SAP attempts to regain control of enterprise data with Dremio and Prior Labs

SAP has just sent a strong strategic signal to the enterprise software market by announcing, almost simultaneously, the acquisitions of Dremio, a specialist in lakehouse architectures built on Apache Iceberg, and Prior Labs, a young German startup focused on AI models applied to tabular data. Beyond the acquisitions themselves, the German software giant is attempting to regain control over the future enterprise data value chain at a moment when artificial intelligence is redefining the very architecture of business software.

The financial terms of the acquisitions were not officially disclosed. However, according to several sources cited by Pathfinders and TechCrunch, the Prior Labs transaction represents an exit valued at more than 425 million euros, primarily in cash, with more than 255 million euros reportedly paid immediately to founders Frank Hutter, Noah Hollmann and Sauraj Gambhir. SAP has also announced an additional investment of 1 billion euros over four years to transform Prior Labs into a global laboratory dedicated to AI applied to structured data.

For several years, a growing share of the value generated around enterprise data has gradually shifted outside traditional ERP systems. Major cloud and analytics platforms such as Databricks, Snowflake and Google Cloud have captured a significant portion of the workloads related to analytics, AI and advanced data exploitation. In this model, ERP systems risk being reduced to transactional backbones feeding far more strategic external platforms, and this is precisely the scenario SAP is now trying to avoid.

Founded in 2017, American company Dremio has emerged as one of the leading players in open data architectures capable of unifying information distributed across multiple clouds, data lakes and analytical databases. Prior to its acquisition, the company had raised approximately 306 million euros from investors including Sapphire Ventures, Lightspeed Venture Partners, Cisco Investments and Adams Street Partners. Its last estimated valuation exceeded 850 million euros.

Its technology enables enterprises to query, structure and govern large-scale datasets through a unified SQL layer while relying heavily on open-source standards, particularly Apache Iceberg. This point is critical because Iceberg is increasingly becoming for data architectures what Kubernetes became for cloud infrastructure: a standardized interoperability layer across multiple environments. The format notably allows companies to manage massive analytical datasets while preserving detailed modification histories, fine-grained governance and stronger data traceability.

The second acquisition involves Prior Labs, founded less than two years ago in Freiburg. The startup had raised approximately 7.9 million euros in a pre-seed round announced in February 2025. The financing was led by Balderton Capital, with participation from XTX Ventures, Atlantic Labs and Hector Foundation. Business angels included Peter Sarlin, founder of Silo AI, Christopher Lynch of AtScale, Guy Podjarny, founder of Snyk and Tessl, Edward Grefenstette of Google DeepMind, Robin Rombach of Black Forest Labs, Ashutosh Kulkarni of Elastic, Thomas Wolf of Hugging Face and Steve Anavi, co-founder of Qonto.

The company develops AI models specialized in tabular data, namely the rows-and-columns structured datasets that form the core of ERP systems. Its TabPFN-2.5 model is designed to generate predictions, detect anomalies and automate certain business analyses directly from databases or large spreadsheets.

This positioning differs significantly from large language models. Since the rise of generative AI, the market has largely focused on LLMs capable of generating text or supporting conversational interactions. Yet within enterprises, most critical data remains deeply structured: accounting, supply chain management, procurement, human resources, treasury operations, vendor management and inventory control.

SAP appears to believe that the next wave of enterprise AI value creation will come less from conversational interfaces than from the intelligent exploitation of these massive transactional databases. The company had already initiated this trajectory with RPT-1, its Relational Pre-Trained Transformer launched last year. The acquisition of Prior Labs now enables SAP to accelerate this strategy dramatically while also acquiring a highly regarded research team specialized in tabular models.

But behind these acquisitions lies a much broader battle: the rise of AI agents.

Major software vendors now understand that the traditional interface of enterprise applications could gradually disappear in favor of agents capable of interacting directly with information systems. In this model, whoever controls the orchestration layer for data, permissions and workflows may ultimately control the future enterprise interface itself.

SAP is therefore seeking to prevent a third party, whether OpenAI, Anthropic or a new generation of agentic platforms, from capturing this strategic relationship.

The recent tightening of its API policy reflects this objective. The company now restricts certain unapproved AI agents from accessing its systems and increasingly favors SAP-validated architectures, particularly around its own agentic layer, Joule.

Ultimately, these acquisitions reflect a common concern shared by historical software giants: preventing AI from transforming their products into invisible infrastructure layers powering far more powerful external platforms. With Dremio and Prior Labs, SAP is attempting to rebuild a central position within the future enterprise technology stack, no longer solely as an ERP vendor, but as a unified infrastructure provider for data, AI and eventually autonomous agents.