Our Take
This is SAP buying its way into modern data architecture rather than building it, betting that data unification matters more than AI model quality for enterprise adoption.
Why it matters
Most enterprise AI pilots fail because data sits fragmented across systems, not because models are inadequate. SAP is positioning to solve the integration problem that kills AI projects before they scale.
Do this week
Data teams: Audit your current lakehouse strategy before Q3 2026 so you can evaluate whether SAP's unified platform beats your multi-vendor approach.
SAP pays undisclosed amount for Dremio's lakehouse platform
SAP announced it will acquire Dremio, an Apache Iceberg-native data lakehouse platform, in a deal expected to close Q3 2026 pending regulatory approval (terms undisclosed). The acquisition will integrate Dremio's platform with SAP Business Data Cloud to create what SAP calls a unified enterprise lakehouse supporting both SAP and non-SAP data sources.
The combined platform will use Apache Iceberg as its native table format, eliminating data movement and format conversion requirements. SAP will build a universal catalog on Apache Polaris and the Apache Iceberg REST Catalog API, serving as the discovery and semantic layer for SAP Business Data Cloud.
Dremio's serverless, elastic architecture scales automatically with demand spikes and scales down during low usage periods. The platform has been a leading contributor to Apache Iceberg, Apache Arrow, and Apache Polaris open-source projects.
Data fragmentation kills more AI projects than model limitations
SAP CTO Philipp Herzig framed the core problem: "Enterprise AI doesn't stall because the models aren't good enough; it stalls because the data isn't ready for AI agents." The company identifies data fragmentation, proprietary format lock-in, and missing business context as the primary barriers to AI project success at enterprise scale.
The acquisition addresses a structural issue where AI pilots cannot scale due to slow integration of new data sources, duplicated engineering work, and compliance risks when organizations cannot explain AI-driven decisions. By creating an Apache Iceberg-native foundation, SAP aims to eliminate the data preparation bottleneck that precedes AI model deployment.
SAP plans to embed the unified catalog as the foundation of its Knowledge Graph, incorporating business relationships, organizational hierarchies, regulatory classifications, and cross-system lineage as native properties.
Evaluate integrated vs. best-of-breed data architecture
The deal signals SAP's recognition that it needed to acquire rather than build modern lakehouse capabilities. Organizations currently running multi-vendor data stacks should assess whether a single-vendor approach offers sufficient advantages over their existing best-of-breed combinations.
The Apache Iceberg foundation means existing Iceberg-compatible tools should continue working with the unified platform. However, practitioners should verify that SAP maintains Dremio's open-source commitments and doesn't introduce proprietary extensions that create new lock-in scenarios.
Data teams should also consider timing: the Q3 2026 closing date provides a window to evaluate alternative lakehouse platforms and negotiate existing vendor contracts before SAP's unified offering becomes available.