SAP Drops 1 Billion Euros on Prior Labs to Build an AI Lab That Does Not Make Chatbots
While everyone else is racing to build the next large language model, Europe's largest enterprise software company just spent one billion euros on a startup that makes AI for spreadsheets and ERP databases.
Three Key Takeaways
One billion euros over four years. SAP officially completed its acquisition of German AI startup Prior Labs on July 17, committing over one billion euros to build it into Europe's leading frontier AI lab. Prior Labs will continue to operate independently. The company was founded just 18 months before the acquisition — a startup-to-exit speed nearly unprecedented in European AI.
Not language models — tabular models. Prior Labs pioneered Tabular Foundation Models (TFMs), a category of AI purpose-built for structured data. It does not process natural language text; it processes the rows and columns inside enterprise databases, the orders and inventory in ERP systems, the numbers in financial reports. This is a track entirely overshadowed by the LLM hype: the most valuable data in global enterprises is overwhelmingly not documents and conversations, but structured business data.
SAP's strategic logic. SAP's customers account for 77% of global commercial transactions — meaning the world's largest pool of enterprise structured data already lives inside SAP systems. Acquiring Prior Labs is not about chasing the LLM trend; it is about building an AI layer on top of SAP's own data moat. While competitors deploy ChatGPT as customer service bots, SAP chose a different path: making AI natively read operational data to forecast, decide, and detect anomalies.
WangDou's Take
Paying one billion euros for an 18-month-old company sounds like a sucker's deal. But SAP's math is actually straightforward: the entire world is training AI that writes poetry, chats, and generates images, yet if you ask any CFO what they actually want AI to do, the answer is probably "read my financial data and tell me what is wrong." ChatGPT cannot do that — feed it a 100,000-row ERP table and it will hallucinate numbers that do not exist. Prior Labs' Tabular Foundation Model is built to solve exactly this problem. SAP sits on 77% of the world's commercial transaction data but has always lacked a native AI engine to digest it — now it spent one billion euros to fill that gap. This deal is not sexy, but it might be the most pragmatic AI acquisition of the year.
Source: SAP News Center, Tech.eu, EU-Startups