Enterprise software giant SAP is making a bold bet on structured data AI. On Monday, the company announced its intention to acquire Prior Labs, a German startup founded just 18 months ago that specializes in tabular foundation models (TFMs). The acquisition price was not disclosed, but SAP plans to invest €1 billion (approximately $1.16 billion) into the business over the next four years to transform it into a dedicated AI lab for enterprise data.

The move comes at a critical time for SAP. The company's stock has dropped significantly in 2026, partly due to the so-called 'SaaSpocalypse' — a market correction impacting software-as-a-service valuations. Meanwhile, the broader tech industry is racing toward agentic AI, where autonomous software agents perform tasks across enterprise systems. SAP is both embracing and controlling this trend: while it builds its own AI capabilities, it is also tightening restrictions on which agents can access its products.

The Prior Labs Deal

Sources close to the deal told Pathfounders that the acquisition was a healthy exit for Prior Labs' founders — Frank Hutter, Noah Hollmann, and Sauraj Gambhir. The deal was 'almost all cash,' with well over half a billion dollars paid upfront. The trio founded Prior Labs in late 2024 with a focus on building AI models that could understand and predict from tabular data — the rows and columns of databases and spreadsheets that form the backbone of enterprise operations.

Their flagship model, TabPFN, has gained significant traction in the developer community. According to a blog post from the founders, its open source models have been downloaded over three million times. This popularity stems from the fact that most enterprise AI challenges involve structured data, not just unstructured text. Large language models (LLMs) like GPT-4 excel at language tasks but often struggle with the precise numerical and relational reasoning needed for financial forecasting, inventory management, or customer churn prediction.

SAP's CTO Philipp Herzig emphasized this point: 'Early on, SAP recognized that 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.' Prior Labs' models are designed to complement language models by handling the tabular component, and SAP plans to integrate them into its existing AI infrastructure, including SAP AI Core and SAP Business Data Cloud.

Agentic AI and the NemoClaw Decision

While investing in its own AI lab, SAP is also taking a defensive stance on agentic AI. The company has updated its API policy to explicitly prohibit AI agents from accessing its products unless they use 'SAP-endorsed architectures.' This effectively blocks OpenClaw, an open-source agent framework, and any other unauthorized agents. The Information first reported on this restriction.

However, SAP has greenlit one specific agent framework: NemoClaw, which is built on Nvidia's Agent Toolkit. Nvidia announced in March that SAP's Joule agents support this toolkit, making NemoClaw the go-to choice for customers who want to deploy agents within SAP's ecosystem. NemoClaw is described as an enterprise-ready, security-focused method for deploying OpenClaw agents, but it is a separate offering controlled by Nvidia and SAP.

This approach contrasts sharply with that of Salesforce, another incumbent caught in the SaaSpocalypse. Salesforce has opted for a more open strategy, allowing enterprise customers to choose their own agents — including OpenClaw if they wish — through its new Headless 360 architecture. SAP's stricter approach reflects its desire to maintain control over its platform and ensure security, especially as AI agents become more autonomous.

Background on Tabular Foundation Models

Tabular foundation models represent a growing niche in AI research. Unlike LLMs trained on text, TFMs are trained on millions of tables to learn patterns in relational data. They can perform tasks like classification, regression, and imputation without needing task-specific training examples. Prior Labs' TabPFN, for instance, is based on a transformer architecture that processes table-like inputs and outputs predictions.

The potential for enterprise applications is enormous. Companies like SAP, Oracle, and Microsoft rely on structured data for critical business processes — from supply chain optimization to financial reporting. Yet most AI investment has flowed into language models. Prior Labs aimed to fill that gap, and SAP's backing gives it the resources to scale.

SAP has not been idle in other AI areas. It previously invested in OpenAI rival Anthropic, as well as Aleph Alpha and Cohere, which now intend to merge into a combined European AI powerhouse. The company also developed its own relational pretrained transformer model, SAP-RPT-1. But the Prior Labs acquisition is a significant shortcut to becoming a leader in structured data AI.

Looking Ahead

Under the deal, Prior Labs will operate as an independent unit to preserve its research velocity. The founders have promised to keep the open source versions of their models available, while SAP invests in productization. Founder and CEO Frank Hutter expressed optimism on social media, saying the 'massive boost' from SAP could help Prior Labs become a 'globally-leading frontier AI lab for structured data — in Europe, in the open.'

Prior Labs had previously raised a modest $9.3 million in pre-seed funding from Balderton Capital in February 2025. This exit is seen as one of Germany's biggest venture outcomes, according to Balderton partner James Wise. Meanwhile, SAP's stock has shown a slight uptick since the announcement, suggesting investor approval.

The broader context is that enterprise AI is still in its early stages. As OpenAI's COO admitted last February, 'we have not yet really seen AI penetrate enterprise business processes.' SAP is betting that structured data AI will be the key to unlocking that penetration, and it is willing to spend over a billion dollars to make it happen. With its strict agent policy, SAP is also signaling that it intends to manage the transition to agentic AI on its own terms.

Source: TechCrunch News