Establishing TFM Leadership: Could Prior Labs Acquisition Be Key To SAP’s AI Strategy Refresh?

SAP has been re-formulating its AI strategy to more effectively support enterprise AI agent adoption. It aims to address the problem of technology stack immaturity and data readiness that continues to constrain AI agent implementation at scale, whilst simultaneously moving away from the increasingly commoditized data layer towards the more attractive and monetizable agentic layer. Here’s what you need to know about SAP’s existing AI product strategy:

• Product: SAP has introduced the Autonomous Enterprise, a unified platform for building, contextualizing and governing AI agents. This comprises the Business AI platform for technical users; the SAP Knowledge Graph for contextual grounding; and Joule Studio, a low-/no-code environment enabling agent, application and workflow development.
• Partnerships: SAP has expanded implementation partnerships with Accenture and Palantir to support large-scale data migration and transformation.
• Customer ecosystem: SAP is launching a €100 million fund to accelerate AI agent development, aiming to drive near-term experimentation, adoption and ecosystem growth around its Business AI platform.

While this strategic repositioning is significant, it only tells part of the story. When viewed alongside two recent acquisitions, it becomes clear that SAP is also reinforcing capabilities deeper in the data and model layer to address core AI agent technology constraints:

• Dremio.
  This is a largely defensive move which aims to help SAP keep pace with competitors such as Databricks and Snowflake by improving the unification of SAP and non-SAP data. Strengthening this layer is critical to enabling AI agents to access, query and reason across distributed datasets.
• Prior Labs.
  This is a more offensive and potentially differentiating acquisition. It positions SAP as a leader in tabular foundation models (TFMs), with plans to establish a frontier model lab backed by approximately $1 billion in investment over the next four years, building on its release of SAP-RPT-1 in 2025.

TFMs are emerging as a compelling complement – or alternative – to LLMs for enterprise AI agent use cases. Their value lies in their ability to handle structured data and statistical relationships more effectively, while delivering strong in-context learning, capturing numerical relationships, producing more deterministic outputs and enabling high-quality predictive capabilities through inferred relationships within data.

This makes TFMs particularly critical in data-rich domains such as finance, logistics and manufacturing – areas where SAP has deep market penetration and where AI agent deployments are already demonstrating strong ROI.

Rather than replacing LLMs, TFMs are likely to operate within hybrid architectures. In these frameworks, LLMs excel at interpreting user intent and generating natural language responses, while TFMs handle structured data evaluation and provide more reliable, deterministic outputs.

While some competitors already support tabular models (for example, Databricks hosts TabPFN), this announcement offers new functionality.  SAP’s acquisition of Prior Labs – combined with scaled R&D – creates an opportunity to build truly differentiated value through tight integration with the Autonomous Enterprise platform, enhancing AI agent performance and overall value creation. This advantage will become increasingly critical. SAP is not only competing with Databricks and Snowflake – which offer largely homogeneous AI capabilities – but will also be competing at the AI agent layer beyond its traditional market boundaries as data ecosystems become more unified.

SAP’s strategy represents a clear view of the future of AI agents and the role of database providers: one defined by unified data, more deterministic models that effectively handle structured data, and orchestrate agent workflows within a cohesive, simple-to-use platform. For more information on the AI agent platform market, please read Strategic Focus: Five Essential Questions When Selecting An Enterprise AI Agent Platform.