SAP buys Dremio and Prior Labs

Enterprise AI is running into a boring problem that turns out to be the whole problem — the data is scattered, structured, and hard to trust. SAP’s answer on May 4 was to buy two different pieces of that stack at once: Dremio for the data layer and Prior Labs for the model layer. That matters because SAP isn’t pitching another chatbot. It’s trying to build a system where AI agents can actually do work across finance, supply chain, and operations without wandering off into bad data. (news.sap.com) ### Why these two companies? Dremio gives SAP a lakehouse platform built to query data across systems, including non-SAP sources, in real time. Prior Labs works on tabular foundation models — AI models tuned for rows-and-columns business data, which is the format mos(news.sap.com)ver it. (news.sap.com) ### Why is tabular data such a big deal? Because enterprise AI is not mostly about images or internet text. It is about invoices, ledgers, purchase orders, inventory tables, HR records, and compliance logs. Large language models are great at language, but business sy(news.sap.com)fit for that reality — and it says Prior Labs builds on work already started with SAP-RPT-1. (news.sap.com) ### What does Dremio fix? The ugly gap between SAP data and everything else. Most big companies do not run on one vendor’s stack, even if SAP is the system of record for core processes. Dremio’s pitch is that it can unify SAP and non-SAP data for analytics and AI workloads (news.sap.com)iness Data Cloud and for real-time agentic AI. The Dremio transaction is expected to close in Q3 2026, pending approvals. (news.sap.com) ### Why mention agents at all? Because the market is shifting from “AI that answers” to “AI that acts.” SAP’s releases frame both deals around agentic AI — software agents that can retrieve data, make decisions inside guardrails, and trigger workflows. But agents are(news.sap.com)nfrastructure layer suddenly matters more than the demo layer. (news.sap.com) ### Is SAP alone here? Not really. Microsoft made Agent 365 generally available on May 1 as a control plane to observe, govern, and secure AI agents across an organization. Citi also introduced Arc, its internal platform for building and scaling AI agents inside the (news.sap.com) visible, and tied to existing controls. (microsoft.com) ### What is SAP really building? Basically, a full enterprise agent stack. Business Data Cloud is the shared data layer. Dremio broadens the reach into non-SAP environments. Prior Labs strengthens the model side for struct(microsoft.com)our years around that effort. That is not a small feature add — it is a platform bet. (news.sap.com) ### Why should finance teams care? Finance is where bad data and weak controls get expensive fast. If AI agents are going to reconcile accounts, prepare analysis, route approvals, or surface exceptions, the winning systems will be the ones that leave an audit trail and pull(news.sap.com) enterprise records. (news.sap.com) The bottom line is simple — SAP just spent to solve the two hardest parts of enterprise AI at the same time: getting the data in one place and making models that understand business tables. If that works, agents stop looking like assistants and start looking like infrastructure.