SAP buys Prior Labs to boost enterprise AI research

SAP has agreed to acquire Prior Labs, bringing the German artificial intelligence specialist into its orbit while allowing it to remain an independent unit.

SAP plans to invest more than EUR €1 billion over the next four years to expand Prior Labs into a Europe-based AI research lab focused on structured business data. Financial terms were not disclosed, and the transaction remains subject to regulatory approval.

The deal centres on tabular foundation models, or TFMs, a branch of artificial intelligence designed to work with structured data in tables, spreadsheets and business databases. SAP is betting these models will prove more useful than large language models for many business tasks that depend on numerical and statistical data.

According to the companies, those tasks include predicting payment delays, supplier risk, upselling opportunities and customer churn. The acquisition also builds on SAP's earlier work in the field through its SAP-RPT-1 model.

SAP argues that conventional large language models have a limited understanding of tables, numbers and statistical relationships. TFMs, by contrast, are built specifically to analyse tabular data and generate predictions from it.

Germany-based Prior Labs is best known for TabPFN, an open-source tool for tabular AI that has recorded more than 3 million downloads, according to SAP. Its latest model, TabPFN-2.6, was described by SAP as the top-performing model on the TabArena benchmark for tabular foundation models.

Research focus

If completed, the deal would give SAP a dedicated AI research operation focused on structured enterprise data, an area central to its software business across finance, procurement, human resources and supply chains. Prior Labs will continue to operate independently to preserve the pace of its research work.

SAP also plans to connect Prior Labs' research more directly to its software products, including SAP AI Core, SAP Business Data Cloud and Joule, its AI assistant.

Philipp Herzig, SAP's Chief Technology Officer, linked the acquisition to the company's longer-term approach to business AI.

"Early on, SAP recognised that the greatest untapped opportunity in enterprise AI wasn't large language models; it was AI built for the structured data that runs the world's businesses," Herzig said. "We built SAP-RPT-1 to prove that conviction for enterprise data. Prior Labs has built a leading TFM on public benchmarks and built one of the leading research teams in this category. Combining their frontier model work with enterprise data and customer reach is how we intend to lead this category globally."

Prior Labs has built its reputation on academic and open-source work in tabular models, an area of AI drawing increasing interest as companies look for systems that can work directly with operational datasets rather than documents or general text. The field aims to improve machine reasoning over structured records at the heart of most corporate software systems.

SAP said the acquisition would help users ask questions in natural language, select datasets and run what-if scenarios without specialist machine learning expertise. It added that the models could make predictions from provided records without additional training, allowing a single model to adapt across different business use cases.

Open-source backing

SAP intends to continue supporting Prior Labs' open-source strategy. That matters because TabPFN has attracted a broad developer community, giving the start-up visibility well beyond its size and helping its models gain adoption among researchers and practitioners.

Prior Labs was founded by Frank Hutter, Noah Hollmann and Sauraj Gambhir. According to SAP, the three lead a team of researchers and practitioners from companies and institutions including Google, Apple, Amazon, Microsoft, G-Research, Jane Street, Goldman Sachs and CERN.

Prior Labs' scientific advisory board is set to include Yann LeCun, Executive Chairman at Advanced Machine Intelligence, and Bernhard Schoelkopf, Director of the Max Planck Institute for Intelligent Systems and ELLIS President. Their involvement underscores SAP's effort to position the new lab as a significant European base for AI research rather than simply an internal product team.

For Prior Labs, the agreement offers access to SAP's data environment and customer base, which could help test tabular models at larger scale in live business settings. That is often a difficult step for AI start-ups whose models perform well in benchmarks but need access to enterprise systems and proprietary datasets to prove commercial value.

Frank Hutter, Prior Labs' Chief Executive, said the transaction would give the company broader backing for its work.

"Over the last 18 months, Prior Labs has built an incredible team, increasing the velocity in tabular foundation models," Hutter said. "Joining the SAP family gives us the resources, data environment and customer reach to take this category to its full potential."