SAP closed its acquisition of Prior Labs on July 17, 2026, completing a deal first announced in May and committing more than €1 billion to scale the Freiburg-based startup into a globally leading frontier AI lab for structured business data.

The deal is notable for what it is not: it is not about large language models, chatbots, or voice assistants. It is a bet that the most valuable untapped opportunity in enterprise AI sits inside the spreadsheets, ERP tables, and operational databases that run actual businesses.

What Prior Labs Built

Prior Labs is the pioneer of Tabular Foundation Models (TFMs), a category of AI purpose-built for structured data rather than free-form text. Founded roughly 18 months ago by researchers Frank Hutter, Noah Hollmann, and Sauraj Gambhir, the company’s TabPFN model series was published in Nature and set state-of-the-art results on tabular benchmarks across hundreds of independent academic studies.

Where a large language model learns from text on the internet, a tabular foundation model learns from the kind of data that every business already has: rows and columns representing customers, transactions, invoices, and operations. The model can then make predictions directly from that structured data without the preprocessing and feature engineering that conventional machine learning requires.

SAP CTO Philipp Herzig put it plainly: “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.”

What This Means Practically

For SAP’s 300,000 plus customers, the acquisition accelerates capabilities that matter in day-to-day operations. Tabular AI can predict:

• Payment delays before they happen
• Supplier risk before a disruption occurs
• Upsell opportunities based on purchase history
• Customer churn risk weeks or months in advance

These are not generic AI use cases. They are predictions made directly from the data already living inside ERP, CRM, and supply chain systems. No custom model training, no data science team required.

SAP has been moving in this direction already with its SAP-RPT-1 model, and the Prior Labs acquisition accelerates that effort significantly. Prior Labs will continue operating as an independent entity, giving it room to run frontier research while benefiting from SAP’s compute infrastructure and enterprise distribution.

Why the Timing Makes Sense

The deal closed at a moment when enterprises are questioning the ROI of AI projects they rushed into in 2024 and 2025. Many of those projects involved deploying LLMs against unstructured documents, internal wikis, or customer service chat. The results have been mixed. The harder problem, predicting what a business needs to do next based on what has already happened, remained largely unsolved.

Tabular AI addresses that problem directly. It does not replace conversational AI tools. It works alongside them, turning historical business data into actionable predictions rather than just summaries or responses.

SAP’s €1 billion commitment over four years covers compute infrastructure, research hiring, and long-term model development. The scale of that commitment reflects how seriously the company views structured data AI as the next major battleground in enterprise software.

What This Means for Business

If you run a business on SAP systems, or plan to, the Prior Labs acquisition signals where AI assistance is heading: from answering questions about your documents to predicting what happens in your operations before it happens.

For business leaders evaluating AI investments right now, this deal reinforces a principle that tends to get lost in the noise around generative AI: the best AI application for most companies is one trained or adapted to the patterns in their own business data. Language models alone cannot do that. Tabular models, built on the structured data your business already generates, can.

This is also relevant for data and analytics teams. As tabular foundation models mature, the skill most in demand will not be prompt engineering. It will be understanding what structured data tells you, knowing how to prepare it, and knowing which predictions are worth acting on.

Enterprise DNA’s data training programs cover exactly this territory: Power BI, Python, SQL, and the data fundamentals that let teams understand and act on model outputs, not just consume them.

The winners from advances like Prior Labs’ TFMs will be the organizations whose people already understand their data.