SAP buys Prior Labs for more than €1 billion

SAP has acquired Prior Labs, a German-founded AI company focused on structured enterprise data, in a deal valued at more than €1 billion. The acquisition comes 18 months after Prior Labs was founded and gives SAP control of a young research lab working on tabular foundation models, a category aimed at prediction tasks on business data rather than language generation.

The companies did not disclose Prior Labs' revenue, headcount, prior funding, or the precise structure of the transaction. SAP said its investment will support computing infrastructure, hiring and longer-term research work at Prior Labs.

Prior Labs is built around TabPFN, a pre-trained model designed to work across structured datasets without requiring customers to train a separate model for each use case. The company says the technology can be used for tasks including forecasting demand, estimating supplier risk, predicting churn and identifying payment delays.

For SAP, the rationale is direct: its software sits on large volumes of enterprise data, much of it structured in tables rather than documents or chat logs. Buying Prior Labs gives the company a research asset in a part of AI that is closer to ERP, supply chain and finance workflows than the large language model race that has dominated enterprise AI budgets.

Prior Labs to keep its own brand

SAP said Prior Labs will continue to operate under its existing name, leadership, research agenda and customer relationships. The company also said Prior Labs will continue to publish research and make its models openly available with SAP's backing.

That independence claim will be tested over time. Prior Labs is now owned by one of the largest enterprise software companies in the world, and SAP's stated plan is to combine Prior Labs' model work with its enterprise data ecosystem and customer reach.

Prior Labs says its technology is already being used by Hitachi to help prevent train failures and by TD to improve financial forecasting. The company also says its models have been used in hundreds of published research projects, including work on pancreatic cancer diagnosis, wildfire prediction and battery materials.

The company also pointed to TabPFN-3-Thinking, its latest model, which it describes as state of the art and enterprise-grade for prediction tasks. SAP did not provide independent benchmark details in the announcement, and the companies did not disclose commercial adoption metrics for the model.

SAP frames the deal as a structured-data AI bet

Frank Hutter, Prior Labs' co-founder and CEO, said the company began as a research project 18 months ago and now has the resources to work on problems it could not previously address. He said advancing tabular foundation models requires stronger data environments, deployment channels and long-term research funding, and that SAP can provide those.

SAP CTO Philipp Herzig said the company identified structured enterprise data, rather than large language models, as a major open opportunity in enterprise AI. He said Prior Labs had defined the tabular foundation model category and built a leading research team in it, citing the company's performance on public benchmarks.

SAP said the acquisition will let Prior Labs pursue multi-year programs in enterprise AI, scientific discovery, causality, relational data and agentic systems. The company also cited potential work in medical data and materials science, but did not attach timelines, budgets or product release dates to those ambitions.

This story draws on original reporting from Tech.eu.