Concept card

SAP AI Core

As of 2026-07-24T14:00:00Z

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

What is SAP AI Core?

Generative AI Hub, not direct OpenAI/Anthropic integration, is the EU AI Act anchor — bypassing it breaks Art. 10 training-data governance because there's no SAP-side audit log of what data crossed which boundary.

What it is and why it matters

SAP AI Core is the BTP-managed runtime that hosts and serves AI models for the SAP ecosystem — both SAP-built models such as Joule and embedding services, and bring-your-own models, whether a fine-tuned open-weight model or a proxied call to a hyperscaler foundation model through Generative AI Hub. Without AI Core and its companion AI Launchpad console, every AI feature an SAP customer builds becomes a one-off integration with no shared governance, no cost attribution, and no audit trail. With them, inference becomes a platform capability that every team draws on rather than reinventing.

The four things it actually does

Inference serving exposes REST endpoints for chat, completion, embedding, and classification workloads. Training orchestration runs fine-tuning and retrieval-augmented-generation pipeline jobs against governed data. Generative AI Hub provides pre-built, SAP-managed proxy connections to OpenAI, Anthropic, AWS Bedrock, and Google Vertex, with SAP handling billing aggregation and audit logging on top of the underlying provider call. Model registry and versioning records every deployment with metadata that feeds directly into the EU AI Act Article 11 technical documentation dossier.

Why Generative AI Hub is the regulatory anchor

Why it matters

Direct-to-provider integration for SAP business data has no audit trail — any AI feature touching S/4, Datasphere, or SuccessFactors should route through GenAI Hub instead.
One resource group per business domain × environment is what makes cost attribution and audit trails align with governance committees — a single shared resource group makes both impossible.
AI Core's model registry and deployment metadata feed the EU AI Act Art. 11 technical dossier directly — the only path that delivers that documentation out of the box.

Key points

Runtime + cockpit for SAP-native AI features
GenAI Hub = EU AI Act Art. 10 anchor
Resource group per business domain × env
Pin model version (not 'latest')
Budget alerts in week 1 (5× runaway risk)
Audit log export feeds Art. 11 dossier
Staging → prod gate mandatory
Cost stable WoW = healthy operation
SAP AI Core is mastered only when it changes a named buyer decision.
Start with the semantic contract and control model before demonstrating the tool.

Terms used on this page

AI Core: BTP-managed runtime hosting + serving AI models
AI Launchpad: BTP UI for AI Core configuration + monitoring
Generative AI Hub: Proxy layer to external foundation models with SAP audit + billing
Resource group: Cost-attribution + governance scope inside AI Core
Compute unit: AI Core consumption metric
Serving template: Deployment manifest defining model + scaling + monitoring
Art. 11 dossier: EU AI Act technical documentation per system
Cost guard: Per-skill or per-tenant budget alert preventing runaway

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 AI Core pricing: what is public, what is not