How SAP's mid-2025 unbundling of Joule, Business AI, and Datasphere changes your licensing obligations, and how to negotiate consumption-based AI and data terms effectively.
This article is part of the SAP S/4HANA: Complete Licensing Guide pillar series. See also: Negotiating BTP Credits with S/4HANA · S/4HANA Add-Ons and HANA Database: Hidden Costs
In mid-2025, SAP revamped its cloud ERP packages and unbundled its AI and sustainability features. The old "Premium Plus" tier of RISE with SAP, which had included advanced AI and analytics, was retired. Tools like the Joule AI assistant and SAP Datasphere became optional add-ons requiring separate licensing.
SAP positions this as "flexibility" (pay only for what you use), but it introduces new cost lines for IT and finance to manage. Where a top-tier SAP subscription might previously have included Datasphere capacity and AI functionality, those now require explicit add-on contracts. Companies that assumed these tools were built-in could face unplanned costs under the new model.
This shift is part of SAP's broader move toward modular, pay-as-you-go licensing. CIOs and CFOs must now explicitly budget for AI and data services. The upside: you avoid paying for AI or analytics you will not use. The downside: if you do need these capabilities, it is critical to negotiate terms upfront to lock in predictable pricing and avoid surprises.
SAP's AI capabilities, including the Joule generative AI copilot and various SAP Business AI features embedded in applications, are offered on a consumption-based model. Enterprises can access these services in three main ways.
| Access Method | How It Works | Best For |
|---|---|---|
| Included Basic Usage | Some AI features come with your SAP licence with a small free allowance (e.g., predictive analytics in S/4HANA Cloud) | Getting started, basic scenarios |
| AI Units (Credit Purchase) | Purchase blocks of AI Units (min ~100/year). Each AI action consumes a fraction of a unit based on SAP's conversion rules. | Advanced AI, generative queries, large-scale automation |
| BTP Cloud Credits | Pay with prepaid BTP credit balance. AI services on SAP Business Technology Platform consume credits on a pay-per-use basis. | Custom AI scenarios, extensions, API-based solutions |
No rollover. SAP tracks AI Unit consumption annually, and unused AI Units expire at year-end. This makes accurate demand forecasting critical. Over-purchasing wastes budget, while under-purchasing risks overage fees or service interruption.
What does it cost? One AI Unit is roughly a few euros at list price. SAP often sets minimum purchase quantities (e.g., 100 units). A starter pack might list at ~EUR 500-1,000 for 100 units before discounts. Each unit covers dozens or hundreds of transactions depending on complexity. The key metric: if EUR 1 of AI spend yields EUR 5 in efficiency, the investment is justified.
SAP Datasphere (formerly SAP Data Warehouse Cloud) is SAP's cloud data warehouse solution, now licensed separately from core ERP subscriptions. There are two primary models.
| Model | How It Works | Cost Drivers | Best For |
|---|---|---|---|
| Capacity Subscription | Commit to fixed capacity (e.g., 128 GB storage + compute) for a flat annual fee | Data volume and compute tier selected | Heavy, predictable analytics usage. Large-scale BI, enterprise data consolidation. |
| Consumption via BTP Credits | Pay-per-use: every runtime hour and GB stored/processed deducts from your BTP credit pool | Actual storage, processing hours, query complexity | Variable/lighter analytics. Pilots, occasional reporting, unpredictable workloads. |
Recommended approach. Many enterprises begin with consumption-based Datasphere to gauge needs, then transition to a capacity subscription once demand stabilises. During negotiations, ask SAP to include a small Datasphere trial or discount in larger deals. Otherwise, budget for this as a separate line item.
SAP's shift to consumption-based licensing means costs directly follow your usage. Both AI services and Datasphere use this model, but their metrics differ.
| Service | Licence Model | Usage Metric | Main Cost Factors |
|---|---|---|---|
| SAP Joule (AI copilot) | Add-on via AI Units | Number of AI queries/tasks converted to AI Units | Frequency of AI queries and interactions |
| Embedded Business AI | Included up to a limit; then AI Units | Count of AI-driven business transactions | Volume beyond the free quota |
| SAP Datasphere | Subscription or BTP consumption | Data storage (GB) and processing hours | Data set size and analytics complexity |
To avoid surprises, align AI and data usage with clear business outcomes. If an AI feature is not delivering enough value to justify its cost, reduce usage or explore alternatives. When a tool delivers strong ROI, increasing usage and cost is justified. The critical discipline: track consumption alongside the business benefits it delivers.
With SAP's new add-on model, negotiating smartly can make a significant difference in total cost.
| # | Action |
|---|---|
| 1 | Inventory needs: List SAP AI features (Joule, document AI, predictive planning) and data services (Datasphere, Analytics Cloud) expected in next 1-2 years |
| 2 | Estimate usage: Project monthly AI queries, document processes, and GB of analytics data using historical patterns |
| 3 | Review entitlements: Check current contracts for included AI credits, BTP credits, or existing data warehousing rights |
| 4 | Engage SAP early: Discuss needs well before renewal. Share usage estimates and request pricing options. Early dialogue reveals promotions. |
| 5 | Negotiate and document: Include AI/Data services at favourable terms. Get all terms in writing: unit counts, costs per unit/GB, free allowances, overage handling. |
SAP retired the Premium Plus tier of RISE with SAP and unbundled AI tools (Joule, Business AI) and data services (Datasphere) into optional add-ons. These capabilities now require separate licensing and explicit budgeting rather than being included in top-tier subscriptions.
AI Units are a prepaid currency for SAP AI consumption. You purchase blocks (minimum ~100 units/year), and each AI action, chatbot queries, automated invoice matching, etc., draws down your pool based on SAP's conversion rates. Unused units expire annually with no rollover.
Two ways: a capacity subscription (fixed storage/compute for a flat annual fee) or consumption via BTP credits (pay-per-use based on actual storage, processing hours, and query complexity). Most enterprises start with consumption to gauge needs, then shift to subscription once demand stabilises.
Partially. BTP Enterprise Agreement credits can be applied to Datasphere and certain AI services. RISE credits may offset some usage. But for substantial consumption, dedicated AI Unit purchases or extra credits are typically needed. Do not assume general pools cover everything.
No. SAP tracks AI Unit consumption annually, and unused units expire at year-end. This makes accurate demand forecasting critical. Over-purchasing wastes budget, while under-purchasing risks overage fees or service interruption.
Bundle into major deals (RISE, S/4HANA renewals), secure volume discounts with price locks, negotiate pilot periods before committing, leverage total SAP spend for concessions, and maintain flexibility with annual adjustment rights and exit clauses.
Unpredictable usage spikes (especially for Datasphere), AI Unit expiration without rollover, overage fees when exceeding allotments, and scope creep as more teams adopt AI features. Mitigation requires monitoring, alerts, and pre-negotiated burst or overage terms.
Start with a pilot. Negotiate reduced-cost access for 6 months, define success criteria, and agree on pre-negotiated expansion pricing. This proves value with data before committing to larger spend, and gives you negotiating leverage for the scale-up phase.