From a Nature paper to a billion-dollar investment by SAP

The machine learning algorithm TabPFN, developed by a team led by BrainLinks-BrainTools member Frank Hutter, uses small tabular datasets to make precise predictions. The researchers published the details in Nature in 2025; since then, the paper has been cited more than 1,000 times. Shortly before, Hutter, together with Noah Hollmann and Sauraj Gambhir, founded the startup PriorLabs to further develop the model and bring it to market. Now, the software giant SAP is stepping in: It will acquire PriorLabs and invest more than one billion euros over the next four years. The goal: to expand PriorLabs into a world-leading frontier AI lab for structured data.

TabPFN uses learning methods inspired by large language models. One key difference: TabPFN learns causal relationships from synthetic data, enabling it to make more accurate predictions than the algorithms typically used to date.

“TabPFN delivers better results faster than traditional machine learning methods, without any manual tuning. It thereby allows data scientists to be more efficient, and to focus their efforts on more interesting tasks than parameter tuning.”

Prof. Frank Hutter

Co-founder of Prior Labs and member of BrainLinks-BrainTools

The SAP Group has also recognized this potential. The global market leader in enterprise applications and business AI plans to integrate TabPFN into its products in the future, enabling business users to run “what-if” scenarios, without having to train their own models. The AI is designed to adapt spontaneously to any business use case, which, according to SAP, should lead to a faster return on investment while ensuring compliance with the GDPR.

With over 3 million downloads, PriorLabs Tab PFN is a widely used open-source tool for tabular AI that supports a dynamic developer ecosystem. SAP intends to continue supporting this open-source strategy.

PriorLabs’ statement: https://priorlabs.ai/blog-posts/priorlabs-next-chapter

SAP’s press release: https://news.sap.com/germany/2026/05/sap-plant-prior-labs-zu-uebernehmen-und-europas-fuehrendes-frontier-ki-lab-aufzubauen/

Original Paper: Hollmann, N., Müller, S., Purucker, L. et al. Accurate predictions on small data with a tabular foundation model. Nature 637, 319–326 (2025). https://doi.org/10.1038/s41586-024-08328-6