SAP closes Prior Labs deal with €1 billion investment plan

SAP has completed its acquisition of Prior Labs, the Freiburg AI startup, for an undisclosed purchase price and said it will put more than €1 billion into the company through 2030. The deal is a notable European deep tech exit because Prior Labs moved from university research to acquisition in about 18 months, with only €9 million in outside funding before SAP stepped in.

SAP announced its plan to buy the company on May 4, 2026, and confirmed the close on July 17, 2026. The company has not broken out how much of the €1 billion commitment is acquisition consideration versus future operating investment, hiring, compute or product integration.

From TabPFN research to SAP ownership

Prior Labs was formed in 2024 by Frank Hutter, a University of Freiburg professor known for work in automated machine learning, Noah Hollmann, a computer scientist with experience at Google and Boston Consulting Group, and Sauraj Gambhir, whose background includes venture capital and M&A.

The company commercialized TabPFN, a foundation model built for tables, spreadsheets and databases. That puts it in the category of tabular foundation models, a branch of AI aimed at structured enterprise data rather than text generation.

Prior Labs raised €9 million in pre-seed funding in February 2024. Balderton Capital and XTX Ventures led the round, with Atlantic Labs, Hector Foundation and Galion.exe also participating. Angel investors included Robin Rombach, co-founder of Black Forest Labs, and Thomas Wolf, co-founder of Hugging Face. No later private round was disclosed before SAP’s acquisition.

Why SAP wants tabular AI

SAP’s rationale is direct: its customers run large structured datasets across finance, operations, supply chains and HR. Philipp Herzig, SAP’s chief technology officer, said SAP saw the main enterprise AI opening in systems built for business data rather than large language models. He also said Prior Labs had defined the tabular foundation model category and led public benchmarks from the start.

Prior Labs says its TabPFN series, published in Nature, has reached 4 million downloads, compared with more than 3 million in May. The company says the model has been used in research areas including pancreatic cancer diagnosis and wildfire prediction. It also says Hitachi has used the technology for train-failure prediction, while TD has applied it to financial forecasting.

The company’s advisory bench includes Bernhard Schölkopf and Yann LeCun, a Turing Award winner. Prior Labs says it will keep its brand, leadership and research agenda inside SAP, while continuing to publish research and release models openly. SAP has not said how it will measure that commitment once the lab is tied to product priorities.

A fast exit in a competitive field

Dealroom said European deep tech investment reached $20.3 billion in 2025, but large outcomes have remained less common than in the U.S. Prior Labs gives Europe a rare fast-cycle example: academic research, a single seed-stage financing and a strategic software buyer in less than two years.

The category is also getting more crowded. Kumo.ai has developed KumoRFM and says it outperforms TabPFN on multi-table tasks, while TabPFN leads on single-table accuracy. Fundamental has raised $255 million in Series A funding for a large tabular model, although peer-reviewed benchmarks have not yet been published, according to the company information available.

SAP has also agreed to acquire Dremio, another AI-focused company, during the same period. For founders and investors in European AI infrastructure, the Prior Labs deal will be watched less for the headline investment pledge than for whether SAP lets the team keep publishing and shipping at research-company speed.

This story draws on original reporting from TechFundingNews.