Data access and integration: SAP realigns its AI strategy

Perhaps SAP has long felt comfortable with a self-image that Nvidia CEO Jensen Huang once aptly formulated: „SAP is sitting on a goldmine of enterprise data that can be transformed into customized Gen AI agents to help customers automate their business processes.“ This correct assessment falls short: by no means all business-critical data from large corporations that is relevant for AI projects is located in SAP environments.

Acquisitions shift AI focus

With its latest acquisitions, SAP appears to be realigning its AI strategy, which has had only limited success to date. The focus is shifting more towards the work that precedes the actual use of AI: the utilization of company-owned data. Initial approaches to this already exist with BDC Connect - an interface technology that enables zero-copy integration with platforms such as Snowflake, Databricks, Microsoft Fabric or Google BigQuery and thus allows SAP and non-SAP data to be processed on an equal footing.

With the acquisitions of Reltio and Dremio, SAP is now incorporating this cross-silo data access much more strongly into its own portfolio. Reltio in particular is a good strategic fit for this realignment. The cloud specialist for MDM, entity resolution and golden records ensures that companies have consistent and trustworthy master data - i.e. uniform customer, product, supplier or financial data across system boundaries.

This direction is even clearer with Dremio. The company brings modern lakehouse and data federation technologies into the SAP stack, in particular around Apache Iceberg, federated queries and direct access to distributed data sources without traditional replication. Technologically, this is a significant step, as modern AI and analytics architectures are increasingly emerging outside of traditional ERP data models.

The strategy is complemented by Prior Labs. Founded in Freiburg at the end of 2024, the start-up develops AI models that are specifically optimized for structured company data. While traditional LLMs are primarily language-based, Prior Labs focuses on tables, relational data structures, time series and ERP logic - precisely the data worlds in which SAP is traditionally strong. SAP's commitment to invest more than one billion euros over the next four years to develop Prior Labs into a world-leading frontier AI laboratory for structured data is correspondingly consistent.

Reltio and Dremio are reminiscent of earlier developments involving Signavio and Celonis. Celonis was long regarded as the technological leader in process mining, but became too expensive for SAP. Instead, SAP took over Signavio and established its own, deeply integrated counter-model. A similar pattern could now be repeated in the data and AI environment - as Reltio and Dremio and their platforms are in direct competition with strategic SAP partners such as Databricks, Snowflake and Google BigQuery.

This results in new strategic directional decisions for customers. Many companies have invested in modern lakehouse architectures and are now faced with the question of whether they should also build their own SAP data platforms. The advantages lie in deeper integration, less interface complexity and central governance - however, vendor lock-in is also increasing in the long term.

Effects on SAP service partners

Solution partners are thus once again receiving the signal that SAP wants to occupy attractive growth markets itself, which could limit the potential for independent data integration services in the medium term. For the SAP service partner ecosystem, the potential for complex data integration services could also be reduced in the medium term if SAP attempts to integrate larger parts of the data architecture directly into its own platform as a standard feature.

To the partner entry: