SAP Acquires Prior Labs to Build a European Frontier AI Research Lab
May 4, 2026 – 4:11 pm
Eighteen months after securing €9m in pre-seed funding, the Freiburg-based startup pioneer of TabPFN, Prior Labs, is being acquired by SAP for over €1bn over four years. While the terms remain undisclosed, the strategic move is clear.
When Frank Hutter, Noah Hollmann, and Sauraj Gambhir founded Prior Labs in early 2024, AI discussions centered around language models, attracting significant capital and attention. However, Prior Labs focused on structured data—tables that power businesses—which was largely overlooked in the quest for frontier labs.
On Monday, the founders announced that this narrative is about to change. Prior Labs has entered into a definitive agreement with SAP to become what the latter terms as a globally leading frontier AI lab. This marks the most substantial European enterprise AI research investment by a European company.
The transaction details have not been revealed. SAP’s official announcement describes Prior Labs as pioneers of Tabular Foundation Models, highlighting the acquisition as an extension of SAP’s own work on a model called SAP-RPT-1.
Prior Labs will maintain its independence, retaining its brand, Freiburg headquarters, Berlin and New York offices, open-source commitments, customer relationships, and scientific advisory board comprising Yann LeCun and Bernhard Schölkopf. The deal is subject to regulatory approval and is expected to close in Q2 or Q3 of this year.
In a joint blog post, the founders spoke of the "next chapter," emphasizing Prior Labs’ achievements:
- TabPFN, their flagship model, was published in Nature in 2025 and has since been cited over 1,000 times and downloaded more than three million times.
- The latest version, TabPFN-2.5, can handle datasets of up to 50,000 samples and 2,000 features, surpassing the accuracy of tuned tree-based models in benchmark tests.
Prior Labs’ breakthrough lies in its ability to process general-purpose data efficiently, without task-specific training, marking a significant technological advancement for enterprise AI applications.