Licensing strategy for the cloud, the complexity does not leave
Every enterprise cloud strategy carries a licensing assumption, and most assumptions are quietly wrong. Cloud changes the metric, the contracting party, the renewal cadence, and the audit motion for every vendor on the estate. It does not shorten the vendor list, and the post cloud position is usually more complex than the one before it.
Prepared by Redress Compliance · August 6, 2026 · Cross vendor licensing advisory. Based on the migration engagements of the eleven vendor practices.
Executive summary
The cloud simplification narrative is structurally seductive: consumption pricing replaces licensing, managed services absorb the compliance conversation, and the hyperscaler contract displaces the old vendor relationships. None of these assertions survives contact with the actual contract base.
Oracle still negotiates Oracle terms whether the workload runs in OCI, Azure, or AWS. Microsoft still controls the Microsoft commercial on every cloud. SAP still owns the entitlement whether the conversion is to RISE or a hosted estate.
IBM still applies Passport Advantage across the public cloud footprint.
What actually happens is addition: the hyperscaler relationship adds a vendor, it does not remove any.
Most enterprises manage more commercial conversations after migration than before, with the hyperscaler now sitting in the middle of conversations the customer used to have directly, and a marketplace channel whose convenience carries its own pricing consequences.
The rules that decide the money are vendor specific and precise. Oracle counts two vCPUs as one processor license in the authorized clouds, with the core factor gone.
Microsoft's bring your own license rights run through Azure Hybrid Benefit and are restricted toward the listed providers. IBM's sub capacity discount survives in the cloud only where the measurement tooling follows the workload.
Each rule is small; together they decide whether a migration saves money or silently reprices the estate.
The strategy that holds is the same one that works on premises, applied earlier: model the license position per vendor before the workload moves, not after.
Every one of the expensive surprises in our migration engagements, doubled Oracle counts, lapsed mobility rights, broken sub capacity eligibility, was knowable from the contract base before the first workload shifted.
The simplification myth, and what migration actually changes
Cloud changes four things about every software relationship, and none of them is the vendor list:
| What changes | On premises | In the cloud |
|---|---|---|
| The metric | Processors, cores, named users, installations | vCPUs, OCPUs, consumption meters, and conversion ratios between old and new |
| The contracting party | You and the vendor, directly | You, the vendor, the hyperscaler, and often a marketplace in the middle |
| The renewal cadence | Multi year enterprise agreements | Annual commits, consumption true ups, and co terming across parties |
| The audit motion | Scripts and site measurement | Policy documents, cloud provider inventories, and metric conversion disputes |
The post cloud position is usually more complex than the pre cloud position. The same estate now spans two or three counting regimes per vendor, plus the hyperscaler's own commit.
Complexity is not an argument against migrating; it is an argument against migrating on the account team's licensing summary.
The rules that decide the money, vendor by vendor
Four rule sets cover most of the enterprise estate, and each one has a trap that surfaces only after the workload moves:
- Oracle. In the authorized cloud environments, AWS, Azure, and Google Cloud, Oracle counts two vCPUs as one processor license with hyperthreading enabled, and the core factor table does not apply. OCI converts at two OCPUs per processor license, deliberately more favorable. The same database can carry three different license counts on three clouds, and the cloud environment rules decide which one you pay.
- Microsoft. Bring your own license runs through Azure Hybrid Benefit, which requires active Software Assurance or subscription licenses, and license mobility toward the listed providers, AWS and Google among them, is restricted. The route a workload takes changes what an existing SQL Server estate is worth in the move.
- SAP. The RISE conversion trades your perpetual estate for a subscription measured in Full User Equivalents, and the conversion moment is the negotiation: entitlement credit, the user tier mapping, and the exit posture all price at signature, not later.
- IBM. Sub capacity eligibility travels with the measurement tooling, ILMT on virtual machines and the License Service in containers, and the discount collapses to full capacity where the tooling does not follow the workload into the cloud.
The Azure licensing cost optimization playbook
The BYOL route map, the Hybrid Benefit math, the listed provider rules, and the workload placement decisions that decide what an existing Microsoft estate is worth in the cloud.
Get the white paper →The hyperscaler in the middle, commits and marketplaces
The hyperscaler relationship brings its own commercial machinery: committed spend agreements with their own discount curves, and marketplaces that let third party software draw down that commit. Both are real levers and both cut two ways.
The commit concentrates leverage into a single periodic negotiation, covered in the AWS EDP playbook and the Google Cloud CUD analysis, and the marketplace route can pay software bills with already committed dollars while quietly changing whose discount applies to the transaction.
The strategic point is that the hyperscaler commit and the software estate now interact: an Oracle workload placed on Azure consumes the Microsoft relationship differently than the same workload on OCI consumes Oracle's, and a marketplace purchase moves spend between negotiations.
Workload placement has become a licensing decision, and the organizations that treat it as purely an architecture decision leave one side of the equation unmanaged.
- Percentile standing for your exact deal size and industry, from real closed transactions
- Scenario simulation before the call: test alternative terms and see the financial impact of each
- A negotiation playbook, talking points, and a two page executive brief on day one
The audit motion follows the workload
Migration does not retire audit exposure; it changes its shape. Oracle's cloud policy disputes turn on vCPU counting and hyperthreading flags. Microsoft's reviews reach into Hybrid Benefit eligibility and Software Assurance continuity.
IBM's findings follow the tooling gap into every cluster the License Service never reached. And the evidence base shifts from your data center to a mixture of your records and the cloud provider's inventories, which you need to be able to produce on the vendor's timeline.
The defensible position is the same artifact it always was, a dated reconciliation of deployments against entitlements, extended with the cloud specific facts: which counting rule applied to each workload, which enabling program covered each BYOL placement, and when each workload moved.
Estates that keep that record treat vendor letters as correspondence. Estates that do not, negotiate from memory against a counterparty with logs.
The strategy that holds
Across the migration engagements of our vendor practices, the expensive surprises shared one property: every one was knowable in advance from the contract base. The doubled Oracle count was in the policy document. The lapsed mobility right was in the expiring Software Assurance.
The broken sub capacity eligibility was in the tooling plan that stopped at the data center wall. The discipline that prevents them is sequencing:
- Model the license position per vendor before the workload moves. The counting rule, the enabling program, and the metric conversion, priced for each target cloud.
- Make workload placement a licensing decision. The same workload carries different license counts on different clouds; place with both equations visible.
- Keep the enabling programs alive through the transition: Software Assurance for Hybrid Benefit, measurement tooling for IBM, and the entitlement records for everything.
- Negotiate the conversion moments deliberately. RISE, subscription transitions, and marketplace routes each reprice the estate once; the terms set there persist for the decade.
- Extend the baseline into the cloud. One dated reconciliation covering both worlds, refreshed on the renewal calendar. The vendor negotiation practice and cross vendor benchmarking keep it standing.
Frequently asked questions
Does moving to the cloud simplify software licensing?
No. It changes the metric, the contracting party, the renewal cadence, and the audit motion for every vendor, and it adds the hyperscaler as a new commercial relationship without removing any existing one. Most enterprises manage more licensing conversations after migration than before.
How does Oracle licensing work on AWS and Azure?
Under Oracle's authorized cloud environment policy, two vCPUs count as one processor license where hyperthreading is enabled, and the core factor table does not apply. OCI converts more favorably at two OCPUs per processor license.
The same database can carry materially different license counts depending on which cloud runs it.
What is required to bring existing Microsoft licenses to the cloud?
Azure Hybrid Benefit, which requires active Software Assurance or subscription licenses, covers the Azure route. Mobility toward listed providers like AWS and Google is restricted, so the placement decision changes what an existing Windows Server or SQL Server estate is worth.
Letting Software Assurance lapse mid migration forfeits the right.
Does IBM sub capacity licensing survive a cloud migration?
Only where the measurement tooling follows the workload: ILMT for virtual machines and the IBM License Service for containers. Where the tooling does not reach, eligibility collapses to full capacity counting. The discount is portable; the discipline that earns it has to be ported deliberately.
Do cloud marketplace purchases affect software negotiations?
Yes, in both directions. Marketplace purchases can draw down committed cloud spend, paying software bills with already promised dollars, but the route changes whose discount applies and moves spend between negotiations.
Marketplace routing belongs in deal design, evaluated against the direct route before signature.
When should the licensing analysis happen in a cloud migration?
Before workloads move, per vendor and per target cloud. Every expensive surprise in our migration engagements, doubled counts, lapsed rights, broken eligibility, was visible in the contract base in advance. After the move, the same facts are still true but the leverage to negotiate them is gone.
