SAP’s €1B AI bet targets business data, not chatbots
▼ Summary
– SAP completed the acquisition of Prior Labs, a European AI startup, for over €1 billion over four years.
– Prior Labs builds foundation models for tabular data (spreadsheets and databases), not text or images, and its TabPFN model set a state-of-the-art for tabular prediction.
– The acquisition fills a gap in SAP’s enterprise AI, as most companies struggle with structured-data AI despite heavy investment in chatbots.
– The deal is a rare European win, keeping Prior Labs independent in Freiburg with open-source models and an advisory board including Yann LeCun.
– The acquisition signals tabular AI’s emergence as a serious category, correcting the industry’s chatbot focus by prioritizing predictions from existing business data.
The global artificial intelligence conversation remains largely dominated by models that generate text and speech. SAP, however, has just closed a major acquisition that signals a very different priority. With the completion of its purchase of Prior Labs, the German software giant is now placing a €1 billion bet on AI designed for the world of spreadsheets and databases, not chatbots.
With all regulatory hurdles cleared, the Freiburg-based lab is now fully integrated into SAP’s operations. The company has committed over €1 billion to this initiative over the next four years, underscoring the strategic weight of the move.
The Core of Prior Labs’ Technology
While most foundation models are trained on text, images, or audio, Prior Labs focuses exclusively on tabular data. Its flagship model, TabPFN, was published in the journal Nature and has set a new benchmark for predictive analysis on structured data across hundreds of independent studies. Instead of parsing prose, it forecasts and classifies information directly from spreadsheets and databases.
This seemingly niche capability has surprisingly broad applications. The technology is already deployed in areas like financial risk assessment, loan approvals, predictive maintenance for railways, and even cancer diagnostics.
Why SAP Made the Move
The synergy between the two companies is almost seamless. SAP’s entire ecosystem revolves around the structured enterprise data that flows through its clients’ systems. That is precisely the type of data Prior Labs’ models are built to read and interpret.
This acquisition also fills a critical void. The structured-data layer remains one of the most difficult parts of enterprise AI to master, even as companies spend heavily on conversational agents. SAP has been aggressively playing catch-up, building an autonomous enterprise framework powered by over 200 AI agents and embedding new automation into its AI studio. To fund this pivot, the company even froze hiring and cut travel expenses.
A Rare Victory for European AI
The strategic implications extend beyond pure technology. A major European software firm has acquired a frontier European AI lab and chosen to keep it open and independent.
This outcome is precisely what EU policymakers have been trying to engineer for years: a smooth transition from academic research to commercial scale. Prior Labs will retain its brand, its Freiburg headquarters, its open-source models, and an advisory board that includes Meta’s Yann LeCun.
It stands in stark contrast to Europe’s otherwise thin record in AI. Other notable names from the continent have typically been absorbed or dissolved into larger, often American, entities rather than being allowed to anchor a homegrown champion.
The Broader Signal
This deal marks the arrival of tabular AI as a serious category. Tech giants like Microsoft, Google, and AWS are all moving into structured-data models. By acquiring one of the most credible independent players, SAP has secured a leading position.
It also represents a quiet correction to the industry’s chatbot obsession. The most valuable AI for a business may not be the one that writes emails. It could be the one that predicts which loan will default or which train part is about to fail.
SAP is betting that the next frontier of enterprise AI lies in the data companies already own. The models will remain open, the lab will stay in Freiburg, and the wager is now over €1 billion in real terms.
(Source: The Next Web)
