Concept card

SAP × Prior Labs — Frontier AI Lab Europe

As of 2026-10-10

Edited by Cédric Mary, sole editor and publication director — 27 years in the SAP ecosystem. How these pages are produced, and where automation is used.

What is SAP × Prior Labs?

SAP announced the acquisition of Prior Labs on 4 May 2026, closed it on 17 July 2026 and committed more than €1 billion over four years to turn the Freiburg start-up behind TabPFN into a frontier AI lab for structured data; the first product consequence, TabPFN-3.5 Plus, became generally available in SAP AI Core on 15 September 2026.

What happened, in order

On 4 May 2026 SAP announced its intention to acquire Prior Labs, a start-up headquartered in Freiburg, Germany, with offices in Berlin and New York City. Prior Labs is the company behind TabPFN, the best-known tabular foundation model (TFM): at the time of the announcement SAP cited more than three million downloads of the open-source TabPFN package and described TabPFN-2.6 as the top-performing model on the TabArena benchmark. SAP committed to invest more than €1 billion over the following four years, stated that Prior Labs would continue to operate as an independent entity, and framed the goal as building "a globally leading frontier AI lab" for the structured data that runs businesses. Deal terms were not disclosed; closing was expected in Q2 or Q3 2026, subject to regulatory approval.

SAP completed the acquisition on 17 July 2026. The completion release repeats the investment commitment and positions the deal as an accelerator of SAP's own TFM work, which "started with SAP-RPT-1" — the relational pretrained transformer SAP made generally available in the generative AI hub in December 2025 (see C165 and C237).

Why it matters

The dates and amount are now fixed and verifiable: announced 4 May 2026, closed 17 July 2026, more than €1 billion committed over four years, Prior Labs kept as an independent entity.
The deal already has a product consequence: TabPFN-3.5 Plus is generally available in SAP AI Core since 15 September 2026, with tabular orchestration planned for it from November 2026.
Customers now get two tabular foundation-model families (SAP-RPT and TabPFN) behind one AI Core integration point — the consulting value is the benchmark and the decision framing, not the tuning.

Key points

4 May 2026: SAP announces the acquisition of Prior Labs (Freiburg; offices in Berlin and New York City), with more than €1 billion to be invested over four years.
17 July 2026: acquisition completed; Prior Labs continues as an independent entity and frontier AI lab for structured data.
TabPFN predicts on tables by in-context learning — no training, no hyperparameter search; TabICL is a separate open model from Inria, not a Prior Labs product.
15 September 2026: TabPFN-3.5 Plus generally available in SAP AI Core; tabular orchestration for it planned from November 2026.
SAP positions the deal as accelerating the tabular foundation-model work that started with SAP-RPT-1 — benchmark both families on your own data before choosing.
SAP-RPT's own release cadence is now roughly monthly: SAP-RPT-1 (GA December 2025) → SAP-RPT-1.5 (GA August 2026) → SAP-RPT-1.6 and 1.6-large (GA 7 September 2026, plus a Tabular Orchestration workflow and an RPT Playground API capped at 1,000 requests/hour).
Beyond Plus, Prior Labs also ships TabPFN-3.5 base, Thinking (most accurate, for grouped and temporal data) and an alpha Fast variant; only Plus is generally available inside SAP AI Core as of 23 September 2026.
A 15–29 September 2026 promotion cut standard TabPFN-3.5-family token rates by 50% for API/MCP users — a time-boxed pricing signal, not the ongoing rate.

Terms used on this page

Tabular foundation model (TFM): A model pretrained once on large amounts of tabular data that predicts on new tables without task-specific training.
TabPFN: Prior Labs' tabular foundation model, pretrained on synthetic tabular data; predicts by in-context learning in a single forward pass.
In-context learning (tabular): Passing labelled context rows together with the rows to predict in one request, so the model infers the pattern at inference time.
TabPFN-3.5 Plus: Variant of TabPFN-3.5 that adds processing for text-rich tables; generally available in SAP AI Core since 15 September 2026.
Tabular orchestration: Managed SAP AI Core workflow that retrieves and selects context rows from a registered data source before calling a tabular model.

Sources

Full card available to members. What the full card adds: the full decision framework · the SAP vs Snowflake / Databricks / Fabric comparison · the common pitfalls and their fix · the cheat sheet · the architecture schemas · the code blocks · the facts worth quoting.

Open in the app →