SAP reset the AI and data licensing model in July 2025. AI Units, Joule consumption, and Datasphere capacity now bill outside the classic S/4HANA user metric. Six moves cap the buyer side exposure.
SAP reset the AI and data licensing model on July 1, 2025. AI Units, Joule access, and Datasphere consumption now bill outside the classic S/4HANA Full Use Equivalent metric. Six buyer side moves cap the exposure across the three pools.
The reset matters because the new pools sit outside the historic SAP user count. The customer that does not separate the pools at the negotiation table pays the SAP default opening across all three lines.
The July 2025 reset moved AI and data consumption out of the classic SAP user metric and into separate consumption pools. The change matters at the negotiation table because each pool now requires its own line item, its own commit, and its own discount band.
SAP AI Units are the consumption metric for embedded AI capabilities across the SAP application estate. The unit measures consumption of large language model calls, document processing, embedding generation, and inference operations. The price falls as the committed volume rises.
| Annual commit tier | List rate per AI Unit | Negotiated band |
|---|---|---|
| Pay as you go | 0.40 USD | No discount |
| 100K Units per year | 0.36 USD | 4 to 8 percent |
| 500K Units per year | 0.32 USD | 10 to 16 percent |
| 2M Units per year | 0.28 USD | 18 to 26 percent |
| 10M plus Units per year | 0.24 USD | 26 to 40 percent |
Joule is the SAP conversational AI assistant. Joule access bills separately from the AI Units that the Joule conversations consume. The customer pays both for the access seat and for the units consumed.
SAP Datasphere is the cloud data warehouse and data fabric service. The Datasphere capacity bills on three separate blocks. Compute, storage, and data movement each meter independently against the customer commitment.
The buyer side captures the most value when the three pools negotiate separately and the commitment lands at the right volume tier for each pool. Six moves recur across the SAP AI advisory engagements.
The forecast is the foundation of the commit. The customer that commits on a guess overpays. The defense is to model the AI Units consumption across the top five SAP use cases, the Joule access seats per persona, and the Datasphere capacity per workload.
The seller side prefers to bundle the three pools into a single commitment. The buyer side captures more by negotiating each pool against its own discount band. The negotiation calendar runs the three pools in parallel rather than as a stacked bundle.
The discount band steps occur at defined volume thresholds. The customer that commits just below a threshold misses the next band. The defense is to model the consumption at threshold ratios and to commit at the next band when the forecast supports it.
The AI Units commitment expires at the end of each subscription year. The customer that under consumes loses the unused units. The defense is to negotiate a 30 percent carry over clause that protects up to 30 percent of the unused commitment.
The SAP default carries a 7 to 9 percent annual renewal uplift on subscription services. The buyer side captures a 0 to 3 percent cap on the AI and data pools at signing. The cap protects the unit price across the commitment term.
SAP runs a December 31 calendar year fiscal end. The seller side carries fiscal year quotas tied to that date. The Q4 timing for SAP runs October through December rather than April through June.
| Pool | Metric | List anchor | Top tier band |
|---|---|---|---|
| AI Units | Per unit consumed | 0.40 USD | 26 to 40 percent |
| Joule access | Per user per month | 12 to 18 USD | 14 to 24 percent |
| Datasphere compute | Per Compute Unit per month | Tier specific | 20 to 35 percent |
| Datasphere storage | Per TB per month | 30 USD | 15 to 25 percent |
| Datasphere movement | Per TB ingested or extracted | Tier specific | 10 to 18 percent |
The SAP AI and data checklist runs the three pools in parallel.
SAP separated AI Units, Joule access, and Datasphere consumption from the classic S/4HANA user metric. The three pools now bill against their own commitment lines with their own discount bands. The change moved the AI and data conversation out of the renewal bundle and into a per pool negotiation.
The customer that does not separate the pools at the negotiation table pays the SAP default opening across all three lines. The buyer side captures 18 to 40 percent on the AI Units pool alone when the commitment lands at the right volume tier.
SAP AI Units list at approximately 0.40 USD per unit at pay as you go. The volume commitment bands move the rate down to 0.24 USD at the 10M plus units per year tier. The discount band corresponds to 26 to 40 percent below the pay as you go list at that tier.
The unit is consumed by Joule conversations, document AI processing, embedding generation, predictive forecasting, and generative content workflows. Each operation has a defined unit count that varies by complexity and token volume.
No. Joule access bills as a per user per month subscription. The AI Units consumed by Joule prompts meter against a separate commitment. The customer who buys Joule access without modeling the unit consumption faces an unexpected meter bill.
The defense is to forecast the Joule consumption per user, model the unit count per Joule interaction, and commit on the AI Units pool to cover the forecast. The pool commitment protects the unit price across the term.
SAP Datasphere bills three separate blocks. Compute Units for query and transformation processing, storage capacity per TB per month, and data movement per TB ingested or extracted. Each block has its own commitment line and its own discount band.
The customer that commits on storage alone and underestimates the compute requirement faces a meter bill on compute. The defense is to model the three blocks together and to commit at the right ratio across the three lines.
Across 40 SAP AI advisory engagements, the median overpay against the optimized commit was 22 percent. The range was 12 percent at the low end and 38 percent at the high end. The overpay typically lands in the AI Units pool where the volume tier discount band is widest.
The two recurring failure modes are bundling the three pools into a single commitment and committing just below a volume tier threshold. The defense is to negotiate the three pools separately and to commit at the threshold ratios.
The SAP default does not include a carry over clause on AI Units. The buyer side captures a 30 percent carry over at signing on negotiated commits at the 500K units per year tier and above. The clause protects the unused commitment up to the negotiated cap.
The defense is to model the consumption ramp across the term. A new AI implementation typically consumes less in year one than in year two. The carry over protects the year one under consumption when the year two ramp lands.
Redress runs SAP AI advisory inside the Vendor Shield subscription, the Renewal Program, and the dedicated SAP service line. The work covers the consumption forecast, the pool separation, the volume tier modeling, the negotiation, and the audit defense.
Typical engagements deliver 18 to 40 percent discount across the three pools with the carry over and uplift protections negotiated.
Redress runs this practice inside the Vendor Shield subscription, the Renewal Program, the SAP Hub, and the Software Spend Assessment.
Read the related case studies, the benchmarking service, the Benchmark Program, the management team page, the about us page, and the contact page.
The companion playbook covers the SAP RISE commercial model, the migration credit mechanics, and the negotiation moves that protect the customer through transition.
Independent. Written for CIOs, CFOs, and procurement leaders. No vendor partner affiliation.
Open the playbook in your browser. Corporate email only.
Open the Paper →The July 2025 reset moved AI and data outside the classic SAP user contract. The customer that does not separate the pools at the negotiation table pays the SAP default.
We have advised on 40 SAP AI and data licensing reviews since the July 2025 reset. Every engagement starts with one conversation.
Cost benchmarks, license rightsizing patterns, and the negotiation moves that worked. Written for buyer side teams running active vendor decisions.
Once a month. Audit patterns, renewal benchmarks, vendor commercial signals across Oracle, Microsoft, SAP, Salesforce, IBM, Broadcom, AWS, Google Cloud, ServiceNow, Workday, Cisco, and the GenAI vendors. No follow up sales pressure.
Free providers (Gmail, Yahoo, Outlook) cannot subscribe. Work email only. Unsubscribe in one click.
