LLMs) in pure numerical works. Deployment in the BTP via the Generative AI Hub plus BDC is standardized and can be done quickly by cloud developers.

Nevertheless, skepticism is warranted: In the SAP community, the model is sometimes referred to as AI chaos theory and the suspicion is voiced that SAP may simply be repackaging classic machine learning algorithms (such as those from the Hana Predictive Analysis Library, PAL) and monetizing them under the marketing hype of the „Foundation Model“. As in-context learning forces the permanent transfer of ERP data streams via API (SAP BDC) to the BTP, existing customers face a high initial integration effort - and a deep vendor lock-in into the lucrative BTP and BDC ecosystem of the Walldorf-based ERP group SAP.