The AI industry is obsessed with models that write and talk. SAP has just completed a large bet on a different kind entirely, one built for structured data, as its acquisition of Prior Labs closed.
The deal is now done, with regulatory approvals secured and the Freiburg-based lab operating inside SAP. The German software giant is putting more than €1bn behind it over four years.
What Prior Labs actually does
Most AI foundation models handle text, images or audio. Prior Labs builds foundation models for tables, the rows and columns of structured data that businesses actually run on.
Its TabPFN model, published in Nature, set the state of the art on tabular prediction across hundreds of independent studies. It forecasts and classifies from spreadsheets and databases rather than from prose.
That mundane-sounding capability has enormous reach. It has already been used for financial risk, loan approvals, predictive maintenance on railways, and cancer diagnostics.
Why SAP wants it
The fit is almost too neat. SAP’s entire business is the structured enterprise data sitting in company systems, exactly the kind of data Prior Labs’ models are built to read.
It also fills a real gap. The structured-data layer is the part of enterprise AI most companies still cannot do well, even as they pour money into chatbots.
SAP has been spending hard to catch up, from an autonomous enterprise built on 200-plus AI agents to embedding new automation into its AI studio. It even froze hiring and travel to fund the push.
A rare European win
The strategic angle is what makes this notable. A European software giant has bought a European frontier lab and kept it open and independent.
That is precisely the academic-to-commercial outcome EU policymakers have spent years trying to manufacture. Prior Labs keeps its brand, its Freiburg base, its open-source models, and an advisory board that includes Meta’s Yann LeCun.
It stands out against a thin European record. The continent’s other marquee AI names have tended to merge or fold into bigger entities rather than anchor a champion at home.
The bigger signal
The deal marks tabular AI’s arrival as a serious category. Microsoft, Google, and AWS are all moving into structured-data models, and SAP has bought one of the most credible independent efforts.
It is also a quiet correction to the chatbot fixation. The most valuable AI in a business may not be the one that writes emails, but the one that predicts which loans default or which train fails next.
SAP is betting that the next enterprise-AI frontier is the data companies already have. The models stay open, the lab stays in Freiburg, and the wager is now more than €1bn real.
