Concept card

BDC Connect for Snowflake (+ SAP Snowflake solution extension)

As of 2026-10-10

Edited by Cédric Mary, sole editor and publication director — 27 years in the SAP ecosystem. How these pages are produced, and where automation is used.

What is BDC Connect for Snowflake (+ SAP Snowflake solution extension)?

Unlike Databricks' OEM-inside-BDC model, Snowflake stays a partner-connected architecture — dual commercial relationship, Snowflake-credit pricing, Cortex Analyst/Agents on Snowflake's side — meaning Tier-1 customers increasingly run both platforms side by side rather than choosing one.

What it is

SAP and Snowflake announced two parallel paths on 2025-11-04: SAP Snowflake (a solution extension where SAP packages Snowflake) and SAP BDC Connect for Snowflake (the bidirectional zero-copy channel). Both reached GA within the H1 2026 target window: the BDC Connect zero-copy connector went GA on 2026-05-04 (per Snowflake's own release notes), and SAP confirmed GA for the SAP Snowflake solution extension around the same time, post-Sapphire 2026. Unlike Databricks (C166, OEM-inside-BDC), Snowflake is the canonical 'partner-connected' model — the buyer's commercial relationship is dual (SAP + Snowflake), pricing remains in Snowflake credits (C233), and Cortex Analyst (C175) + Cortex Agents (C176) sit on the Snowflake side of the boundary. The trade-off: Snowflake brings best-in-class warehouse separation of storage/compute that Databricks doesn't match on pure-BI workloads; Databricks brings deeper Joule integration via the OEM model. Tier-1 customers in 2026 will run BOTH — trilateral architecture (C245).

The distinction from C169 matters at the scoping table: BDC Connect for Snowflake is the technical channel — the zero-copy, bidirectional path that moves governed data between Datasphere and Snowflake without a physical export/import step — while SAP Snowflake (the solution extension) is a separate commercial vehicle for buying Snowflake capacity through SAP's own invoicing and support. A client can adopt the connector without ever buying Snowflake through SAP, and conversely a client who buys Snowflake through SAP still configures the same connector for the technical integration; the two decisions are independent and a consultant should never let a client conflate them into one purchase conversation.

Why it matters

The commercial-independence point matters specifically to CFOs: Snowflake's credit-based billing stays separate from SAP, avoiding deeper single-vendor lock-in.
The named trade-off is concrete and testable: Snowflake wins on storage/compute separation for pure-BI, high-concurrency workloads; Databricks wins on native Joule integration via its OEM status.
Both SAP Snowflake (solution extension) and BDC Connect for Snowflake (zero-copy channel) reached GA within H1 2026 — the connector on 2026-05-04 per Snowflake's release notes — two related but distinct offerings to keep straight when scoping.

Key points

SAP and Snowflake announced two parallel paths on 2025-11-04: SAP Snowflake (a solution extension) and BDC Connect for Snowflake (the bidirectional zero-copy channel).
Unlike Databricks' OEM-inside-BDC model, Snowflake stays partner-connected: dual commercial relationship, Snowflake-credit pricing, Cortex Analyst and Agents on Snowflake's side.
The consequence for large customers is coexistence — both platforms side by side — rather than a single-platform choice.
Check the current GA status of each path in the release notes before committing a delivery date; the two paths did not ship on the same day.
The connector is a data bridge, not a model bridge — Cortex Analyst and Cortex Agents run entirely on the Snowflake side, never through SAP's generative AI hub.
SAP's Data Access Controls on Datasphere remain the single row/column security rule set, enforced identically whether the query originates from Snowflake, Databricks or SAC.
In a trilateral estate (SAP + Snowflake + Databricks), route by workload shape, not by vendor preference: process-aware questions to Joule, deterministic declared metrics to Cortex Analyst, ML-training-shaped work to Databricks.
Publish an explicit register of which system owns which metric's definition before running three AI surfaces over the same business — the connectors make coexistence technically possible, they do not resolve semantic conflicts.

Terms used on this page

Partner-connected model: SAP's commercial pattern for Snowflake, distinct from Databricks' OEM-inside-BDC: the customer holds two separate commercial relationships (SAP and Snowflake) rather than one bundled contract.
Trilateral architecture: A Tier-1 pattern (C245) where SAP, Snowflake and Databricks all run side by side against the same governed Datasphere data products, each serving the workload shape it fits best.
Cortex Analyst: Snowflake's native text-to-SQL capability (C175), governed by a formally declared semantic model — the deterministic NL-to-SQL surface on the Snowflake side of a BDC Connect for Snowflake deployment.
Cortex Agents: Snowflake's multi-step agentic orchestration layer (C176) that chains Cortex Search, Cortex Analyst and SQL/Python tasks — the agentic surface on the Snowflake side, entirely separate from Joule Agents.
Data Access Control (Datasphere): Datasphere's catalog-level row and column security rule set, enforced identically on a shared data product regardless of whether the reading platform is Snowflake, Databricks or SAP's own SAC.
Metric-ownership register: An explicit, published document naming which system's semantic layer is authoritative for each business metric, kept current at the same cadence the metric definition changes — the practical fix for cross-platform number disagreement.

Sources

Full card available to members. What the full card adds: the full decision framework · the SAP vs Snowflake / Databricks / Fabric comparison · the common pitfalls and their fix · the cheat sheet · the architecture schemas · the code blocks · the facts worth quoting.

Open in the app →

Guides that answer with this page

These guides cite this page as one of the sources their answer rests on.

SAP Datasphere vs Snowflake: the real decision