Technology
November 18, 2026

SAP Databricks Integration: How the Partnership Reduces ETL and Makes SAP Data AI-Ready

The SAP and Databricks partnership eliminates custom ETL pipelines using zero-copy Delta Sharing to make enterprise SAP data AI-ready.

For many enterprises, SAP data integration follows a familiar pattern. A sales order lives in SAP S/4HANA. To analyze it alongside CRM, IoT, or market data, someone extracts it, rebuilds the business logic lost on the way out, loads it into a lakehouse, and keeps that pipeline healthy every night. Multiply that by hundreds of tables and several business units, and integration becomes one of the costliest layers of the data estate.

SAP Databricks integration is changing that pattern. Since early 2025, SAP and Databricks have moved from an announced partnership to production integrations that let teams work with governed SAP data in Databricks without building a new extraction pipeline for every use case. Here is what has shipped, how the architecture works, and where data engineering still matters.

Why SAP data integration has been so hard?

SAP data is rich in business meaning, but that meaning is hard to carry outside SAP. Table and field names are cryptic, business rules live in the application layer, and every extract forces teams to rebuild context by hand.

Databricks itself describes the traditional route as custom extraction pipelines that copy data, lose business context, and fragment governance. Every copy also creates a reconciliation problem: which version of "revenue" is the right one?

The SAP and Databricks partnership, from 2025 to today

SAP Databricks inside SAP Business Data Cloud

SAP Business Data Cloud (SAP BDC) is SAP's managed data platform. It brings together SAP Datasphere, SAP Analytics Cloud, and SAP Business Warehouse around curated, SAP-managed data products: ready-to-use datasets that keep SAP's business semantics.

SAP Databricks is a fully managed version of Databricks that runs as a service inside BDC and is sold by SAP. BDC data products are made available to it through Delta Sharing, so data scientists and engineers can start from SAP data that already carries its business context.

BDC Connect for existing Databricks environments

Many enterprises already run Databricks. BDC Connect for Databricks links an existing Unity Catalog–enabled workspace to SAP BDC for bidirectional, zero-copy sharing. According to SAP's reference architecture, this protects existing Databricks investments while giving teams curated SAP data without new ETL.

The practical choice: SAP Databricks suits teams starting fresh inside BDC; BDC Connect suits teams with an established Databricks platform.

The October 2026 expansion

At SAP TechEd 2026, the companies announced SAP Application Data Foundation, which makes SAP application data, from cloud and on-premises systems, available to BDC engines including SAP Databricks. They also outlined plans to bring Databricks Genie, Lakeflow, Unity Gateway, and Lakebase to SAP Databricks, along with a two-way Genie–Joule integration. SAP's own announcement also lists Agent Bricks.

No release dates were published for these additions, so treat them as roadmap items. SAP Databricks and BDC Connect are available today.

Zero-ETL vs zero-copy: related, but not the same

These terms are often used interchangeably. They should not be.

  • Traditional ETL extracts data, transforms it in a separate layer, and loads it into a target. ELT loads raw data first and transforms it inside the target. Both create a copy and a pipeline your team owns. ‍
  • Zero-ETL describes an outcome: you do not build or maintain pipeline code for the integration step. Some zero-ETL products still replicate data behind the scenes; the platform simply manages that work. ‍
  • Zero-copy describes a mechanism: the consumer reads data where the provider stores it, so no second persistent copy exists. ‍
  • Delta Sharing is the open protocol that makes zero-copy work between SAP BDC and Databricks. In the Databricks documentation, shared data stays in place and is governed and audited through Unity Catalog.

Two caveats matter. Zero-copy is not zero movement: data still travels over the network when it is queried, and cross-cloud reads can add egress cost. And neither approach removes transformation. Data still needs cleansing, modelling, and business rules; that work moves downstream rather than disappearing.

How the SAP Databricks integration architecture works?

  1. Source: SAP applications such as S/4HANA feed SAP BDC, where data is published as SAP-managed or custom data products. ‍
  2. Share: A BDC administrator shares a selected data product. BDC acts as the Delta Sharing provider; the connection is secured with mutual TLS and OpenID Connect. ‍
  3. Govern: In Databricks, the share is mounted as a catalog in Unity Catalog, which controls who can query it and records access for auditing. ‍
  4. Use: Engineers and analysts join SAP data with non-SAP sources using SQL, notebooks, and machine learning. ‍
  5. Return: Results, such as a demand forecast, can be shared back to BDC for use in SAP Datasphere or SAP Analytics Cloud.

Business benefits of SAP Databricks integration

  • Fewer pipelines to build and run for data that is already available as a data product. ‍
  • Less duplication, which means fewer conflicting versions of the same metric. ‍
  • SAP business context preserved, so teams spend less time reverse-engineering tables. ‍
  • Governance in one place, with access controls and audit trails in Unity Catalog. ‍
  • Faster time to analysis, because data scientists start from curated data. ‍
  • Better AI readiness, since models and agents can be grounded in trusted SAP data.

Actual savings depend on your landscape, data volumes, and how many existing pipelines a shared data product can replace.

Industry use cases

  • Manufacturing: Combine years of SAP order history with third-party signals to build demand forecasts, the scenario Databricks highlighted at TechEd. ‍
  • Retail: Blend SAP sales and inventory with e-commerce and loyalty data for customer analytics and replenishment planning. ‍
  • Insurance: Pair SAP finance data with policy and claims data for financial planning and operational reporting. ‍
  • Financial services: Use receivables data with credit and payment-behavior data to predict late payments. ‍
  • Supply chain: Join SAP procurement data with CRM and supplier data for supplier performance analytics.

Implementation challenges to plan for

  • Legacy SAP landscapes: ECC and heavily customized systems need a clear path. Check what Application Data Foundation will cover for your release. ‍
  • Coverage gaps: Custom objects and industry extensions may need custom data products or conventional pipelines. ‍
  • Data quality: Zero-copy sharing does not improve data quality. Issues in the source data products reach Databricks unchanged. ‍
  • Governance and lineage: You now manage access across the BDC catalog and Unity Catalog. Define ownership, approvals, and lineage across both. ‍
  • Performance and cost: Querying in place shifts cost to compute. High-traffic dashboards may still need precomputed tables. ‍
  • Contracts and compliance: Databricks' documentation notes it may share certain usage information about BDC-shared workloads with SAP. Legal and compliance teams should review this. ‍
  • Coexistence: Existing extraction pipelines will run alongside zero-copy sharing for some time. Plan a phased migration. ‍
  • Skills: SAP and data platform teams often work separately. This model needs both.

Where enterprise data architecture is heading?

Available now: zero-copy SAP data sharing and unified governance through Unity Catalog. ‍

Announced: SAP Application Data Foundation, more Databricks capabilities inside SAP Databricks, and Genie–Joule integration. SAP also agreed to acquire Dremio in May 2026 to make BDC natively support Apache Iceberg. The Register reports SAP will keep supporting Delta Lake.

Likely direction (our view): AI agents that act on shared business semantics, more automated data engineering, near-real-time operational intelligence, and open table formats that let several engines work on one copy of data. The common thread is a governed, semantic data foundation.

How Hexaview Technologies can help?

Zero-ETL simplifies data access, but it does not design your data platform. Hexaview Technologies helps enterprises assess SAP and Databricks integration options, design Databricks Lakehouse architecture, modernize and automate ETL/ELT pipelines, and put data quality and governance in place so SAP data is ready for analytics and AI.

Planning your SAP data strategy? Talk to Hexaview's data engineering team about where zero-copy sharing fits, which pipelines to retire, and how to make your SAP data AI-ready.

Sources:  

https://news.sap.com/2025/02/sap-business-data-cloud-databricks-turbocharge-business-ai/

https://www.databricks.com/company/newsroom/press-releases/databricks-announces-launch-sap-databricks

https://www.databricks.com/blog/announcing-general-availability-sap-databricks-sap-business-data-cloud

https://www.databricks.com/blog/announcing-general-availability-sap-business-data-cloud-connect-databricks

https://www.databricks.com/blog/databricks-and-sap-unleash-end-end-data-ai-capabilities-level-enterprise-ai-transformation

https://news.sap.com/2026/10/sap-teched-ai-developers-build-what-matters-autonomous-execution/

Frequently Asked Questions

What is SAP Databricks?

SAP Databricks is a fully managed version of Databricks that runs as a service inside SAP Business Data Cloud and is sold by SAP. It gives data teams governed access to SAP data products through Delta Sharing.

What is the difference between SAP Databricks and BDC Connect?

SAP Databricks runs inside SAP BDC. BDC Connect links SAP BDC to a Databricks workspace you already run, for bidirectional zero-copy sharing.

Is zero-ETL the same as zero-copy?

No. Zero-ETL means you do not build or maintain integration pipelines, though the platform may still replicate data behind the scenes. Zero-copy means data is read in place, with no second persistent copy.

How does Delta Sharing work with SAP Business Data Cloud?

SAP BDC acts as the Delta Sharing provider. Each shared data product is mounted as a catalog in Databricks Unity Catalog, which controls and audits access.

Does SAP Databricks integration eliminate data engineering?

No. It reduces extraction and pipeline maintenance, but data modeling, quality checks, governance, and use-case transformations remain.

Can it work with SAP ECC or on-premises SAP?

SAP announced SAP Application Data Foundation in October 2026 to make cloud and on-premises SAP data available to SAP Databricks. Check availability for your release with SAP.

About Author

Shubham Rai
Shubham Rai is an Application Engineer at Hexaview Technologies and a Microsoft Fabric Community Super User specializing in modern data engineering, Power BI, SQL, and PySpark. He holds four Microsoft certifications and one Databricks certification, reflecting his deep expertise in building scalable, insights-driven data solutions. Passionate about community knowledge sharing, he actively helps organizations transform complex data into meaningful business insights.
View Profile

Blogs you may like

Cookie Preferences