SAP BTP for Analytics: Complete Architecture Guide
SAP Business Technology Platform is a comprehensive, multi-cloud platform-as-a-service layer designed to extend SAP enterprise applications, run custom development workloads, host AI and machine learning models, and orchestrate complex integration scenarios. This guide maps the BTP services landscape, establishes architectural boundaries, and clarifies where BTP genuinely delivers value for analytics teams.
The BTP Services Landscape
November 2026 update: Since SAP Sapphire 2026, SAP describes BTP capabilities as a core part of the new SAP Business AI Platform, which brings SAP BTP, SAP Business Data Cloud, and SAP Business AI together. BTP still spans services across data, AI, integration, extension, runtime, and platform administration. For analytics teams, the most relevant services cluster around data management (SAP HANA Cloud; Datasphere, licensed through SAP Business Data Cloud for new contracts since January 2026), AI (SAP AI Core with its generative AI hub, Joule Studio), integration (Integration Suite, Event Mesh, Open Connectors with more than 160 prebuilt connectors), and extension (SAP Build, SAP Build Process Automation).
SAP BTP is organized into six major capability domains. Understanding the services within each domain—and their interconnection points—is essential for designing platform strategies that scale, integrate cleanly, and avoid over-engineering.
SAP BTP Services Landscape: Six integrated capability domains supporting runtime, AI, data, integration, extension, and DevOps workloads around SAP Analytics Cloud and Datasphere.
Core Services Deep Dive
Runtime Services: Cloud Foundry & Kyma
Cloud Foundry is an application platform for deploying microservices, RESTful APIs, and event-driven workloads. It abstracts infrastructure, scales application instances, and binds services to applications. Cloud Foundry excels when you need custom business logic that supplements analytics workflows—approval chains, notification systems, or data transformation jobs.
Kyma is SAP's managed Kubernetes runtime, based on the open-source Kyma project. Teams work with Kubernetes directly and add modules such as Serverless, Eventing, and API Gateway (Istio comes as a default module)—useful for complex data pipeline orchestrations, containerized ML inference servers, or headless analytics jobs requiring precise resource allocation.
Destination Service manages credentials and connectivity to on-premise systems, cloud databases, and third-party APIs. It centralizes authentication, endpoint management, and certificate handling. For analytics, Destinations enable secure connectivity from BTP runtimes to SAP Analytics Cloud, Datasphere, HANA Cloud, and external data warehouses without hardcoding credentials.
AI & Machine Learning: AI Core, the generative AI hub, Joule Studio
SAP AI Core runs AI assets on BTP: it executes training and batch-inference pipelines as jobs, serves trained models as inference endpoints, and tracks executions, metrics, and artifacts through one API. You bring your own code and containers through your Git repository and Docker registry, or use models from its model library, such as SAP-RPT-1.6 for tabular predictions. SAC does not call AI Core on its own: an application writes predictions to a model or table that SAC reads.
The generative AI hub in SAP AI Core gives governed access to foundation models from OpenAI, Anthropic, Google, Mistral AI, Amazon, Meta, Cohere, Perplexity, and SAP. Its orchestration service offers a harmonized API (switch models without changing client code), prompt templates and prompt lifecycle management, grounding on your own documents, data masking, and content filtering. Analytics teams use it to add conversational interfaces, generated explanations, and semantic search to custom apps without building integrations to several model providers.
Joule Studio is SAP's environment for building custom AI agents, with low-code and pro-code options and more than 70 AI models. Its first edition in SAP Build shipped the skill builder (GA July 2025) and the agent builder (GA December 2025); general availability of the new Joule Studio, launched at SAP Sapphire 2026, has been rolling out in phases since October 2026. SAP also ships prebuilt agents: at Sapphire 2026 it described more than 50 domain-specific Joule Assistants orchestrating over 200 specialized agents. In analytics, agents handle investigation (a user asks a question, the agent fetches data, runs analysis, and returns an insight) without requiring dashboard design.
SAP AI Launchpad is the management UI: a multitenant SaaS application for managing AI scenarios, executions, deployments, and their metrics across AI runtimes such as SAP AI Core, including the generative AI hub. Use it to govern analytics-adjacent ML models in production, such as propensity scores, next-best-action engines, and demand forecasts; accuracy and drift checks are metrics your own pipelines compute and log.
Data Services: Datasphere, HANA Cloud, Data Integration
Datasphere is SAP's business data fabric and a core component of SAP Business Data Cloud. It connects to SAP and non-SAP sources, builds a semantic layer (business semantics, relationships, transformations), and shares data products inside SAP Business Data Cloud. From BTP, you access Datasphere through its OData consumption APIs—building custom discovery UIs, lineage views, or quality monitoring dashboards. Datasphere handles the "what data exists and where" question; BTP handles the "how do teams consume it" question.
HANA Cloud is an in-memory multi-model database. It supports relational queries (SQL), graph analytics (finding entities and relationships), spatial queries (geo-based analysis), JSON document storage, and vector search, and since 2026 it is also available natively inside SAP Business Data Cloud. For analytics on BTP, HANA Cloud is used as the analytical cache or mart—loading subsets of Datasphere data, running complex joins, and serving results quickly to downstream applications.
Data Integration within Integration Suite handles connectivity to source systems. Cloud Integration (formerly Cloud Platform Integration, CPI) runs integration flows that move and transform payloads and handle errors. Event Mesh delivers business events to subscribers, which write them to SAP HANA Cloud or Datasphere tables for near-real-time analytics and event-driven alerting.
Integration Suite: Cloud Integration, API Management, Event Mesh
Cloud Integration (formerly CPI) is the workhorse for system-to-system data flows. It handles SAP ERP, SAP SuccessFactors, cloud connectors, REST/SOAP endpoints, and file-based integrations. For analytics: Cloud Integration extracts data from transaction systems, applies business rules, normalizes payloads, and lands clean data into Datasphere or HANA Cloud.
API Management exposes analytics data and models as managed APIs. You manage API revisions, apply rate limiting, track usage, and monetize API products, publishing them in the Developer Hub. Analytics teams publish APIs from SAP Build, custom runtimes, or AI Core inference endpoints—allowing business users and external partners to consume insights programmatically.
Event Mesh enables asynchronous, publish-subscribe messaging: the Event Mesh capability of Integration Suite covers low-to-moderate volumes, and advanced event mesh covers high-volume, distributed scenarios. Rather than polling for updates, subscribers receive business events (orders created, invoices posted, inventory movements) as they happen. It decouples systems, reduces latency, and enables event-driven analytics refresh once events land in a store that your analytics reads.
Extension Services: SAP Build, Build Process Automation, Build Work Zone
SAP Build is SAP's unified low-code and pro-code offering. With SAP Build Apps you assemble web and mobile apps from prebuilt components and bind them to OData and REST services through BTP destinations; SAP Build Code and ABAP Cloud cover pro-code work. Build custom data-entry forms, approval interfaces, KPI portals, and discovery applications with little or no code.
Build Process Automation orchestrates business processes using visual workflows. Define steps, conditions, assignments, and notifications. Use it to automate data quality checks (reject incomplete submissions, notify owners), approval chains (executive sign-off before publishing analytics), and data handoff workflows between teams.
Build Work Zone provides a branded portal for end users. Embed SAP Analytics Cloud dashboards, apps built with SAP Build, and collaboration tools. Work Zone handles authentication, role-based access, and single-sign-on to a unified analytics experience.
DevOps & Governance
CI/CD Pipelines automate testing, building, and deploying custom code. Integrate with GitHub, GitLab, or Bitbucket. SAP BTP provides the SAP Continuous Integration and Delivery service, whose pipelines trigger on code commit, run tests, and deploy to dev/test/prod environments. For analytics: automate deployment of custom services, APIs, and integration flows.
SAP Cloud Transport Management orchestrates multi-environment promotion. Move custom code, configuration changes, integration content, and SAP Analytics Cloud content packages from dev to test to production along defined transport routes, with scheduled imports into each target node.
Logging, Monitoring, and Audit: SAP Cloud Logging collects logs, metrics, and traces from applications in different BTP environments, SAP Alert Notification routes alerts, the SAP Audit Log service records audit events such as access to personal data, and SAP Cloud ALM adds exception monitoring. Build dashboards that track API latency, integration errors, model inference costs, and data flow volumes, and keep audit trails for compliance (GDPR, SOX, industry regulations).
Architectural Patterns & Use Cases
When BTP Genuinely Delivers Value
Use BTP when analytics consumption requires orchestration—multi-step approval chains, conditional data delivery, or business rule application beyond what dashboards alone provide. Example: A financial planning process requires executive approval before publishing new budget forecasts. Build a custom approval workflow in Build Process Automation, integrate with SAP Analytics Cloud, and control publication access programmatically.
Extend SAP Analytics Cloud's built-in predictions with proprietary models. Deploy a custom demand forecasting model to AI Core, have an application call it and write the forecasts where SAC reads them, and manage executions and metrics in AI Launchpad. For natural-language questions on SAC models, use just ask and Joule's analytical insights; use the generative AI hub when a custom app needs its own conversational interface.
When batch analytics refresh is too slow, use Event Mesh to stream change events from source systems. A customer places an order; a subscriber writes the event to SAP HANA Cloud or a Datasphere table; dashboards on a live connection show it on their next query. Operations get current inventory, sales teams see live pipeline, and executives see rolling KPIs—not stale end-of-day snapshots.
Build a single source of truth when your data lives in SAP and non-SAP systems. Use Integration Suite (Cloud Integration) to connect ERP, cloud applications, and third-party systems. Land normalized data in Datasphere, publish a semantic layer, and serve analytics from a single trusted version.
When SAP Analytics Cloud alone doesn't meet UX requirements, build a custom portal with SAP Build and Build Work Zone. Embed dashboards, add custom data-entry interfaces, apply branding, and provide role-based personalization. Control what each user sees without relying solely on SAP Analytics Cloud's built-in sharing model.
Joule Studio, whose new edition has been rolling out in phases since October 2026, enables copilot-style analytics. Users ask questions in natural language; agents fetch data, run analysis, and respond. Agents can use different foundation models for different tasks (Joule Studio offers more than 70), and custom apps reach models through the generative AI hub. Build domain-specific agents for procurement, supply chain, and finance that investigate questions and suggest actions, with human approval for sensitive steps.
When Teams Overreach
BTP becomes expensive and slow when teams treat it as a substitute for solid analytics fundamentals:
- Weak data modeling: Building custom APIs and orchestration to work around poor semantic design. Fix the model; don't build platform code to hide it.
- Presentation logic in code: Moving dashboard formatting, calculations, or drill-down logic into microservices. The dashboard tool should handle presentation; BTP should handle business logic only.
- Over-governance: Creating custom approval and audit workflows when role-based access control would suffice. Every approval step adds latency and operational overhead.
- Duplicate data flows: Building custom integrations when Integration Suite covers the use case. Consolidate; don't replicate.
- Bespoke infrastructure for commodity problems: Running Kubernetes clusters to host batch jobs when Cloud Foundry's autoscaling would work. Choose the simplest tool that solves the problem.
Pricing & Operational Considerations
Consumption-Based Pricing (BTPEA and Pay-As-You-Go)
Under the consumption-based model (BTPEA, CPEA for existing contracts, or pay-as-you-go), each service is metered by its own metric, listed in the SAP Discovery Center. Typical cost drivers:
- Compute: Cloud Foundry runtime is billed by the runtime memory actually consumed; Kyma and other runtimes have their own metrics.
- Data: SAP HANA Cloud and other storage services are metered by their own capacity metrics; Datasphere and SAC are sized in SAP Business Data Cloud capacity units for new contracts, not in BTP credits.
- AI/ML: SAP AI Core usage for training and inference, and generative AI hub consumption, which is priced in capacity units.
- Integration: Integration Suite editions and add-ons, with advanced event mesh sized in broker tiers (for example by client connections).
- Commitment: BTPEA credits are prepaid with an annual commitment and overages are billed at list price; pay-as-you-go is billed monthly in arrears.
This model rewards efficient architecture. Oversized deployments, inefficient queries, and redundant data copies drive costs quickly. Monitor consumption in the BTP cockpit (global account and subaccount usage views) and review costs at least quarterly.
Operational Maturity Required
BTP's power comes with operational responsibility:
- Security patches: BTP patches infrastructure automatically, but you manage code, dependencies, and configurations.
- Monitoring & alerting: Set up dashboards to track API latency, integration failures, and data quality. BTP provides the tools; you define what to monitor.
- High availability: Multi-instance deployment, database replication, and disaster recovery require planning and automation.
- Team skills: Cloud Foundry requires DevOps competency. Kyma requires Kubernetes knowledge. SAP Build is low-code, but custom services require developers.
Start with a clear operational model. Define SLAs, escalation paths, and on-call rotations before deploying production workloads.
Architecture Decision Framework
Before committing to BTP, ask these questions in order:
1. What Capability Gap Exists?
Is the problem a missing platform capability (extension, orchestration, AI, real-time data)? Or is it weak fundamentals (poor data model, unclear ownership, inconsistent governance)?
If fundamentals are weak: Fix them first. A stronger semantic model and clearer data governance often eliminate the perceived need for platform build-out.
If the gap is genuine: Proceed to question 2.
2. Can SAP's Native Tools Solve It?
Before building custom, check if SAP Analytics Cloud, Datasphere, Integration Suite, or SAP Build already address the need. "Nearly fits" is not the same as "fits."
If a native tool solves 80% of the problem: Use it. Accept minor gaps rather than building custom infrastructure.
If native tools fall materially short: Proceed to question 3.
3. Which BTP Domain & Service?
Match the use case to the right service:
- Custom business logic: Cloud Foundry microservice or Build Process Automation workflow.
- Custom ML model: AI Core + AI Launchpad.
- Generative AI: the generative AI hub + Joule Studio (agents).
- System integration: Integration Suite (Cloud Integration, Event Mesh).
- Custom UI: SAP Build + Build Work Zone.
- Real-time data: Event Mesh + HANA Cloud.
- Container orchestration: Kyma (if you need Kubernetes-native features) or Cloud Foundry (if you need simplicity).
4. What Is the Operational Burden?
Each choice has operational cost. Cloud Foundry is mostly managed; Kyma requires Kubernetes ops. Integration Suite scales automatically; custom services require monitoring and alerting setup. The generative AI hub is managed; Joule Studio development requires training.
Estimate staffing, skills, and tooling required. If operational cost is high relative to business impact, reconsider or simplify the design.
5. What Is the Cost Profile?
Get a quote from SAP. Provision a sandbox environment and measure actual consumption for a week. Real numbers are better than estimates. Factor in development time, operations staff, and infrastructure costs into the business case.
Key Takeaways
- SAP BTP is a comprehensive platform, not a single tool. Its power lies in the integration of six major capability domains—runtime, AI/ML, data, integration, extension, and DevOps—into a cohesive, multi-cloud infrastructure.
- Cloud Foundry and Kyma provide runtime flexibility. Cloud Foundry suits microservices and stateless workloads. Kyma suits complex orchestrations and Kubernetes-native deployments.
- AI Core, the generative AI hub, and Joule Studio enable AI-driven analytics. Deploy proprietary models, access foundation models from providers such as OpenAI, Anthropic, and Google, and build custom AI agents without building infrastructure.
- Datasphere, HANA Cloud, and Integration Suite form the data foundation. Datasphere governs semantic data products, HANA Cloud serves real-time analytics, and Integration Suite orchestrates complex multi-system flows.
- SAP Build, Build Process Automation, and Build Work Zone provide low-code extension. Create custom UIs, automate workflows, and deliver branded analytics portals with little or no code.
- BTP delivers genuine value for extension, orchestration, AI, and real-time analytics. It becomes expensive and risky when used to paper over weak fundamentals—poor data modeling, unclear ownership, or inconsistent governance.
- Architecture discipline matters more than platform enthusiasm. Ask whether the requirement is genuine platform work (extension, orchestration, AI, integration) or a signal that fundamentals need fixing first.
- Operational maturity is required. BTP's flexibility comes with operational responsibility. Plan for monitoring, high availability, security patching, and team skills before deploying production workloads.
- Start small and measure. Provision a sandbox, build a pilot use case, measure actual costs and operational burden, and then decide whether to scale.
Next Steps for Your Organization
For Architects
Map your current analytics landscape. Identify capability gaps that genuinely require platform investment versus gaps that signal weak fundamentals. Build a BTP pilot for one use case—orchestration, AI model deployment, or real-time integration. Measure cost, operational overhead, and business impact. Use results to justify broader investment or to guide refocusing on fundamentals.
For Analytics Teams
Learn SAP Analytics Cloud and Datasphere deeply before assuming BTP is needed. Understand role-based access, semantic modeling, and data product design. When a "we need BTP" request arrives, ask: Is this a genuine extension need? Or is it a sign we need better data governance? Push back respectfully; deep analytics fundamentals often eliminate the need for platform build-out.
For Enterprise Leaders
BTP enables competitive advantage when used to solve genuine extension, orchestration, or AI needs. It becomes a cost center when it substitutes for weak fundamentals or duplicates tool capabilities. Require a clear business case before BTP investment. Measure pilot outcomes. Scale based on evidence, not enthusiasm.
Trust and Provenance Note
This guide is original architectural guidance and synthesis from publicly available SAP documentation, analyst reports, and practitioner experience. Service claims—particularly around Joule Studio availability, pricing models, and LLM access—should be validated against current official SAP references before use in production decisions. Last updated November 2026.