Today, 6 May 2026
Good morning, Human. Sometimes the most consequential moves in enterprise technology happen not in the flashy world of chatbots and image generators, but in the unglamorous realm of spreadsheets and databases. Yesterday, Germany's most valuable company made exactly that bet.
In Brief
SAP is acquiring Freiburg-based Prior Labs and committing more than €1 billion over four years to build what it calls a globally leading frontier AI lab for structured data. This matters because it represents the largest enterprise AI research investment ever made in Europe by a European company, and it signals a strategic conviction that the next frontier in business AI isn't large language models but tabular foundation models (TFMs) that can actually understand the rows, columns, and statistical patterns where enterprise data lives. For Europe, this is the kind of outcome policymakers have been asking for: academic talent converting into commercial success, with the strategic acquirer itself European.
On 19 May in Vienna, questions like these won't just be topics for discussion. They'll be working questions, in the room where decisions actually get made. Human x AI Europe.
The Lead: SAP's Structured Data Gambit
The numbers alone tell a story. Prior Labs raised €9 million in a pre-seed round in February 2025. Eighteen months later, SAP is acquiring the company and pledging more than €1 billion to scale it into a frontier AI lab. Balderton partner James Wise called it one of Germany's biggest ever venture outcomes. The speed is almost disorienting.
But the strategic logic is clear. Large language models, for all their capabilities, struggle with the kind of data that actually runs businesses: tables, numbers, statistical relationships. As SAP CTO Philipp Herzig put it:
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.
Philipp Herzig
Prior Labs' flagship model, TabPFN, has been downloaded over three million times and currently sits atop TabArena, the standard benchmark for tabular machine learning. The model does something most foundation models cannot: it runs in a single forward pass, without task-specific training, and matches or exceeds the accuracy of tuned tree-based models that have run for hours. For enterprise systems filled with customer records, financial transactions, and manufacturing telemetry, that's the technical inflection point.
The deal structure is notable. Prior Labs will continue operating as an independent entity, retaining its brand, headquarters in Freiburg, offices in Berlin and New York, and its open-source commitments. Yann LeCun and Bernhard Schölkopf will serve on its scientific advisory board. SAP is explicitly committing to continue the open-source strategy around TabPFN. The models will integrate into SAP AI Core, SAP Business Data Cloud, and the agentic layer with Joule.
There's a defensive dimension here too. TechCrunch reports that SAP has blocked OpenClaw and other agent technologies it hasn't explicitly authorized, while endorsing NVIDIA's NemoClaw through its Joule Agents integration. In the emerging agentic AI landscape, SAP is choosing its partners carefully.
The Agent Architecture Question
The SAP-NemoClaw connection deserves unpacking. When NVIDIA announced NemoClaw at GTC 2026 in March, it positioned the stack as the enterprise security layer that OpenClaw was missing. OpenShell provides sandboxing, least-privilege access controls, and a privacy router. SAP is among the launch partners, using NVIDIA Agent Toolkit to enable AI agents through Joule Studio on SAP Business Technology Platform.
This creates an interesting dynamic. SAP customers will be authorized to use NemoClaw agents, but not OpenClaw directly. The company is essentially building a walled garden for agentic AI within its ecosystem, one that prioritizes security and governance over the open experimentation that made OpenClaw the fastest-growing open-source project in history.
The contrast with Salesforce is stark. Salesforce is allowing enterprises to choose their own agents, including OpenClaw, through its new Headless 360 architecture. Two different philosophies for the same problem: how do you let AI agents operate inside enterprise systems without creating security nightmares?
The Policy Situation: EU-Japan Digital Partnership Deepens
While SAP was making its acquisition announcement, Brussels was hosting the fourth meeting of the EU-Japan Digital Partnership Council. Executive Vice-President Henna Virkkunen met with Japanese ministers to agree on new steps in regulatory, research, and industry cooperation across data, AI, quantum, semiconductors, and digital infrastructure.
The joint statement reaffirms what both sides call a strategic partnership as like-minded partners sharing fundamental democratic values, the rule of law, and a human-centric approach to digital transformation. The practical deliverables include improved cross-border data flows, interoperable digital identities, and strengthened cooperation on platform regulation.
This matters for European AI companies operating globally. Japan represents both a market and a regulatory model that aligns more closely with European values than the U.S. or Chinese approaches. The partnership creates pathways for European AI companies to expand into Asia-Pacific markets with regulatory frameworks that don't require fundamental redesigns of their compliance architecture.
The Regulatory Calendar: Protecting Young People Online
The Commission's push on child online safety continues to accelerate. The policy framework now spans the Digital Services Act, the AI Act's prohibitions on manipulative systems targeting children, the Audiovisual Media Services Directive, and the recently announced age verification app.
The statistics driving this push are stark: 97% of young people in the EU use the internet daily, and social media is the main information source for 65% of them. Over 9 in 10 Europeans support mechanisms to restrict children's ability to access age-inappropriate content. The Commission's age verification app, currently being piloted by Denmark, France, Greece, Italy, and Spain, allows users to prove they're over 18 without sharing personal data.
Member states are moving faster than Brussels. Greece has announced a social media ban for children under 15, effective January 2027. France has enacted similar restrictions. Spain, Austria, Denmark, and the Netherlands are preparing their own rules. The risk, as always with fragmented national approaches, is a patchwork that creates compliance headaches for platforms while potentially pushing young users toward less regulated corners of the internet.
The Special Panel on child safety online, advising President von der Leyen, met for its second session in April. The panel includes researchers like Matthias Brand from the University of Duisburg-Essen, whose work on addiction-like internet use informs the scientific basis for policy interventions.
The Numbers That Matter
- €1 billion+: SAP's committed investment in Prior Labs over four years, the largest enterprise AI research investment by a European company in Europe.
- 18 months: Time from Prior Labs' founding to acquisition, one of the fastest venture outcomes in German tech history.
- 3 million+: Downloads of Prior Labs' open-source TabPFN model.
- $17.6 billion: European venture funding in Q1 2026, up nearly 30% year over year, with AI claiming more than 50% of total funding for the first time.
- 97%: Share of young people in the EU who use the internet daily.
- 92%: Europeans who identify cyberbullying as the primary online threat to children.
The Week Ahead
- May 7-8: IBM Think 2026 continues in Boston, with announcements on sovereign AI and enterprise AI operating models.
- May 12: EIC Pathfinder 2026 deadline for breakthrough AI technologies, including neuromorphic computing and explainable AI.
- May 19: Human x AI Europe convenes in Vienna, bringing together policymakers, founders, and researchers to work through the governance questions that matter.
- Q2-Q3 2026: Expected close of SAP's Prior Labs acquisition, pending regulatory approval.
The Thought That Lingers
There's something almost poetic about SAP's bet. While the AI conversation has been dominated by language models that generate text and images, the company that built its empire on enterprise resource planning is doubling down on the data that actually runs businesses: the tables, the spreadsheets, the structured records that don't make for exciting demos but determine whether invoices get paid and supply chains hold together.
The question isn't whether tabular foundation models matter. They obviously do. The question is whether Europe can sustain this kind of outcome: academic talent converting into commercial success, with the strategic acquirer itself European, and the resulting lab remaining on the continent. Prior Labs' founders wrote that their mission has not changed. It just got accelerated. Whether that acceleration produces a globally competitive frontier lab or gets absorbed into corporate bureaucracy is the test that comes next.
Frequently Asked Questions
What are tabular foundation models and why do they matter for enterprise AI?
Tabular foundation models (TFMs) are AI systems specifically designed to understand structured data like spreadsheets, databases, and tables. Unlike large language models that work with text, TFMs can process the rows, columns, and statistical relationships that make up most enterprise data. They matter because most business-critical information exists in structured formats, not natural language.
How does SAP's acquisition of Prior Labs compare to other European AI deals?
At over €1 billion committed over four years, this represents the largest enterprise AI research investment ever made by a European company in Europe. The 18-month timeline from Prior Labs' founding to acquisition is also one of the fastest venture outcomes in German tech history, demonstrating the rapid pace of AI consolidation.
What is the difference between OpenClaw and NemoClaw agent technologies?
OpenClaw is an open-source agent framework that became the fastest-growing open-source project in history, while NemoClaw is NVIDIA's enterprise-focused agent stack that includes security features like sandboxing and privacy controls. SAP has chosen to support NemoClaw while blocking unauthorized OpenClaw implementations, prioritizing security over open experimentation.
Why is the EU-Japan digital partnership significant for European AI companies?
The partnership creates regulatory alignment between Europe and Japan on AI governance, data flows, and digital infrastructure. This allows European AI companies to expand into Asia-Pacific markets without completely redesigning their compliance architecture, as Japan's regulatory approach aligns more closely with European values than U.S. or Chinese frameworks.
What are the key elements of the EU's child online safety framework?
The framework spans multiple regulations including the Digital Services Act, AI Act prohibitions on manipulative systems targeting children, and the Audiovisual Media Services Directive. It includes an age verification app being piloted by five member states and addresses the fact that 97% of EU young people use the internet daily, with social media as the main information source for 65% of them.