Reserved instance, savings plan, on demand. The buyer side commitment model for RISE.

The hyperscaler infrastructure that sits underneath a RISE with SAP contract is priced in three primary modes. Reserved instances, savings plans, and on demand. The mode that SAP applies inside the RISE bundle determines a substantial portion of the seven year cost, often twenty five to thirty five percent of the infrastructure component. The buyer side commitment model, the question of how much capacity to commit, at what tenor, and through which mechanism, is one of the most consequential decisions in the negotiation. It is also one of the least visible. The RISE order form does not generally itemise the infrastructure pricing model, and the SAP account team does not generally surface the question. The buyer team has to surface it deliberately.

The three commitment modes, in plain terms.

A reserved instance is a fixed capacity commitment for a defined term, typically one or three years, paid in full, partially upfront, or monthly. The reservation buys a discount against the on demand rate, ranging from thirty percent for a one year all upfront to sixty five percent for a three year all upfront on AWS, with broadly similar ranges on Azure and Google Cloud. The reservation is tied to a specific instance family and a specific region. If the workload moves, the reservation does not automatically follow.

A savings plan is a financial commitment to spend a certain dollar amount per hour for a defined term. The commitment buys a discount that approaches the reserved instance discount but applies across instance families and, in the case of compute savings plans, across regions. The trade off is that the savings plan does not guarantee capacity in the way a reservation does. The buyer commits to spend, not to specific machines.

On demand is the unreserved hourly rate. It is the highest unit price but the most flexible, with no commitment, no tenor, and no penalty for stopping. On demand is appropriate for variable workloads, short term test landscapes, and the buffer above the committed base.

How SAP applies these inside RISE.

Under RISE with SAP, the infrastructure capacity is procured by SAP from the hyperscaler and resold to the buyer inside the bundled subscription price. The buyer does not see the underlying reservation, savings plan, or on demand allocation directly. The bundle pricing includes an assumed mix that SAP has constructed to optimise its own margin. Where SAP can build the bundle on three year reservations against predictable workloads, the SAP margin is highest. Where SAP must rely on on demand capacity to absorb buyer variability, the SAP margin is squeezed.

The buyer leverage point is that SAP needs the buyer commitment to be predictable in order to build the bundle on reserved capacity. The longer the RISE term, the larger the production landscape, and the more stable the consumption profile, the more SAP can lean on reserved instances. The buyer who provides that predictability has a legitimate claim to a share of the resulting discount. The buyer who does not surface the question rarely receives it.

The buyer side commitment model.

The buyer side commitment model is a worksheet that maps the production landscape, the non production landscape, and the burst capacity required across the seven year RISE term. The worksheet identifies the stable base, the predictable growth, and the variable peaks. The stable base, typically sixty to seventy percent of the production landscape, is the candidate for three year reserved instances. The predictable growth, ten to twenty percent, is the candidate for one year reservations or savings plans. The variable peaks, ten to twenty percent, are the candidate for on demand pricing.

The worksheet is presented to the SAP account team during the negotiation as the basis for the infrastructure pricing inside the bundle. The buyer position is that SAP should construct the bundle against the commitment profile the buyer has documented, and the resulting reservation discount should be reflected in the bundle price. The position is not theoretical. It maps to the actual procurement mechanics on the hyperscaler side, and the SAP commercial team understands the mechanics even when the account team does not surface them.

Discount benchmarks across AWS, Azure, Google Cloud.

For SAP workloads in 2026, the typical discount ranges off the on demand rate are as follows. AWS three year all upfront EC2 reserved instances for SAP HANA certified instance families, sixty to sixty five percent. AWS compute savings plans for the same workload, fifty four to fifty eight percent. AWS one year all upfront, thirty five to forty percent. Azure three year reserved virtual machine instances for SAP, sixty to sixty four percent. Azure savings plans for compute, fifty to fifty five percent. Azure one year reserved, thirty two to thirty eight percent. Google Cloud three year committed use discounts, fifty seven to sixty two percent. Google Cloud one year committed use, thirty four to thirty seven percent.

These benchmarks are the reference points the buyer team uses to test whether the SAP bundle is reflecting reserved pricing economics. Where the SAP infrastructure component, isolated from the application and operations layers, prices materially above the reserved benchmark, the bundle is carrying spread that should be negotiated down. The benchmark is also the reference point for the hyperscaler exit conversation. A buyer who exits RISE and procures infrastructure directly recovers the discount the bundle was suppressing.

The seven year scenario, in dollars.

Consider a mid sized SAP landscape, two thousand FUE, fifteen terabyte HANA database, three application servers, with a seven year RISE term. The infrastructure component inside the bundle, at on demand equivalent pricing, runs at roughly four hundred and twenty thousand dollars per year. The same capacity on three year reserved pricing runs at roughly one hundred and sixty thousand dollars per year. The difference, two hundred and sixty thousand dollars per year, is one and eight tenths million dollars across the seven year term. That is the discount the buyer is entitled to claim if the commitment profile justifies it.

The realistic outcome of a well prepared negotiation is that the buyer captures sixty to seventy percent of the reservation discount. The remaining thirty to forty percent is retained by SAP as margin for assuming the commitment risk on behalf of the buyer. The buyer who does not surface the question captures none of it. The difference between capturing the discount and not capturing it can be more than a million dollars across the term.

The bundle pricing includes an assumed mix that SAP has constructed to optimise its own margin. The buyer who provides predictability has a legitimate claim to a share of the resulting discount.

Mid term true ups and reservation drift.

Reservations and savings plans are made against a forecast. The forecast turns out to be wrong in some direction over a seven year term. The buyer team should negotiate a mid term true up mechanism inside the RISE contract that allows the reservation profile to be adjusted at year three and year five. The true up should be bilateral. Where buyer consumption is below the forecast, the SAP bundle price reduces to reflect the unused commitment. Where buyer consumption is above the forecast, the buyer pays the differential at the reserved rate, not at the on demand rate.

Without the true up mechanism, the buyer is exposed to drift in both directions. Overcommitted reservations become stranded cost. Undercommitted capacity becomes on demand spend at full price. The true up mechanism is rarely offered. It has to be requested, drafted, and negotiated, and it is one of the most valuable clauses the buyer team can insert into a RISE infrastructure schedule.

Conclusion.

The commitment mode underneath the RISE infrastructure layer is one of the largest unaddressed cost levers in most RISE negotiations. Reserved instances, savings plans, and on demand pricing each apply to different parts of the production and non production landscape. The buyer who builds a commitment profile, presents it to SAP, and negotiates a share of the reservation discount captures one to two million dollars of value across a typical seven year term. The buyer who treats the infrastructure layer as opaque captures none of it. The mechanics are public, the discount benchmarks are public, and the SAP commercial team understands the conversation. The question is whether the buyer team surfaces it.