Why the SAP Cloud Business Case Keeps Coming Up Short

Ask an SAP organization still running ECC why it has not committed to SAP S/4HANA, and the answer is increasingly about money rather than technology. SAPinsider’s ERP Migration and Transformation 2026 report found that 63% of organizations with no SAP S/4HANA plans cite rising cloud infrastructure and deployment costs as their biggest barrier, ahead of the 50% who cannot build a value proposition at all. That should make cloud costs the most scrutinized number in any migration plan. Instead, it often ranks low on the organization’s priority list.

The reason is structural. Most business cases compare what an organization pays today for licenses and hosting with what it will pay for a subscription, and cloud often wins that comparison. However, the costs that decide whether a migration pays back sit in integration architecture, operations teams that never leave, the parallel landscapes staged migrations create, and the AI consumption bill that agentic workloads will add. These costs land in different budgets, surface after signing, and mostly stay with the customer regardless of who hosts the infrastructure.

The Costs That Arrive After the Signature

Integration is usually where the gap opens first. SAPinsider’s analysis of the BTP-to-S/4HANA integration decision argues that where an organization draws its integration boundary sets TCO over a five-to-ten-year horizon. Point-to-point connections look inexpensive at go-live, which is when the business case is written. Still, complexity grows exponentially as systems are added, and in-app customization hardens into technical debt that makes every later upgrade slower and costlier.

Operations follow a similar pattern. A managed contract moves infrastructure ownership, not accountability. SAPinsider’s RISE with SAP 2025 benchmark found only 62% of live users rigorously follow shared responsibility guidelines, and just one-third conduct regular monitoring and auditing. The Basis, CloudOps, and SecOps skills that a business case retires remain essential, and their cost resurfaces in the IT budget instead of the migration model.

Then there is the landscape itself. Only 34% of organizations have fully completed their SAP S/4HANA transition, which means most will run old and new environments side by side for years. SAPinsider’s recent coverage on hybrid SAP landscapes describes the result: Solution Manager for ECC, Cloud ALM for Cloud ERP, hyperscaler-native tools, and standalone automation, each with its own licenses, storage, and maintenance cycles, all before a single SAP workload has moved.

The newest omission is AI. SAP concentrates Joule and SAP Business AI on Cloud ERP contracts, making cloud the route to its innovation roadmap. As agentic workloads spread across finance, procurement, and supply chain, every model call adds to a token bill that grows with adoption rather than contract terms. Because most organizations don’t yet have that bill, it rarely appears in the model.

Making the Number Honest

Hyperscaler pricing adds further distortion. Core SAP finance and logistics systems run steady and around-the-clock loads. This workload profile is least suited to consumption pricing. Dell’s three-year TCO study comparing on-premises SAP HANA and SAP S/4HANA on VxRail with AWS shows how Direct Connect, NAT gateway data processing, and S3 backup charges accumulate outside the headline compute price. Add a variable AI inferencing layer onto the same model, and the CFO inherits two unpredictable lines instead of one.

So the correction isn’t about abandoning the cloud; it’s about matching each workload to the pricing model that fits its behavior. A credible five-year model places steady-state SAP and high-frequency inferencing on customer-controlled, consumption-priced capacity and reserves public cloud for burst demand and frontier models. SAP has built for this scenario. Its Customer Data Center option for Cloud ERP Private supports deployment where local hyperscaler options are unavailable, keeps payload data inside the customer’s physical landscape, and retains local access to Joule and SAP Business AI through a secure BTP connection. It also treats operations as a division of labor the customer chooses, forcing the retained-operations question into the open before contracts are signed.

Dell is SAP’s named infrastructure provider for that option, delivering SAP Cloud ERP Private on Dell APEX platform-as-a-service. It supplies, installs, monitors, and manages SAP-certified infrastructure in the customer’s data center or colocation facility on a pay-per-use model under a single SAP-assured end-to-end SLA. Paired with Dell AI Factory with NVIDIA for inferencing on owned capacity, the approach preserves SAP’s cloud operations and AI access while giving finance leaders costs they can forecast.

What This Means for SAPinsiders

Model the costs the subscription does not replace. CIOs and ERP program managers should price retained operations, integration architecture, and parallel tooling explicitly before signing, since the subscription replaces licensing and hosting lines, not operating costs.

Treat AI consumption as a TCO line item now. Finance and IT leaders should estimate inferencing volume per agentic use case and decide which workloads belong to owned capacity versus public APIs.

Price every deployment path like-for-like. Enterprise architects should evaluate the Customer Data Center option and hybrid models alongside the hyperscaler default on an identical five-year basis, including egress, connectivity, backup, and staffing.