SAP AI Budget & Forecasting: What Enterprises Need to Know

The SAP AI Sales Pitch vs. Commercial Reality

SAP has invested heavily in positioning itself as an AI-first ERP vendor. Every sales presentation, every SAPPHIRE keynote, every account review features SAP Business AI prominently. The message is consistent: AI is embedded, AI is included, AI will transform your enterprise. This is partially true and substantially misleading when you examine the commercial terms beneath the marketing language.

Here is what SAP AI actually means for your budget. SAP offers AI capabilities under the SAP Business AI umbrella, which includes SAP Joule (the generative AI copilot), embedded AI features in S/4HANA Cloud (predictive cash management, intelligent document routing, automated matching), and custom AI development capabilities via SAP AI Core on BTP. Each of these has a different commercial model, and the interaction between them is where budget complexity compounds.

The fundamental truth that SAP sales teams underemphasise: the AI features you see demonstrated in SAP product videos require a BTP infrastructure layer that is metered and charged separately from your S/4HANA subscription. The AI capability is real. The "included" description is selectively accurate. The cost to run it at enterprise scale is a separate conversation that SAP prefers to have after you've committed to RISE.

For enterprises currently evaluating RISE or approaching RISE renewal, understanding SAP AI licensing before signing is the single highest-leverage action you can take. Our complete SAP AI budget and forecasting guide covers the full framework. This article addresses the foundational knowledge every enterprise team needs before entering any SAP AI conversation.

The Three Currencies of SAP AI Billing

Understanding SAP AI costs requires knowing which billing currency applies to which service. This is the first area where enterprise teams consistently get into trouble — and SAP does nothing to simplify it.

Currency 1: BTP Credits

SAP Business Technology Platform credits are the primary billing currency for AI services that enterprises deploy and customise. AI Core (for model training and hosting), AI Launchpad (for managing AI lifecycle), and most custom AI services consume BTP credits. Credits are typically allocated annually and have a use-it-or-lose-it structure. The BTP credit allocation in a standard RISE contract was sized for integration and extension scenarios. AI workloads consume 3–8x more credits per compute hour than typical BTP integration scenarios.

If your RISE contract was signed more than 18 months ago, your BTP credit pool almost certainly insufficient for the AI use cases being discussed in your SAP conversations today. This is not an accident — SAP sold those contracts knowing AI consumption was coming and did not pre-provision adequate credits.

Currency 2: Cloud Credit Units (CCUs)

Introduced in 2023, Cloud Credit Units are a billing mechanism for certain embedded AI features in S/4HANA Cloud and RISE. CCUs are functionally similar to BTP credits but exist in a separate entitlement pool with different pricing. The critical issue: contracts signed before SAP introduced CCUs do not contain a CCU allocation. Any AI feature that runs on CCU billing in your environment will therefore trigger unbundled charges at SAP's list rates — which are substantially higher than negotiated rates.

We have seen enterprises discover mid-year that specific SAP AI features they believed were "included" were actually consuming CCUs from a pool that didn't exist in their contract. The resulting invoices are treated as unbundled consumption rather than contract overages, which means standard overage protections (if any) do not apply. Before activating any AI feature in your RISE environment, demand written confirmation from SAP of which billing currency it uses.

Currency 3: Capacity SKUs

For enterprise-scale Joule deployments and high-volume AI use cases, SAP offers capacity-based SKUs that provide a fixed allocation of AI interactions per month. These are purchased separately from your RISE subscription and are not published in SAP's standard price list. Pricing is negotiated deal-by-deal and varies significantly based on organisation size, industry, geography, and negotiating leverage. Enterprises that accept SAP's first capacity SKU proposal typically pay 40–60% more than those who engage in structured negotiation. See our guide on the SAP AI negotiation approach for the tactical framework.

What "Included AI" Actually Means in RISE

When SAP says AI is included in RISE with SAP, they mean: a curated set of Joule scenarios is available in the RISE user interface for the user types covered by your subscription, subject to the usage constraints in your specific commercial schedule. This is a meaningful but limited inclusion, and the constraints are where the cost pressure emerges.

The included Joule scenarios in RISE as of late 2025 cover approximately 50 embedded AI capabilities: guided processes (helping users navigate transactions), natural language queries (asking questions about your SAP data), automated document summaries, and simple workflow recommendations. For a small team of finance and procurement super-users testing AI capabilities, these included scenarios are genuinely useful and carry no additional cost.

The problems start when you try to roll Joule out to an enterprise population of 5,000+ users, or when you want to customise scenarios for your specific processes, or when you activate AI capabilities that process high document volumes (invoice AI, purchase order matching, cash allocation). At that scale, the included allocation is exhausted quickly, and the cost model shifts to metered consumption.

The practical test: if your AI use case is one that SAP can demonstrate in a 30-minute product demo without mentioning BTP credits or additional SKUs, it is probably included in RISE. If it requires a follow-up call with the commercials team, you are in additional-cost territory. This heuristic is imprecise but directionally accurate. Proper validation requires a contract and entitlement review.

The Cost Escalation Risk Profile for SAP AI

SAP AI cost escalation follows a predictable pattern that our advisory team has observed across dozens of enterprise deployments. Understanding this pattern allows you to anticipate and mitigate cost risks before they materialise.

Phase 1: Pilot (Months 1–3)

AI usage is low. Consumption runs within the included allocation. Finance teams have no visibility of pending costs. This phase generates enthusiasm and accelerates the business case for broader deployment. SAP actively supports this phase and encourages expansion.

Phase 2: Activation (Months 4–9)

Broader deployment begins. Development environments and production environments both consume BTP credits. Background AI processes (automated matching, predictive runs) start accumulating consumption that is invisible to business users. Consumption climbs past the included allocation. SAP's BTP cockpit shows overages but the dashboards have a 24–72 hour lag, making real-time management difficult. This is where most enterprises first discover their AI budget assumptions were wrong.

Phase 3: Scale (Months 10–18)

AI is embedded in core processes. Consumption is now structurally committed — disabling AI would disrupt operations. At this point SAP holds significant commercial leverage. Any renegotiation is from a position of dependency. The enterprise either absorbs the overage costs or faces a difficult decommissioning project. SAP's account team is acutely aware of this dynamic.

The intervention point that prevents Phase 3 lock-in: establish a consumption tracking framework during Phase 1 and conduct an independent commercial review before Phase 2 begins. This is the window where negotiating leverage is highest and cost outcomes are most controllable.

Five Questions Every Enterprise Must Answer Before Deploying SAP AI

We use these five questions as a pre-deployment diagnostic with every enterprise client considering SAP AI. If you cannot answer all five confidently, you are not ready to deploy at scale — and you should not sign any additional AI commercial commitments until you can.

1. What is our current BTP credit allocation and what does it cost us if we exhaust it?

Pull your contract and find the BTP credit allocation. Identify the overage rate (typically expressed as a per-credit charge or a percentage uplift on the contracted rate). Model the financial exposure if consumption runs at 150%, 200%, and 300% of the allocated amount. If the overage rate is not explicitly stated in your contract, SAP will charge list price — which is typically 200–300% of the negotiated rate.

2. Which AI features in our use case plan require CCU consumption?

Ask SAP to provide a written list of every AI feature in your deployment roadmap, tagged with its billing mechanism (BTP credits, CCU, or included). Do not accept verbal assurances. If SAP cannot provide this in writing, it is because the answer creates commercial liability for them.

3. Have we modelled AI consumption independently of SAP's estimates?

SAP's consumption estimators consistently understate real-world usage. Before committing to any AI deployment budget, build an independent consumption model using SAP's published billing coefficients (available in the BTP service catalogue) and your own process volumes. Apply a 30% uplift buffer. Compare against SAP's estimate and investigate any discrepancy greater than 15%.

4. What contractual overage protections do we have?

Review your contract for: an explicit overage cap (limits your financial exposure if consumption exceeds entitlement), a consumption carryover provision (allows unused credits to roll into the next year), a right to purchase additional credits at contracted rather than list rates, and a quarterly consumption review cadence that obligates SAP to provide forward projections. Many enterprise contracts contain none of these protections — they are all negotiable before signing. See our SAP contract negotiation service for guidance.

5. Do we have governance controls in place to prevent runaway consumption?

Governance for AI consumption is different from governance for traditional SAP licences. The consumption model requires real-time monitoring, alerting, and the operational authority to pause or throttle AI workloads. If your organisation has no team with both the technical access to implement BTP consumption controls and the commercial authority to act on alerts, you have a governance gap that will generate cost surprises.

How SAP Approaches the AI Commercial Conversation

Understanding SAP's commercial strategy in AI conversations gives you the information you need to position your own negotiation effectively. SAP's primary objective in AI discussions is to move enterprises from the "exploring" stage to the "committed" stage before the cost implications are fully understood. The tools they use for this include:

• Demo environments with unlimited credits: SAP demo and trial environments for Joule and AI Core do not consume your contracted BTP credits. The generous experience in demo mode creates expectations that do not reflect your production cost structure.
• Phase-gated pricing disclosure: SAP does not proactively disclose the full AI cost model in initial RISE presentations. Cost questions typically trigger escalation to a separate commercial team — which itself creates a multi-week delay that is used to build organisational momentum before the numbers are visible.
• Bundling complexity: By having three billing currencies (BTP credits, CCUs, capacity SKUs) and dozens of AI services, SAP creates a complexity that makes independent modelling difficult. Most enterprises simply accept SAP's estimates because reverse-engineering the cost model requires expertise that is not available internally.

The countermeasure: engage an independent SAP licensing advisor before any AI commercial discussion with SAP, and commit internally to not proceeding with any AI commercial commitment until you have an independent cost model in hand. Our SAP licence optimisation team specialises in exactly this type of pre-commitment review.