SAP’s AI Pivot: A Technical Deep Dive into the Post-Migration Architecture

Executive Summary

• Technical Summary: SAP is transitioning its strategic focus from the "RISE with SAP" ERP migration framework toward a decoupled, AI-first architecture centered on the Joule platform and Business Data Cloud, introducing consumption-based commercial models to offset a €2 billion shortfall in cloud transition targets.
• Failure of the "Rip-and-Replace" Model: As of 2025, on-premise support revenue stood at €10.5 billion—€2 billion higher than the €8.5 billion target—indicating that over 10,000 major organizations remain on legacy ECC (Enterprise Core Component) systems.
• AI Decentralization: SAP is softening its "cloud-only" innovation stance, preparing to port its flagship AI, Joule, to customers on SAP ERP Private Edition subscriptions, regardless of whether they have completed a full S/4HANA transformation.
• New Commercial Logic: The company is moving away from per-employee seat licensing toward consumption-based metrics to protect revenue as AI-driven automation reduces the necessary human headcount for enterprise tasks.

Technical Architecture: From ERP Core to AI Orchestration

Under the previous "RISE with SAP" strategy, the technical roadmap was a linear "lift and shift" from legacy ECC environments to S/4HANA in the cloud. The new pivot, led by the newly formed Customer Value Group, reflects a more modular, "innovation-led" architecture.

1. The Joule AI Layer

The centerpiece of this shift is Joule, SAP's generative AI assistant. Architecturally, Joule is being positioned as a cross-portfolio orchestration layer. Initially restricted to S/4HANA Public Cloud, SAP is now extending Joule’s reach:

• Integration with Private Cloud: Support is expected later in 2026 for customers on SAP Cloud ERP Private Edition (formerly RISE with SAP S/4HANA Cloud Private Edition).
• Decoupled Innovation: By providing AI capabilities to ECC-in-cloud environments, SAP is decoupling the "Innovation" layer from the "Core ERP" layer. This allows customers to leverage LLM-based automation without a full "brownfield" or "greenfield" ERP migration.

2. Business Data Cloud (Project Atlas)

To feed its AI models, SAP is pushing the Business Data Cloud. This component serves as a semantic data layer that bridges SAP’s structured data with external data lakes.

• Partnerships: Technical integrations with Databricks and Palantir are being leveraged to provide a unified data fabric.
• Functionality: It allows for the training and fine-tuning of domain-specific models using enterprise-specific telemetry, financial data, and supply chain metrics without moving the underlying ERP data out of its sovereign environment.

3. The Shift in Commercial Engineering

SAP is establishing a dedicated unit of hundreds of engineers and sales specialists to implement a Consumption-Based Model.

• The Logic: Legacy licensing based on "Named Users" or "Per Employee" becomes a liability in an AI-driven ecosystem where productivity gains (targeted at 30%) result in fewer human users.
• Technical Metric: While the exact technical unit (e.g., tokens, API calls, or "Business Units of Value") is not yet fully disclosed, the architecture is shifting toward a metered gatekeeper model for AI inference and data processing.

Performance Analysis: The "Cloud Escape" Reality Check

The pivot is driven by significant friction in the planned migration from legacy systems to the cloud. The following table illustrates the gap between SAP’s 2020 projections and the 2025/2026 reality.

Migration Timeline

• 2027: End of mainstream support for ECC.
• 2030: End of extended support (available at a 2% premium).
• 2033: Final support cut-off for customers with a documented migration plan.

Technical Implications for the Ecosystem

For Enterprise Architects

The "rip-and-replace" era is effectively over. Architects should now focus on Brownfield Migrations—moving ECC environments into a private cloud subscription (SAP Cloud ERP Private Edition) to gain access to AI services like Joule and the Business Data Cloud, rather than attempting a full business process transformation.

For Developers and ML Engineers

The focus is shifting from ABAP-based customization within the ERP core to Side-by-Side Extensibility via the SAP Business Technology Platform (BTP).

• AI Integration: Developers will increasingly interact with AI via "bite-sized" innovation chunks rather than monolithic updates.
• Data Strategy: The reliance on the Business Data Cloud means ML engineers must focus on data federation between SAP and platforms like Databricks, rather than traditional ETL (Extract, Transform, Load) processes.

Limitations and Trade-offs

• Fragmentation Risk: By allowing AI innovation on older core versions (via Private Edition), SAP risks maintaining a highly fragmented install base for another decade, potentially slowing the adoption of native S/4HANA features.
• Pricing Complexity: Moving from predictable per-user licensing to consumption-based models often leads to "bill shock" for enterprise customers, potentially slowing adoption of the very AI tools SAP is prioritizing.
• Technical Debt: The "softening" of the migration requirement allows technical debt to persist. Customers may choose the path of least resistance (ECC-to-Cloud) without cleaning up decades of legacy code.

Expert Perspective

The reorganization of SAP's leadership—specifically expanding Thomas Saueressig's role to lead the Customer Value Group—is a clear signal that SAP has accepted it cannot force its customers to move at the speed of its cloud ambitions.

From a technical standpoint, this is a pragmatic retreat. The €2 billion revenue miss in cloud migration targets suggests that the complexity of global manufacturing and industrial ERP environments was underestimated. By pivoting to an AI-upsell strategy, SAP is effectively trying to monetize the existing installed base while they wait for the long-tail migration to eventually complete by 2033. The success of this strategy hinges entirely on whether Joule and the Business Data Cloud provide enough tangible ROI to justify the new consumption costs.

Technical FAQ

How does SAP’s AI pivot affect customers still on ECC?

Customers on ECC who move to the cloud via a SAP ERP Private Edition subscription will likely gain access to Joule and other AI innovations. SAP is no longer strictly gatekeeping these features behind a full S/4HANA migration, allowing "brownfield" environments to access GenAI.

What is the technical difference between "RISE with SAP" and the new AI-focused units?

"RISE with SAP" was primarily a commercial vehicle to bundle hosting, licensing, and migration services into one contract. The new Customer Value Group and AI units focus on post-migration "upselling" of innovation modules (AI, Data Cloud) and shifting the billing from a flat subscription to consumption-based metrics.

Is the per-employee licensing model being replaced entirely?

SAP CEO Christian Klein has indicated a shift away from per-employee licensing, specifically because AI-driven automation reduces headcount. While existing contracts may be honored, new commercial models for AI products will focus on usage/consumption to align with the value provided by automation.

Does Joule run on-premises?

No. Joule and the associated Business AI capabilities are delivered as cloud-based services. Even for customers in Private Edition or legacy environments, Joule acts as a cloud-integrated orchestration layer, necessitating a connection to SAP's cloud infrastructure.

References

• SAP Support Revenue Discrepancy Analysis (The Register)
• S/4HANA Migration Deadlines and Support Policy (Gartner/SAP)
• Business AI Strategy and Productivity Targets (ARC Advisory Group)

Sources

• The Register
• ARC Advisory Group
• Kyndryl Analysis

All technical specifications, pricing, and benchmark data in this article are sourced directly from official announcements. Competitor comparisons use publicly available data at time of publication. We update our coverage as new information becomes available.