SAP buys Prior Labs and invests over €1 billion in tabular AI

SAP bets on tabular foundation models in 2026, not only chatbots

SAP SE and Prior Labs announced a definitive acquisition agreement on May 4, 2026. Prior Labs is expected to remain independent while SAP commits more than €1 billion over four years to a European frontier AI lab for structured data; the transaction remains subject to regulatory approval.

Why tabular AI is different for ERP data

SAP argues that classic large language models have only a limited understanding of tables, numbers and statistics. Tabular foundation models are designed for business tables such as payment delays, supplier risks, upsell opportunities and churn risk.

Prior Labs brings TabPFN and a research team

Prior Labs points to more than 3 million TabPFN downloads and a research team around Frank Hutter, Noah Hollmann and Sauraj Gambhir. CRN frames the deal, together with SAP's Dremio acquisition plan, as part of a broader data and agentic-AI strategy.

What remains open

The purchase terms were not disclosed, and SAP has not announced final closing beyond regulatory approval. This is therefore a product and research bet, not an instant button inside every SAP system.

Why it matters

For companies, the news matters because most real processes do not live in chat logs. They live in ERP tables: invoices, inventory, offers, late payments and master data. If SAP connects TFMs to SAP AI Core, Business Data Cloud and Joule, AI could move closer to core business processes.

Practical example

A DACH machinery company with 800 employees could use SAP payment and supplier data in 2027 to prioritize 50 risk cases every week. Instead of letting an LLM guess over tables, a TFM would evaluate structured patterns from invoice date, supplier, payment terms and complaint history.