On May 4, 2026, SAP said it had agreed to acquire Dremio, the lakehouse company it wants to use to unify SAP and non-SAP data for agentic AI. Three days later, on Thursday, May 7, this is still worth covering because the story is no longer just a deal headline. It is a clearer signal about where enterprise AI is getting stuck: not on model quality alone, but on fragmented data, missing business context, and the cost of making agents work across real systems.
SAP’s announcement matters beyond one acquisition because it reframes the enterprise AI stack. Instead of treating agents as a front-end assistant problem, SAP is moving deeper into the data layer, where access, lineage, open formats, and structured business records decide whether an AI system can actually do useful work at scale.
What SAP agreed to buy on May 4
SAP said Dremio will complement SAP Business Data Cloud and SAP HANA Cloud, with the goal of making Business Data Cloud an Apache Iceberg-native enterprise lakehouse. The company said that would let SAP and non-SAP data live on the same open foundation, with federated access across enterprise sources and without forcing format conversion or large-scale data movement first.
That is the practical center of the deal. SAP is not pitching Dremio as a generic analytics add-on. It is pitching Dremio as infrastructure for AI-ready context: a way to make business data easier for agents, analytics systems, and operational AI workloads to use in real time.
SAP also said the combined stack would use an open catalog built on Apache Polaris and the Apache Iceberg REST Catalog API, creating a single semantic and discovery layer for meaning, relationships, lineage, and access rights. The company tied that directly to its SAP Knowledge Graph, which it says will encode business relationships, hierarchies, regulatory classifications, and cross-system lineage as native properties.
The transaction terms were not disclosed. SAP said the Dremio deal is pending regulatory approval and is expected to close in the third quarter of 2026.
Why this still matters for enterprise AI teams now
The most important line in SAP’s announcement was not about Dremio itself. It was SAP’s claim that enterprise AI projects often fail because the data underneath them is fragmented, trapped in proprietary formats, or stripped of the business context needed to explain and trust an AI-driven decision.
That argument lines up with a broader market shift. Enterprise buyers are getting past the phase where adding another model or chatbot looks like progress on its own. The harder problem is building a data layer that agents can query safely, reason over consistently, and use without forcing every workflow through another brittle ETL project.
In that sense, Dremio gives SAP three things at once:
- An open-format bet: SAP is making Apache Iceberg more central to its AI story, which matters because open table formats are becoming the connective tissue for analytics, machine learning, and agent systems.
- A lower-friction data path: The pitch is that SAP customers should be able to work across SAP and non-SAP systems with less copying, reformatting, and pipeline sprawl.
- A stronger agent context layer: If lineage, semantics, and permissions sit closer to the data, AI systems have a better chance of doing work that is auditable instead of merely impressive in demo form.
For Nerova’s audience, that is the durable search-value angle. The interesting question is not whether SAP bought a lakehouse vendor. It is whether enterprise AI is becoming a contest over who can provide the cleanest governed context for agents to act, not just answer.
Why the Prior Labs deal makes the Dremio story bigger
SAP made a second AI move on the same day, announcing an agreement to acquire Prior Labs. That deal is easy to miss if you only followed the Dremio headlines, but it helps explain SAP’s broader strategy.
Prior Labs works on tabular foundation models, or TFMs, which are designed for structured business data such as payment delays, supplier risk, churn, and other prediction-heavy use cases where large language models are often weak. SAP said Prior Labs will continue to operate independently and that it plans to invest more than €1 billion over four years to scale it into a frontier AI lab focused on structured enterprise data.
The combination is what makes the May 4 news more important than a normal M&A item. Dremio addresses the open data foundation and data access problem. Prior Labs addresses the modeling problem for structured records and prediction-heavy business workflows. Put together, SAP is signaling that its agent story will rely less on generic LLM magic and more on a stack built around enterprise data reality: tables, permissions, lineage, forecasts, and operational context.
That is also why SAP linked Prior Labs back to SAP AI Core, Business Data Cloud, and Joule. The company is trying to connect structured-data prediction and agent execution into one operating model instead of leaving them as separate AI silos.
What changed in SAP’s competitive position
Outside analysts also highlighted why the Dremio move matters. CIO noted that Dremio is younger and less established than Snowflake or Databricks, but said the acquisition could still fill a missing piece for SAP by giving it a more direct open data fabric for AI agents. The Register made a similar point from the infrastructure side, noting that SAP had previously leaned on Databricks for parts of its external-data story and is now pushing much harder into Iceberg-native foundations of its own.
That does not automatically make SAP the winner here. The obvious questions are still unresolved: how quickly the integration lands, whether SAP can keep the openness it is promising, and how mature the combined platform will feel compared with existing cloud data leaders.
But the strategic direction is clearer now. SAP does not want to be only the system of record with an AI assistant attached. It wants to own more of the governed data substrate that future enterprise agents will depend on.
What enterprise teams should watch next
- Integration timing: The Dremio transaction is expected to close in Q3 2026, while the Prior Labs deal is expected in Q2 or Q3 2026. Product detail will matter more than deal language once roadmaps get specific.
- How open the stack stays: SAP is emphasizing Apache Iceberg, Apache Polaris, and open APIs. Buyers should watch whether that openness remains real in day-to-day deployment.
- Partner fallout: SAP already works with major data-platform vendors. The Dremio move could reshape how those partnerships matter for customers building AI and analytics on mixed estates.
- Structured-data AI inside agent workflows: If Prior Labs technology gets productized well, SAP could have a stronger answer for prediction-heavy workflows where plain LLMs still struggle.
The practical takeaway is simple. SAP’s Dremio announcement from May 4 is still relevant on May 7 because it captures a broader enterprise AI shift that searchers are still trying to understand: the next bottleneck is not just model access. It is whether your agents can reach the right data, understand what it means, and act on it without creating another integration mess.